Uncharted Waters - National Academy of Social Insurance (NASI) · 2017. 1. 10. · Study Panel...

Uncharted Waters: Paying Benefits From Individual Accounts in Federal Retirement Policy

Co-Chairs

Kenneth S. Apfel

Michael J. Graetz

Study Panel Final Report

The National Academy of Social Insurance is a nonprofit, nonpartisanorganization made up of the nation’s leading experts on social insurance.The Academy’s mission is to promote understanding and informed poli-cymaking in social insurance and related programs through research,training, and the open exchange of ideas. Social insurance, both in theUnited States and abroad, encompasses broad-based systems for insuringworkers and their families against economic insecurity caused by loss inincome from work and the cost of health care. The Academy’s researchcovers social insurance systems such as Social Security, Medicare, work-ers’ compensation, and unemployment insurance, and related social assis-tance and private employee benefits.

The Academy convenes steering committees and study panels that arecharged with conducting research and policy analysis, issuing findings,and, in some cases, making recommendations based on their analyses.Members of these groups are selected for their recognized expertise andwith due consideration for the balance of disciplines and perspectivesappropriate to the project. The findings and any recommendations arethose of the Study Panel and do not represent an official position of theNational Academy of Social Insurance or its funders.

This report presents new analyses and does not make recommendations.It was prepared by and for the Uncharted Waters Study Panel. In accor-dance with procedures of the Academy, the report has been reviewed bya committee of the Academy’s Board of Directors for completeness, accu-racy, clarity and objectivity.

This project received financial support from The Ford Foundation, theActuarial Foundation, TIAA-CREF Institute, and the Foundation forChild Development.

Suggested Citation:

Reno, Virginia P., Michael J. Graetz, Kenneth S. Apfel, Joni Lavery, andCatherine Hill (eds.), (2005). Uncharted Waters: Paying Benefits fromIndividual Accounts in Federal Retirement Policy, Study Panel FinalReport, Washington, DC: National Academy of Social Insurance,January.

© 2005 National Academy of Social InsuranceISBN: 1-884902-42-1

Board of DirectorsHenry J. Aaron

ChairJerry L. Mashaw

PresidentJanice M. Gregory

Vice PresidentJanet L. Shikles

SecretaryWilliam B. Harman Jr.

Treasurer

John F. BurtonJack. C. EbelerRichard HobbieCharles A. JonesMark Novitch

Patricia M. OwensJohn L. PalmerJoseph F. Quinn

Anna M. RappaportMargaret Simms

Andrew SternBruce C. Vladeck

Founding ChairRobert M. Ball

Honorary AdvisorsBill Archer

Nancy Kassebaum BakerLloyd BentsenJohn H. Biggs

Linda Chavez-ThompsonTeresa HeinzRoger Joslin

Beth KoblinerRobert J. MyersPaul H. O’Neill

Franklin D. RainesStanford G. Ross

Alexander Trowbridge

Executive Vice PresidentPamela J. Larson

VP for Income SecurityVirginia P. Reno

VP for Health PolicyKathleen M .King

VP for AdministrationTerry T. Nixon


Co-Chairs

Kenneth S. Apfel

Michael J. Graetz

Study Panel Final Report

Preface

i

That people are living longer and generallyenjoying better health while doing so is unques-tionably a good thing. But as the elderly becomea larger portion of our population, new chal-lenges arise for making sure that retirees haveenough income to live on. In an effort tostrengthen personal savings for retirement,President Bush and both Democratic andRepublican members of Congress have advancedproposals for establishing some form of univer-sally available individual savings accounts.However, the question whether these individualaccounts should be apart from and in additionto Social Security, or instead integrated withSocial Security and a part of any restructuring ofSocial Security, has sharply divided politiciansand policy analysts alike.

Implementing any system of universally avail-able individual accounts requires answers tothree broad questions: How will money get intothese accounts? How will the funds in theseaccounts be invested and managed? How willindividuals or their families obtain paymentsfrom these accounts? Much analysis has beenproduced attempting to answer these questionsin recent years, but most of the analyses have

focused on the “accumulation” phase — howindividuals, in concert with their employers, agovernment institution or regulated privatefinancial institutions, would get the money intothe accounts, then invest and manage theaccounts’ assets during the person’s workingyears. In contrast, questions surrounding the“payout” phase — how individuals wouldreceive their funds after retirement or upondeath or disability—have frequently been neglected. This report examines these largelyunexplored issues in depth.

Why is it important to examine these “payout”issues? Since a central goal of retirement securitypolicy is to assure some level of adequateincome, it is essential that any debate about cre-ating individual accounts include a completeunderstanding of how the benefits will bereceived. How would the assets accumulated inindividual accounts be paid out during retire-ment? Will individuals have funds available tothem before retirement? Do these answerschange if an individual becomes disabled or diesbefore retirement? What rights does a spouse orformer spouse have to these accounts? Can cred-itors reach the accounts? What institutions—

ii

government or private—will be responsible formaking payments from the accounts? If privateinstitutions are responsible, will the federal orstate governments regulate their conduct? Ifthese new accounts are part of Social Security orintegrated with Social Security reforms, whatwill happen to payouts of Social Security bene-fits? These are the kinds of questions this reportaddresses in detail.

It is crucial that these payout questions be clear-ly understood by policymakers. They mustresolve each of the issues discussed in this reportif they are going to add universally availableindividual accounts to our current system forproviding retirement income. In addressing theissues here, this report has not concentrated onany one plan for revamping our existing system.Rather, it has considered a wide range of pro-posals that have been advanced across the politi-cal spectrum. Since it is important to understandhow any new individual account program wouldfit with traditional Social Security, employer-based pensions and other tax-advantaged indi-vidual retirement savings vehicles to answerthese questions, our decision to consider a vari-ety of individual account proposals has substan-tially complicated our task. Moreover, ratherthan insisting on one “best” answer to the ques-tions we examine here, this panel offers a rangeof potential answers. We hope that this com-plexity does not deter our readers from comingto their own conclusions about answers to thesequestions.

We want to emphasize that it never was theintention of this panel to come to agreement onthe desirability of creating individual savingsaccounts as a part of Social Security. Indeed,panel members hold sharply divergent views onthis issue. Nor was it our intention to present a“blueprint” for how to design payouts fromindividual accounts. Panel members hold quitedifferent views on these issues as well. We did,however, all agree about what are the key issuesthat must be addressed in considering payoutsfrom individual accounts. And our panel mem-bers also found common ground on the poten-

tial implications and trade-offs of various policychoices.

Pulling together more than two dozen knowl-edgeable, strong-minded, energetic and political-ly divergent experts to tackle these issues mighthave been a recipe for disaster. But throughnumerous meetings, untold conference calls andliterally thousands of e-mails extending over aperiod of more than two years, this remarkablegroup of people generously gave their time,knowledge, and judgment to bring this report tofruition. Draft after draft has been subject todetailed debate. Controversies have arisen andbeen resolved over a panoply of issues large andsmall. The commitment of these panel memberscannot be overstated.

Our panel was greatly assisted by the efforts ofthe Social Security Administration and its officeof the Chief Actuary. We are also grateful forthe financial support provided by the FordFoundation, the Actuarial Foundation, theTIAA-CREF Institute and the Foundation forChild Development. That said, the substance ofthe report is that of the members of the paneland not of any of the organizations mentionedhere.

Finally, this report would never have been com-pleted without the sustained effort and expertiseof the staff at the National Academy of SocialInsurance. Particular thanks are due to VirginiaReno, who directed the NASI staff, and to JoniLavery, who, along with Virginia, finalized thisreport. We also benefited from the expertise andwork of Catherine Hill, who served as a staffmember for this report in its early stages. NellyGanesan and Anita Cardwell also provided criti-cal logistical support.

It has been an honor to serve as co-chairs of thisexceptional panel.

Kenneth S. Apfel Michael J. GraetzCo-Chair Co-Chair

Kenneth S. Apfel, Co-ChairSid Richardson Chair in Public AffairsLBJ School of Public AffairsThe University of Texas at Austin

Michael J. Graetz, Co-ChairJustus S. Hotchkiss Professor of LawYale Law School

Lily BatchelderAssistant Professor of Law and Public Policy NYU School of Law

Ray BosharaDirector, Asset Building ProgramNew America FoundationWashington, DC

Jeffrey BrownAssistant Professor of FinanceUniversity of Illinois at Urbana-Champaign

Craig CopelandSenior Research AssociateEmployee Benefit Research InstituteWashington, DC

Douglas ElliottPresidentCenter on Federal Financial InstitutionsWashington, DC

Joan EntmacherVice President and Director of Family EconomicSecurityNational Women's Law CenterWashington, DC

Martha E. FordDirector of Legal AdvocacyThe Arc of the United States andUnited Cerebral PalsyWashington, DC

Douglas ForeDirector of Portfolio AnalyticsTIAA-CREF Investment ManagementNew York, NY

Fred GoldbergAttorneySkadden, Arps, Slate, Meagher & Flom LLPWashington, DC

Stephen C. GossChief ActuarySocial Security Administration

Karen C. HoldenProfessor of Public Affairs & Consumer ScienceUniversity of Wisconsin

J. Mark IwryNon-Resident Senior FellowThe Brookings InstitutionWashington, DC

Howell JacksonJames S. Reid, Jr. Professor of LawHarvard Law School

Charles A. JonesDirector of ReengineeringMichigan Family Independence AgencyLansing, MI

Kilolo KijakaziProgram OfficerFord FoundationNew York, NY

John H. LangbeinSterling Professor of Law and Legal HistoryYale Law School

Maya MacGuineasDirector, Fiscal Policy Program &Retirement Security ProgramNew America FoundationWashington, DC

Lisa MensahExecutive Director Initiative on Financial SecurityThe Aspen InstituteNew York, NY

Uncharted Waters Study Panel

Project StaffVirginia P. Reno, Vice President for Income Security

Joni Lavery, Income Security Research Associate

Anita Cardwell, Income Security Research Assistant

Jill Braunstein, Director of Communications

Simona Tudose, Communications Associate

Catherine Hill, Income Security Research Associate (until March 2004)

Nelly Ganesan, Income Security Research Assistant (until August 2004)

Acknowledgements

Many people were instrumental in helping make this report possible. The National Academy of SocialInsurance would like to thank all the Uncharted Waters Panel members for their expertise and theirgenerous commitment of time and energy in bringing this report to fruition. Special thanks to theSocial Security Administration’s Office of the Chief Actuary for their professional expertise and coop-eration. And finally, the Academy would like to thank Laura Haltzel, Dawn Nuschler, and Gary Sidorfor their help in analyzing components of individual account proposals.

Peter R. OrszagJoseph A. Pechman Senior Fellow Economic StudiesThe Brookings InstitutionWashington, DC

Pamela J. PerunAfflilated Scholar at the Urban InstituteBerkeley, CA

Eric RodriguezDirector, Economic Mobility InitiativeNational Council of La RazaWashington, DC

Jane RossDirector, Center for Social and Economic StudiesNational Research CouncilWashington, DC

Dallas L. SalisburyPresidentEmployee Benefit Research InstituteWashington, DC

Bruce SchobelVice President and ActuaryNew York Life Insurance CompanyNew York, NY

Sheila ZedlewskiDirector, Benefits Policy CenterThe Urban InstituteWashington, DC


v

PREFACE…………………………………………………………………………….. ........................................................i

UNCHARTED WATERS STUDY PANEL……………………………………………......................................................iii

LIST OF FIGURES………………………………………………………………….....................................................….vii

CHAPTER ONE: INTRODUCTION AND SUMMARY…………………………......................................................…..1

Social Insurance and Property……………………………………………………..............................................2

Framework for Analyzing Payout Rules…………………………………………. ...........................................5

Examples of Individual Account Plans……………………………………………............................................5

Social Security Finances and Solvency Projections……………………………….. .........................................7

Benefit Changes for Solvency and Benefit Offsets………………………………… ......................................8

Financial Demographics…………………………………………………………….............................................9

Payments at Retirement……………………………………………………………… ......................................10

Institutional Arrangements for Providing Life Annuities……………………………… .............................12

Pre-Retirement Access to Individual Accounts………………………………………....................................14

Spousal Rights……………………………………………………………………….. .........................................15

Disabled Workers and their Families………………………………………………….. ..................................17

Children, Life Insurance, and Bequests……………………………………………….....................................18

Worker-Specific Offsets…………………………………………………………………....................................19

Individual Account Taxation……………………………………………………………. ..................................20

Concluding Remarks…………………………………………………………………….....................................21

CHAPTER TWO: FINANCIAL DEMOGRAPHICS…………………………………………..........................................23

Components of Retirement Income………………………………………………………. .............................23

Role of Social Security—Past and Future…………………………………………………..............................27

Tax Favored Retirement Savings………………………………………………………….. ..............................28

Other Measures of Financial Security and Stress…………………………………………. ...........................34

How Big Might Individual Accounts Be?…………………………………………………..............................37

Are Americans Linked to Potential Administrators of Accounts?………………………….......................39

Summary………………………………………………………………………………… .....................................42

CHAPTER THREE: PAYMENTS AT RETIREMENT………………………………………............................................45

Financial Risks in Retirement…………………………………………………………….. ................................46

Framework for Analyzing Retirement Payout Options………………………………….............................48

A Primer on Life Annuities………………………………………………………………..................................51

Retirement Payout Options……………………………………………………………….................................56

Design and Implementation Issues………………………………………………………................................62

Summary…………………………………………………………………………………. ....................................72

CHAPTER FOUR: INSTITUTIONAL ARRANGEMENTS FOR PROVIDING ANNUITIES.. .........................................75

U.S. Market in Life Annuities……………………………………………………………..................................76

A Framework and Existing Annuity Arrangements………………………………………. ..........................82

Providing Inflation-Indexed Annuities……………………………………………………..............................85

vi

Hybrid Models to Provide Inflation-Indexed Annuities………………………………….............................90Summary…………………………………………………………………………………. ....................................94

CHAPTER FIVE: PRE-RETIREMENT ACCESS TO INDIVIDUAL ACCOUNTS…………. ...........................................97Framework for Analyzing Access Rules........................................................................................................97Precedents for Access to Retirement Funds……………………………………………….............................99Options for Early Access………………………………………………………………….................................101Design and Implementation Issues…………………………………………………………..........................103Summary………………………………………………………………………………….. .................................111Appendix: Precedents for Early Access………………………………………………….. ............................113

CHAPTER SIX: SPOUSAL RIGHTS………………………………………………………….. .....................................117Precedents for Spousal Rights…………………………………………………………….. ............................117Framework for Analyzing Spousal Rights………………………………………………… ..........................124Policy Choices for Spousal Rights………………………………………………………….............................125Implementation Issues……………………………………………………………………................................133Summary………………………………………………………………………………….. .................................135Appendix: Precedents for Spousal Rights…………………………………………………...........................139

CHAPTER SEVEN: DISABLED WORKERS AND THEIR FAMILIES……………………….......................................143Introduction ..................................................................................................................................................143Payment Options for Disabled Workers…………………………………………………… .........................146Design and Implementation Issues…………………………………………………………..........................152Summary…………………………………………………………………………………… ...............................154

CHAPTER EIGHT: CHILDREN, LIFE INSURANCE, AND BEQUESTS……………………........................................157Current Social Security Benefits for Children……………………………………………... .........................157Defined-Benefit Payment Options for Young Survivor Families…………………………. ......................159Payment Options for Children of Retired and Disabled Workers…………………………. ....................161Children’s Rights to their Parents’ Individual Accounts…………………………………...........................162Bequests to Heirs other than Spouses and Dependent Children………………………….. ....................163Disabled Adult Children—A Profile……………………………………………………….............................164Summary………………………………………………………………………………….. .................................165Appendix: Profile of Disabled Adult Children...........................................................................................167

CHAPTER NINE: WORKER-SPECIFIC OFFSETS…………………………………………..........................................169Intention Behind Offsetting Benefits………………………………………………………..........................169Design of Worker-Specific Offsets……………………………………………………….. .............................171Applying Offsets at other Life Events……………………………………………………. ............................177Offset Administrative and Legal Issues……………………………………………………...........................182Summary………………………………………………………………………………….. .................................185

CHAPTER TEN: INDIVIDUAL ACCOUNT TAXATION…………………………………….......................................189Tax Equivalencies………………………………………………………………………….................................190Models for the Tax Treatment of Individual Accounts…………………………………............................192Summary of Tax Policy Choices…………………………………………………………. ...............................193Potential Designs of Individual Accounts…………………………………………………...........................194Implications of the Design and Purpose of Individual Accounts………………………...........................195Summary………………………………………………………………………………… ...................................201Appendix A: Current Rules for Taxing Savings……………………………………….. ...............................203Appendix B: Tax Treatment in Account Proposals………………………………………............................206

GLOSSARY…………………………………………………………………………………….......................................211REFERENCES…………………………………………………………………………………. ......................................217

vii

List of Figures

1-1: Categories of Individual Account Plans by Source of Funds and Nature of Participation .......................62-1: Shares of Income from Specified Sources, 2002:

Married Couples and Unmarried Persons Age 65 and Older ...................................................................242-2: Percent Receiving Specified Sources of Income, 2002:

Married Couples and Unmarried Persons Age 65 and Older ....................................................................242-3: Shares of Income from Specified Sources by Income Level, 2002:

Married Couples and Unmarried Persons Age 65 and Older ....................................................................252-4: Social Security Benefits Compared to Past Earnings by Earnings Level, 2004:

Retired Workers Age 65 ................................................................................................................................262-5: Role of Social Security in Total Income, 1976 and 2002:

Married Couples and Unmarried Beneficiaries Age 65 and Older ...........................................................272-6: Social Security Benefit and Replacement Rate, 2004 and 2030:

Scaled Medium Earner at Age 65 and at Normal Retirement Age ..........................................................282-7: Social Security Net Replacement Rate, 2003 and 2030: Medium Earner Age 65....................................292-8: Pension Coverage Trends, 1979-1998: Private Wage and Salary Workers ...............................................302-9: Pension Coverage by Occupation, 2000: Private Employees .....................................................................312-10: Ownership of Tax-Favored Retirement Accounts by Family Income, 2001 .............................................322-11: Ownership of Tax-Favored Retirement Accounts by Age, Race and Ethnicity, 2001 ..............................332-12: Asset Ownership by Age, Race and Ethnicity, 2001....................................................................................332-13: Homeownership, Market Value, and Home Equity by Age, Race and Ethnicity, 2001 ..........................352-14: Credit Card Debt by Family Income, 2001 ...................................................................................................352-15: Annual Earnings in Social Security Covered Employment, 2001: Wage and Salary Workers ...............372-16: Wage Levels and Age-65 Replacement Rates, 2003 and 2030: Scaled Illustrative Earners ....................382-17: Size of Account Balance and of Life Annuity at Age 65 as a Percent of Annual Earnings

by Contribution Rate and Duration of Contributions: Scaled Medium Earner.......................................392-18: Percent of Families without a Checking or Savings Account by Race and Ethnicity,

Age, and Income, 2001 ..................................................................................................................................402-19: All Tax Units and Non-Filers of Personal Income Tax by Income, Family Status, and Age, 2002...........413-1: Cohort Life Expectancy of Americans Age 65 in 2005 ...............................................................................463-2: Cost-of-Living Adjustments in Social Security Benefits, 1975-2004...........................................................473-3: Retirement Payout Rules in Selected Individual Account Proposals.........................................................493-4: Five Strategies to Make Income Last a Lifetime:

Purchasing Power of Income Stream from Age 65 to 115 ........................................................................503-5: Effect of Inflation Indexing and Survivor Protection on Initial Monthly Annuity

for $10,000 Premium: 65-Year-Old with 65-Year-Old Spouse....................................................................523-6: Effect of Inflation Indexing and Guarantee Features on Initial Monthly Annuity

for $10,000 Premium: 65-Year- Old ..............................................................................................................533-7: Joint-Life Annuity and Traditional Social Security for One-Earner Couple Age 65:

Payments as a Percent of Amount for a Single Worker.............................................................................623-8: Joint-Life, Inflation-Indexed Annuities: Dual-Earner Couples Both Age 65.............................................643-9: Effect of Spouse’s Age on Initial Joint-Life, Inflation-Indexed Annuity for $10,000

Premium: Retiree with Spouse Age 53 or Age 77 ......................................................................................653-10: Timing of Annuity Purchase and Widowhood: Joint-Life Annuities or Delaying

Annuity Purchase Beyond Widowhood ......................................................................................................663-11: Full and Partial Annuities, Bequests, and Guarantees: Single Individual.................................................683-12: Relative Size of Fixed, Single-Life Annuity by Age Purchased...................................................................694-1: Current Private Annuity Market and State Regulation: Actual Division of Responsibilities..................834-2: Annuity Arrangements in the Thrift Savings Plan: Actual Division of Responsibilities.................................84

viii

List of Figures (continued)

4-3: Private, Inflation-Indexed Annuities, Backed by TIPS: Hypothetical Division of Responsibilities ..........91

4-4: Private Risk Bearing and Federal Administration: Hypothetical Division of Responsibilities ................93

4-5: Federal Annuity Provider and Private Investment Management:

Hypothetical Division of Responsibilities .....................................................................................................94

5-1: Pre-Retirement Withdrawal Rules in Selected Individual Account Proposals........................................101

5-A: Federal Rules and Tax Treatment of Withdrawals and Loans From IRAs, 401(k)s,

and the Thrift Savings Plan ........................................................................................................................113

6-1: Number of Beneficiaries by Basis for Entitlement, December 2001 .......................................................118

6-2: Type of Social Security Benefits Received 2000 and 2040: Women Age 62 and Older .......................119

6-3: Spousal Protection Rules in Selected Individual Account Proposals ......................................................126

6-A: Spousal Rights and Benefits in Various Retirement Savings and Pension Plans....................................138

7-1: Disabled Workers’ Reliance on Social Security as a Share of Family Income.........................................146

7-2: Disabled Workers’ Reliance on Social Security by Living Arrangements................................................146

7-3: Probability of Becoming Disabled or Dying Before Age 67: Persons Age 20 in 2004 ..........................147

7-4: Disability Benefit Rules in Selected Individual Account Proposals..........................................................148

7-5: Blended PIA for Worker Disabled One-Quarter of Working Life ...........................................................151

7-6: Blended PIA for Worker Disabled Three-Quarters of Working Life .......................................................152

7-7: Annual Income From $55,000 UK Lifestyle Annuity Purchase.................................................................154

8-1: Social Security Benefits for Children of Deceased Workers Compared to Past Earnings, 2004...........159

8-2: Percent of Families with Life Insurance, 2001 ...........................................................................................161

8-3: Disabled Adult Child Beneficiaries, December 2002.................................................................................164

8-A1:Age of Disabled Adult Child Beneficiaries, 2001 ......................................................................................166

8-A2:Age of Disabled Adult Child Beneficiaries Awarded Benefits, 2001 ......................................................166

8-A3:Disabled Adult Child Beneficiaries by Diagnostic Group and

Representative Payee Status, December 2001...........................................................................................166

8-A4:Poverty Status of Disabled Adult Child Beneficiaries, December 1999 ..................................................166

9-1: Categories of Individual Account Plans by Source of Funds and Nature of Participation ...................171

9-2: Examples of Worker-Specific Offset Designs .............................................................................................172

9-3: Examples of Offsets by Size of Hypothetical Account

Relative to Actual Individual Account Annuity.........................................................................................173

9-4: Individual Account Splitting at Divorce .....................................................................................................179

9-5: Blended PIA for Worker Disabled One-Quarter of Working Life ...........................................................180

10-1: Equivalence of Roth IRA and Deductible IRA Models..............................................................................190

10-2: Value of Consumption for Immediate Consumption Versus Saving.......................................................191

10-3: Negative Rates of Tax when Saving Financed Through Borrowing .......................................................191

10-4: Categories of Individual Account Plans by Source of Funds and Nature of Participation ...................196

10-B: Tax Treatment Rules in Selected Individual Account Proposals ..............................................................206

1Chapter

Introduction and Summary

1

Calls for new individual savings accounts as partof federal retirement policy have come from var-ious quarters, either as part of some SocialSecurity reform proposals or as saving vehiclesfor individuals who do not have access toemployer-sponsored pensions. Individual devel-opment accounts and other initiatives to helplow-income people save also demonstrate grow-ing public interest in the issue.

Much of the work on individual accounts aspart of Social Security proposals has focused onhow individuals would save and manage theassets in the accounts during their working lives.Less attention has been paid to how and underwhat circumstances funds could be withdrawnfrom these accounts. For example, what condi-tions, if any, would permit individuals to with-draw funds before retirement? Wouldindividuals be required to convert account bal-ances into lifetime annuities at retirement, orwould they be allowed to access funds at what-ever time and for whatever amount they wished?Payout schedules and pre-retirement with-drawals affect other family members, an issuethat raises the question of what kinds of spousalrights would be recognized, and how these

rights would be applied in the event of divorce,retirement, or death. This study identifies pay-out issues raised by individual accounts in apublic retirement system and analyzes the poten-tial implications of different policy choices.

The Panel did not attempt to reach consensus onthe desirability or feasibility of individualaccounts in federal retirement policy. Panelmembers continue to hold sharply divergentpoints of view about personal accounts, particu-larly with regard to replacing any part of SocialSecurity with individual accounts. Members doagree that the choices described in this reportconstitute essential issues on payouts from suchaccounts.

This report summarizes work conductedbetween October 2002 and November 2004 bya non-partisan panel of nationally recognizedexperts led by co-chairs Kenneth S. Apfel, of theLBJ School of Public Affairs at the University ofTexas, and Michael J. Graetz, of Yale LawSchool. The Panel created a framework for ana-lyzing how benefits might be paid in a nationalsystem of new individual retirement accounts.The Panel considered individual accounts creat-

2 NASI • Uncharted Waters: Paying Benefits From Individual Accounts in Federal Retirement Policy

ed within Social Security as well as proposals foraccounts separate from and supplemental toSocial Security.

This introductory chapter explores key featuresof social insurance and private property, twoimportant components of retirement security inthe United States. The chapter presents a frame-work for analyzing payout issues and offers aclassification of individual account plans basedon some of these attributes. A brief summary ofSocial Security finances and solvency projectionspresents a backdrop for the Panel’s delibera-tions. Distinctions are drawn between reductionsin scheduled benefits in response to solvencyissues and reductions to accommodate the cre-ation of individual accounts. The chapter con-cludes with highlights of report findings thatcover financial demographics of American fami-lies, payout issues at retirement, institutionalarrangements for selling annuities to retirees,issues about access to accounts before retire-ment, spousal rights, implications of account

payouts for disabled workers and their familiesand young survivor families, issues in the designof worker-specific offsets, and potential taxtreatment of accounts.

Social Insurance and Property

Some Social Security proposals call for creatinga system of individual accounts as part of theSocial Security retirement program. Individualaccounts are typically considered to be personalproperty, while the traditional Social Securityprogram is social insurance. Both personallyowned property and social insurance are impor-tant components of retirement security; each hasparticular strengths, but they differ in importantrespects.

PropertyOwning and controlling property is the main-stay of a capitalist economy. Individuals areencouraged to own property – land, buildings,financial resources, or other types of assets – not

Purpose of the Uncharted Waters Study Panel Dispassionate Analysis, Diverse Views, and Varied Expertise

The Uncharted Waters Study Panel was convened by the National Academy of Social Insurance to promotedialogue and analysis by scholars who bring highly diverse expertise, knowledge, and philosophical perspec-tives to a relatively unexplored set of questions about payout issues in individual accounts. The Panelincludes experts in Social Security, pensions, private retirement savings, wealth building for low-incomeworkers, private insurance, social insurance, disability income policy, family benefit policy, tax policy, finan-cial markets, and federal and state regulation of financial intermediaries.

Panel members have very different personal views about the appropriate role of individual savings accountsin Social Security. Some panelists believe strongly that such accounts in some form are a very good idea.Other panelists believe strongly that any such accounts are a very bad idea. The Panel was not asked toresolve these differences, and it did not.

Rather, the purpose of the Panel is to bring its talent and knowledge to analyze in an even-handed way var-ious issues that arise in designing the payout side of any new individual account system. The goals are tohelp policymakers identify and begin to resolve a range of policy questions. The scope of the inquiryincludes accounts that aim to replace part of Social Security and other new savings vehicles – such as indi-vidual development accounts or new retirement savings vehicles – that are outside the Social Security sys-tem altogether. The Panel succeeded in finding common ground to identify and analyze payout issues in aclear, informative, and dispassionate way.

Chapter One: Introduction and Summary 3

only to stimulate economic well being but alsoto help raise one’s standard of living. Propertyownership can enhance self-reliance and person-al wealth can help secure one’s own future andthe future of one’s heirs.

Property ownership is essentially a bundle ofrights created by law. Individual ownership gen-erally implies control of the owned asset (and toexclude others’ rights to that asset), and owner-ship grants the holder wide discretion in assetconsumption. However, these rights may be lim-ited by the nature of the property right,1 by reg-ulations, spousal rights, creditors’ claims, orwhen owner rights would reduce or infringe onthe rights and security of others.

Property ownership carries with it a certainamount of risk. The assumption of risk is a keycomponent of a capitalist ownership system,with greater rewards generally related to greaterrisk. Property owners can buy private insurancefor some types of property risks, such as fires ortheft, but some economic security risks, such asbecoming disabled or living to very old age, areless commonly insured in the private market.

Social InsuranceLike property ownership, social insurance seeksto preserve individual dignity and self-reliance,although methods differ for accomplishing thesegoals. Social insurance emerges, in part, as aresponse to market failure in private insurance(Graetz and Mashaw, 1999). Other rationalesfor social insurance build on the notion that acompetitive economy sometimes fails to providefor all individuals, exposing them to risks out-side their control and not commonly insured bythe private market. Some workers earn lowwages over their entire work careers and cannotsave adequately for retirement, while others facecircumstances that significantly derail their abili-ty to save. A prolonged period of involuntaryunemployment, sickness, or incapacity candeplete whatever savings have been set aside forthe future. Social insurance, through universalparticipation, pools risks broadly to provide abasic level of economic security to all.

Social insurance has played an important role inmany nations by protecting individuals fromrisks inherent in competitive economies. In theUnited States, social insurance programs com-pensate workers who are laid-off from their jobsor are injured on the job. Social Security, thenation’s largest social insurance program, pro-vides workers and families with benefits inretirement as well as protections against eco-nomic insecurity due to prolonged disability orthe death of a family worker. Social Securitybenefits are closely tied to work and past wagesfrom which contributions were paid.

Comparing FeaturesThe Panel recognizes that there are importantdifferences between the social insurance featuresof Social Security and the ownership features ofretirement savings accounts. A brief comparisonof Social Security with voluntary employer-spon-sored 401(k)-type savings plans highlights someof the differences between social insurance andprivate property.

Key Function or Purpose

A 401(k)-type savings plan gives individuals andfamilies an opportunity to save for retirement ona tax-favored basis. Social Security providesbasic wage-replacement income in retirement foralmost all American workers and their spousesand widowed spouses. Social Security also pro-vides basic insurance protection when familieslose wage income due to the disability or thedeath of a worker.

Relationship between Contributions andPayouts

Owners of 401(k)-type accounts get out whatthey and their employers put in, plus investmentreturns, minus administrative costs. Investmentrisk is borne by account holders whose retire-ment payments depend on investment perform-ance. Some individuals may contribute morethan others, by choice or plan design. Further,choices made on fund investments and marketswings may produce substantial variations inreturns and payouts for individuals with similarcontributions.


Returns from Social Security may also vary overtime as legislation adjusts tax rates and benefitsto adapt, for example, to numbers of workersversus numbers of beneficiaries. Social Securitybenefits are based on a formula, with payoutsvarying depending on earnings level, years ofcovered work, and family situation. SocialSecurity pays relatively more for a given level ofearnings and contributions to: (a) low earners,whose monthly benefits replace a larger share ofpast earnings; (b) some widowed and divorcedspouses, who receive benefits without payingadditional contributions; (c) disabled workersand young families of deceased workers, whohave disability and survivor protection againstthese risks; (d) larger families, because addition-al benefits are paid for children without requir-ing additional contributions; and (e) people wholive a long time into advanced old age, who ben-efit from the guarantee of inflation-indexed ben-efits that last for life. By the same token, groupswho receive less relative to past wages and con-tributions have the opposite characteristics; theyare higher earners, dual-earner couples, singleworkers, childless workers, and workers whodie early without family members eligible forsurvivor benefits.

Terms for Contributing Funds

Individuals have a choice whether to contributeto employer-sponsored 401(k)-type savingsaccounts largely because these accounts are inaddition to the basic retirement income providedby Social Security. While matching funds mayencourage workers to contribute, workers retainfree choice about whether to put money into theaccounts. Workers may also choose how muchto contribute, subject to caps in plan rules andfederal tax rules.

In contrast, Social Security contributions (ortaxes) are mandatory. Workers do not have achoice to opt out.2 Employers are required towithhold Social Security contributions fromworkers’ wages and to pay matching amounts.The law sets the level of contributions for allworkers in relation to their wages or self-employment income. Making everyone con-

tribute protects individuals from their ownshortsightedness or bad luck and is consistentwith a system that pools and redistributes funds.If contributions were voluntary, higher-incomepersons who believe they have a less than aver-age likelihood of benefiting from the systemmight opt out, leaving lower earners to pay alarger share of the cost (Diamond, 2004;Langbein, 2004).

Terms for Withdrawing Funds

In 401(k)-type retirement savings plans, accountholders have wide latitude in choosing when andhow to withdraw their funds. Participants canwithdraw money at almost any time, as long asthey pay required taxes and, in some cases ofwithdrawals before a particular age, a 10 per-cent tax penalty. The penalty is designed to dis-courage pre-retirement withdrawals, butparticipants can usually access their funds –either by taking out loans from the accounts orwhen leaving their jobs. At retirement, partici-pants have many choices about the form of pay-outs, including leaving the money in the accountuntil age 701/2, taking it out in phased with-drawals, buying a life annuity, or withdrawing itin a lump sum.

In contrast, the choices for payouts in SocialSecurity are very limited, are set in law, and pro-mote ease of administration. Participants’ onlychoices are whether to accept the benefits theyare entitled to and when to begin retirementbenefits between ages 62 and 70. No optionexists to take the retirement money out early, toborrow against it, or to get it in any form otherthan monthly benefits. The lack of choice couldbe seen as a shortcoming, or as a way to protectindividuals against unforeseeable risks.

Tradeoffs in Blending ConceptsPolicy proposals that blend concepts of socialinsurance and private property face tradeoffs indeciding which model to follow in particular sit-uations or how to fit the two models together. Inthe chapters that follow, a recurring theme inconsidering payout rules for individual accountsthat replace part of traditional Social Security


benefits is how to blend concepts of social insur-ance with concepts of personal ownership.Different perspectives emerge in consideringissues on bequests, longevity insurance throughthe purchase of life annuities, tradeoffs betweenfree choice and mandates in the timing and formof payouts before retirement, spousal rights, andhow to preserve desired disability and life insur-ance for young families if part of Social Securityis being shifted from social insurance to privateproperty.

Framework for AnalyzingPayout Rules

The Panel believes that policymakers’ decisionsabout payout rules for any new system of indi-vidual accounts will differ depending on: theintended use of the accounts; the level of tradi-tional Social Security benefits that accompanythe accounts; the source of funds for theaccounts; and whether participation in theaccounts is mandatory or voluntary.

The Intended Use of Individual AccountsIf the main purpose of individual accounts—when combined with traditional SocialSecurity—is to provide basic financial securityduring retirement to individuals and their familymembers, then individual account payouts mightaim to resemble features of traditional SocialSecurity, with an emphasis on payments for life,family protection, and inflation protection. Yet,if the main purpose of the accounts is to helpbuild financial wealth, then payout rules mightresemble rules that apply to other discretionarysavings, such as individual retirement accounts(IRAs) or 401(k) plans. And, if the main pur-pose is to build funds to invest in human capitalor business enterprise before retirement, thenpayouts should be designed to target these purposes.

The Level of Remaining Traditional SocialSecurity BenefitsPayout rules for individual accounts intendedfor retirement might differ depending on the

level of traditional Social Security benefits thataccompany the accounts. If Social Securitydefined benefits are thought to meet basic ade-quacy goals, more discretion in payouts fromindividual accounts might be called for. Yet, ifthe account proceeds are viewed as an integralpart of basic Social Security retirement incomeprotection, more restrictions on payouts mightbe called for.

The Source of Funding for the AccountsWhether Social Security retirement benefits areadequate, too meager, or too generous is not atopic of this report. However, if a portion of thecurrent scheduled Social Security contributionsare used for individual accounts, there might bea stronger case for designing payouts to providesome of the protections found in traditionalSocial Security benefits. Yet, if accounts arefunded with new contributions from workers,more discretion in payouts might be in order.Also, the source of contributions to the accountsand the tax treatment of those contributions arelikely to affect views about tax treatment of pay-outs from the accounts.

Voluntary or Mandatory ParticipationThe case for flexible payout rules is strengthenedif policymakers want to encourage contribu-tions. Voluntary participation may not be con-sistent with restrictive rules designed to achievebasic security. Highly restrictive payout rulescould discourage individuals from participatingat all or cause them to contribute less than theywould if they had more choices about payouts.

The following section describes a typology ofplans based on some of these attributes.

Examples of Individual AccountPlans

A host of different kinds of individual accountshave been proposed for different purposes andthey could be grouped by any number of criteriadepending on the scope of the discussion. Forsome of its deliberations, the Panel found it use-ful to classify proposals along two dimensions:


whether contributions to accounts would bemandatory or voluntary; and whether theaccounts would be funded with new earmarkedcontributions from workers, or by using current-ly scheduled Social Security taxes, or by someother means, such as general revenues, as illus-trated in Figure 1-1.3

The Panel also agreed that when discussing pay-outs from individual accounts, a key issue iswhether proceeds from the accounts are meantto replace part of traditional Social Securityretirement benefits or are intended to providenew retirement resources. This distinction alsoemerges in the typology in Figure 1-1.

In Figure 1-1, the first category (1) includesplans that create individual accounts withmandatory new contributions. Examples ofSocial Security proposals with these attributesgenerally view the proceeds from the accounts aspart of Social Security retirement benefits. Onesuch plan, the Individual Account plan, was rec-ommended by Chairman Edward Gramlich ofthe 1996 Advisory Council on Social Security.That plan would scale back traditional benefitsto a level that could be financed with currentlyscheduled Social Security taxes of 12.4 percentof wages. The plan would then require workersto pay an additional 1.6 percent of their wagesto individual accounts. Proceeds from thoseaccounts were envisioned as part of SocialSecurity benefits (ACSS, 1996; NASI, 1996).Business leaders associated with the Committeefor Economic Development also proposed aSocial Security solvency plan along these lines in

their 1997 report, Fixing Social Security (CED,1997). Their plan calls for further reductions inscheduled Social Security benefits, and requiresboth workers and employers to pay an addition-al 1.5 percent of workers’ wages (for a total of3.0 percent of wages) to fund individualaccounts. Again, account proceeds were envi-sioned as part of Social Security retirement benefits.

The second category (2) of Figure 1-1 includesplans that call for mandatory participation usingpart of existing Social Security taxes to financeindividual accounts. Proceeds from theseaccounts are also generally viewed as part ofSocial Security retirement income. A subset ofthe 1996 Advisory Council on Social Security,led by Sylvester Schieber and Carolyn Weaver,proposed one such plan, the Personal SecurityAccount Plan. This plan would shift 5.0 percent-age points of employees’ share of Social Securitytaxes to individual accounts and scale back tra-ditional Social Security to a flat benefit. TheNational Commission on Retirement Policy, inits 1999 report The 21st Century RetirementSecurity Plan (NCRP, 1999), also recommendeda plan in this category. Co-chairs of theCommission were Senator Judd Gregg, SenatorJohn Breaux, Representative Jim Kolbe, andRepresentative Charles Stenholm. The planwould scale back traditional Social Security ben-efits so that they could be financed with a SocialSecurity tax of 10.4 percent of wages. Theremaining 2.0 percent of current Social Securitytaxes were allocated to individual accounts on amandatory basis. Subsequent legislation co-

Figure 1-1. Categories of Individual Account Plans by Source of Funds and Nature of Participation

New Earmarked Currently Scheduled Unspecified General Nature of Contributions for Social Security Taxes RevenuesParticipation the Accounts for Accounts for Accounts

Mandatory Participation (1) (2)(5)

Voluntary Participation (3) (4)


sponsored by Representatives Kolbe andStenholm in the 108th Congress (HR 3821)built on the Commission’s recommendations.

The bottom sections of Figure 1-1 include planswith voluntary participation. Category (3)includes proposals that involve voluntary newcontributions from workers and may includematching funds from other sources. All of theseproposals envision the proceeds of the accountsas being separate from Social Security and itsfinancing. In this respect, President Clinton’sRetirement Savings Accounts of 2000 called fornew contributions from households and federalmatching funds for low-income households. Theplan did not address Social Security finances.The Social Security Plus plan, offered by formerSocial Security Commissioner Robert M. Ball in2003, would set up administrative mechanismsfor workers to voluntarily save on top of a sol-vent Social Security system. Ball’s solvency planfor Social Security did not depend on money inthe accounts. Finally, a new and expanded sys-tem of individual development accounts (IDAs)4

would also involve voluntary new contributionsfrom individuals, perhaps with matching funds,and would create accounts independent of SocialSecurity. The main purpose of IDAs has been toexpand opportunities for asset accumulation foreducation, buying a home, or setting up a business, but IDAs could include saving forretirement.

Category (4) of Figure 1-1 includes plans thatpermit workers to shift part of their SocialSecurity taxes into individual accounts. Theseplans generally consider the proceeds from theaccounts to be part of Social Security. At retire-ment, individuals who had chosen to shift taxesto personal accounts would incur an offset (areduction in scheduled Social Security definedbenefits) based on an amount linked to the con-tributions to their accounts. Examples of plansthat fit in this category include recommenda-tions from President Bush’s 2001 Commission toStrengthen Social Security. Other plans in thiscategory include Representative Nick Smith’sRetirement Security Act, introduced in 2002

(H.R. 5734, 107th Congress), then-Representative Jim DeMint’s Social SecuritySavings Act of 2003 (H.R. 3177, 108thCongress) and Senator Lindsey Graham’s SocialSecurity Solvency and Modernization Act of2003 (S. 1878, 108th Congress).

The last column of Figure 1-1, category (5),includes proposals for individual accounts fund-ed by general revenues or other non-earmarkedfunds. Representative Clay Shaw’s SocialSecurity Guarantee Plus Act of 2003 (H.R. 75,108th Congress) would allow workers to becredited with annual contributions from the gen-eral fund of the Treasury for personal accounts.While participation in this plan would be volun-tary, it is assumed that participation would beuniversal. Another plan, Representative PaulRyan’s Social Security Personal SavingsGuarantee and Prosperity Act of 2004 (H.R.4851, 108th Congress), guarantees that thecombination of Social Security benefits and pay-ments from individual accounts would be atleast equal to currently scheduled Social Securitybenefits through transfers from the general fundof the U.S. Treasury. Some plans in Categoryfour (4) also require unspecified general rev-enues to pay scheduled Social Security benefits.

Panel members hold very different views abouthow to analyze plans that rely on unspecifiedgeneral revenue transfers. The disagreement cen-ters largely on whether the need for large gener-al revenue transfers would result in pressure tofurther reduce traditional Social Security bene-fits, or whether the funding for such transferscould be accommodated from other sources,such as income taxes, reduced spending on otherprograms, or from an increase in public debt.

Social Security Finances andSolvency Projections

While the Panel did not evaluate Social Securitysolvency, Panel members agreed that the long-range shortfall in Social Security finances was animportant backdrop for our deliberations. SocialSecurity retired-worker, disability, and survivor


benefits are financed mainly by earmarkedSocial Security taxes. Workers and employerseach pay 6.2 percent of workers’ earnings up to$90,000 in 2005, for a total of 12.4 percent.The earnings cap subject to Social Security taxesrises each year to keep pace with economy-widewages. The tax rate is scheduled to remainunchanged in the future. Currently the SocialSecurity trust funds take in more in revenuesthan are paid in benefits, and consequently arebuilding reserves. The reserves were $1.5 trillionat the end of 2003, according to the 2004 reportof the Social Security Trustees.

The Trustees project that tax revenue flowinginto the trust funds will exceed outgo until2018, under their intermediate, or best estimate,assumptions. After that, Social Security tax rev-enues plus interest earned on the Treasury bondsin the funds will exceed all benefit paymentsuntil 2028. Through the redemption of Treasurybonds plus Social Security tax revenue and inter-est income, scheduled Social Security benefitscan be paid in full until 2042, at which time thetrust funds are projected to be depleted. If nochanges are made to the program, taxes cominginto Social Security are expected to cover about73 percent of the scheduled benefits. By 2078,the end of the 75-year projection period used bythe Social Security Trustees, revenues are pro-jected to cover about 68 percent of scheduledbenefits.

Social Security solvency proposals address thislong-term funding shortfall in various ways,generally by reducing scheduled benefits (such asby modifying the benefit formula, raising the fullbenefit age, or altering automatic cost-of-livingadjustments in benefits) or by increasing rev-enues (such as by raising the Social Security taxrate, lifting the cap on wages subject to SocialSecurity taxes, or earmarking other revenues forSocial Security), or by using a combination ofsuch measures.

This Panel’s charge was not to recommend waysto achieve balance in Social Security. Rather, ourpurpose was to help policymakers think through

payout issues that arise in various types of pro-posals that would introduce individual accountsas part of Social Security. We also consider pay-out issues that might arise if a new system ofindividual accounts were set up separate fromSocial Security.

Benefit Changes for Solvencyand Benefit Offsets

Given this Panel’s focus on payout issues, asopposed to the restoration of solvency to SocialSecurity, we distinguish between reductions inscheduled defined benefits designed solely tohelp achieve solvency, and other reductions intraditional defined benefits that flow from deci-sions to shift part of currently scheduled SocialSecurity taxes to personal accounts. These latterreductions are called “offsets.”

Reductions in Scheduled Benefits toAchieve SolvencyMany of the plans in categories (1), (2), (4) and(5) of Figure 1-1 call for reductions in scheduledbenefits for the purpose of putting SocialSecurity in long-run financial balance. Thesebenefit reductions take many forms and thereductions could apply to all beneficiaries (forexample, by reducing scheduled benefits acrossthe board) or they could target particular sub-sets of beneficiaries, such as early retirees, highearners, dependent spouses, children, and soforth.

Benefit OffsetsPlans that shift scheduled Social Security taxesto individual accounts, as illustrated in cate-gories (2) and (4) of Figure 1-1, call for furtherchanges in scheduled benefits to accommodate,or “offset,” the partial shift of scheduled SocialSecurity taxes to personal accounts. These off-sets may take different forms depending onwhether the shift of taxes is mandatory and uni-versal – as is the case with proposals in category(2) of Figure 1-1 – or whether it is voluntary –as is the case in proposals in category (4) ofFigure 1-1.


Across-the-Board Offsets

If accounts funded with scheduled SocialSecurity taxes are mandatory and universal, theoffset in defined benefits to accommodate thattax shift could also be mandatory and universal.Proposals in category (2) of Figure 1-1 fit thiscategory. That is, all Social Security contributorswould automatically have part of their SocialSecurity taxes put into individual accounts andall workers would be affected by across-the-board changes in defined benefits necessary tobalance the remaining defined benefit systemwith a smaller amount of Social Security taxrevenues. The across-the-board changes couldtake many forms.

Worker-Specific Offsets

If workers have a choice whether to shift part oftheir Social Security taxes to personal accounts,then some mechanism is needed to personalizethe reduction in scheduled benefits. A worker-specific offset would ensure that only individualswho chose to shift their Social Security taxes toindividual accounts would have their traditionalSocial Security benefits reduced for this reason.These worker-specific offsets can be designed ina wide variety of ways and become a key aspectof payout issues. Proposals in category (4) ofFigure 1-1 fit this category and involve worker-specific offsets. These offsets are discussed inChapter Nine.

The Panel believes that the analyses in the chap-ters that follow make important headway inexploring the relatively uncharted waters gov-erning payouts if a new system emerges thatblends property concepts with social insurance.The chapters also provide insights for designingpayouts in property-based systems that are sepa-rate from social insurance.

Financial Demographics

As a backdrop for considering payouts from anew system of individual accounts, Chapter Twoexamines the role of Social Security in theincomes of retirees, recent developments in pen-

sions, and lessons from experiments to help low-income workers save.

Role of Social SecuritySocial Security is the major source of income formost retired Americans. About 90 percent ofpeople aged 65 and older receive benefits. Fortwo in three of those beneficiaries, SocialSecurity is half or more of their total income.Women without husbands are the most relianton Social Security benefits. For three in four eld-erly unmarried women receiving Social Security,the benefits are more than half their income. Fornearly three in ten of such women, SocialSecurity is their only source of income.

Social Security benefits alone do not provide acomfortable level of living. The average benefitfor a retired worker was about $922 a month,or $11,060 a year in 2004. Under current SocialSecurity law, benefits for future retirees arescheduled to rise in real terms. Benefits willgrow somewhat more slowly than earnings,however, because the 1983 law raised the “fullbenefit age” from 65 to 67. That law phases inover the next 20 years. Although the real levelof benefits will be higher, benefits for 65-year-old retirees will replace a smaller share of priorearnings than is the case today or at any time inthe last 30 years. Because Social Security is notin long-run financial balance, other changesmight be enacted that would either raise revenueor lower benefits.

Pension TrendsEmployer-sponsored pensions are an importantsupplement to Social Security for the half ofmarried couples and one third of unmarried menand women age 65 and older who receive pen-sions. At any time over the past 25 years, abouthalf of private-sector workers have been coveredby pension plans. The form of these plans hasshifted dramatically from the 1970s and 1980swhen defined-benefit plans were dominant.Today, defined-contribution plans, such as401(k) plans, are more common. In defined-con-tribution plans, workers have more choicesabout whether to participate and how much to


contribute; they can take the accounts with themwhen they change jobs; and they have morechoices about when and how to withdraw themoney. At the same time, workers take on moreresponsibility for financing the plans and bear-ing the investment risk that employers bear indefined-benefit plans. Today, about half of allU.S. families own a tax-favored retirementaccount. The median value of the accumulatedbalances in those accounts was $29,000 in2001. For the 59 percent of families headed bysomeone aged 55 to 64 who have such accounts,the median value was about $55,000.

Experience with Individual DevelopmentAccountsMany Americans lack experience with financialinstitutions. This lack of financial experiencemerits attention in the design of a new individ-ual account system. The size of the “unbanked”population – those who do not have a checkingor savings account with a bank or credit union –is estimated to be between 10 and 20 percent ofall U.S. families. Low-income and minority fam-ilies are most likely to be without a connectionto a financial institution.

Individual development account experimentshave offered financial education and matchedsavings to low-income workers. The savings areearmarked for specific purposes, such as highereducation, purchase of a first home, or startinga business. Conditions that appear to foster suc-cessful saving include: (a) access to a savingsplan, (b) incentives through matching funds, (c) financial education, (d) ease of savingthrough direct deposit and default participation,(e) clear saving targets and expectations, and (f)restrictions on withdrawals.

Payments at Retirement

Chapter Three examines financial risks retireesface and how life annuities can insure againstthose risks. It offers four illustrative options forpayout rules at retirement and examines theimpact of various annuity features on costs to

retirees, the interests of heirs, and implicationsfor consumer education.

Life Annuities Insure Against FinancialRisksRetirees face at least four sources of financialuncertainty. They do not know how long theywill live (longevity risk), how long their spousemight live (spousal survivorship risk), howprices might rise in the future (inflation risk),nor what returns they will earn on their savings(investment risk). To illustrate longevity risk,while the average 65-year-old woman can expectto live 20 years, she has a 7 percent chance ofdying within five years and a 14 percent chanceof living for 30 years to her 95th birthday. Toillustrate inflation risk, even modest priceincreases of just 3 percent per year will make$100 today worth only about $74 in ten years;after 25 years, the value would drop by morethan half, to $45. High and unexpected inflationcould rapidly erode buying power of any givenamount of money.

A life annuity is a financial product offered byan insurance company that promises paymentsfor as long as the annuitant lives. When an indi-vidual buys a life annuity, the insurance compa-ny has a contractual obligation to pay theannuitant a guaranteed income for life. Theannuity purchase shifts the individual’s longevityrisk and investment risk to the insurance compa-ny. Because insurers pool mortality risk among alarge group of annuitants, the extra funds fromannuitants who die early are used to cover theannuity costs of individuals who live a longtime. From the annuitant’s perspective, thedownside of buying a life annuity is that the fullprice is paid up front and the purchase is irrevo-cable. Other strategies to spread money overone’s remaining life – such as taking phasedwithdrawals – do not guarantee the money willlast for life, but the account holder retains own-ership of the money.

Policy OptionsShould retirees be encouraged or required to buylife annuities with their individual accounts?


Chapter Three presents four illustrative options.The first gives retirees Unconstrained Access totheir account funds. It offers many choices andis based on the federal employees’ Thrift SavingsPlan. The second option, Compulsory Annuitieswith Special Protections, falls at the other end ofthe spectrum. It would require the purchase oflife annuities that are indexed for inflation andthat automatically provide survivor benefits forwidowed spouses. A third option, DefaultAnnuities with Special Protections, makes theannuities of option two a default, but wouldallow other payouts. Finally, option four,Compulsory Minimum Annuities, would requirethe annuities of option two, but only up to agiven level.

Policy choices along this spectrum are likely tobe influenced by the purpose of the accounts,the level of Social Security defined benefits thataccompany the accounts, and whether participa-tion in the accounts is mandatory or voluntary.If the purpose of the accounts is to provide basicsecurity, then policymakers might want payoutsto resemble the mandatory protections of thesecond option. Many proposals for accountsthat aim to replace part of traditional SocialSecurity call for mandatory inflation-indexedannuities with spousal protections. Yet, if theaccounts are discretionary savings on top of tra-ditional Social Security benefits, then payoutsmight resemble the broader choices of optionone.

Additional Protection Costs MoreEach layer of protection for inflation-indexingand survivor benefits lowers the size of theannuity one can buy with a given account bal-ance. With $10,000, a 65 year-old retiree couldbuy a fixed life annuity of about $80 a month.If the annuity were indexed to keep pace withinflation at 3 percent a year, it would start outlower, at about $62 a month. If it would contin-ue to pay for as long as either the annuitant or a65-year-old spouse lived, the annuity wouldstart out lower still, about $50 a month. Theseprices are based on the assumption that every-one would be required to buy life annuities.5

Whether the purchase of life annuities should becompulsory is a key policy issue. Compulsoryannuities assure that people cannot outlive theirmoney, but allow retirees no choice.Compulsory annuities cost less, on average.Optional annuities cost more (or pay less forany given premium) because people with shortlife expectancies tend not to buy them.Compulsory annuities make higher payouts, onaverage, precisely because short-lived people arerequired to buy a product that is not a gooddeal for them.

Joint-Life Annuities Providing joint-life annuities that protect wid-owed spouses will reduce the size of the annuitythat a given premium will buy. Many choices arepossible, such as between symmetric and contin-gent joint-life annuities. For example, if Johnbuys a contingent joint and two-thirds annuity,the payment for his widow will fall to two-thirds of the original amount if he dies, but thepayment will remain the full original amount ifhe is widowed. In contrast, if he buys a symmet-ric joint and two-thirds annuity, the paymentwill always drop to two-thirds of the originalamount when one partner became widowed.Each annuity type has different pros and consthat policymakers might want to address.

Guarantees and Interests of HeirsSome annuity contracts guarantee a payment toa named death beneficiary if the annuitant diesshortly after buying an annuity. A ten-year-cer-tain annuity, for example, guarantees paymentsfor ten years even if the annuitant dies in lessthan ten years. A refund-of-premium annuityguarantees that the annuity will pay out at leastthe nominal purchase price. For example, if theannuitant paid $10,000 for a life annuity anddied after receiving only $1,000, then $9,000would be paid to the death beneficiary.

Guarantees lower the monthly annuity that agiven premium will buy. For $10,000, one couldbuy a single-life, inflation-indexed annuity of$62 a month. Adding a 10-year certain featurewould lower the monthly amount to about $58,


while a refund of premium annuity would lowerthe amount to about $55 a month. Manyexperts believe guarantee features are not a wisepurchase on purely economic grounds. Yetannuity buyers often choose guarantees, perhapsbecause the guarantees help their heirs avoid dis-appointment and serious regret if the annuitantpaid a large amount for a life annuity and diedsoon after.

Timing of Annuity Purchase and HeirsThe interests of heirs could influence the ques-tion of whether and when to buy an annuity.From a strictly selfish perspective, named benefi-ciaries might prefer that the accountholder delaybuying an annuity so that the account wouldremain inheritable. For example, an unmarriedaccount holder might name an adult child,friend or other relative as a death beneficiary. Ifthe account holder dies before buying an annu-ity, the entire balance would go to the heir. If theaccount is used to buy an annuity, the bequest isgone.

The timing tradeoff affects married retirees, too.If one spouse is expected to die relatively soon,the couple might be wise to delay or avoid buy-ing joint-life annuities. The survivor’s income inthe form of a single-life annuity based on thebalance in both accounts would be considerablyhigher than the survivor payment from joint-lifeannuities from both accounts.6 So, both singleand married retirees might want flexibility in thetiming of annuity purchase.

Recap of ChoicesRetirement payout policies present tensionsbetween offering choices and guaranteeingincome for life. Possible questions to be deter-mined by mandates or participant choicesinclude the following: whether to buy an annu-ity at all; how much of the account to spend onan annuity; whether the annuity will be indexedfor inflation; when to buy an annuity; whetherto buy a guarantee feature and, if so, what type;whether to buy a joint life annuity and, if so,whether to choose a contingent or symmetricproduct and what size survivor benefit to buy.

Informed ChoiceIf retirees have choices about buying annuities, akey policy issue becomes who will advise themand answer their questions. To what extentwould the educators or advisors be responsiblefor the consequences if the advice produced dis-appointing results? As retirees have more choic-es about retirement payouts, these questionsgain added importance.

The Social Security Administration has very littleexperience helping retirees make informed choices about payouts, because Social Securityoffers almost no choices. The only choices arewhether and when to take benefits once onebecomes eligible.

The federal government, in its role as employer,informs participants in the Thrift Savings Planabout payout options. Personnel offices provideseminars and explanations to employees whoare planning to retire. While some large privateemployers might be equipped to help employeesunderstand annuity choices in individualaccounts, many small employers would not havethe resources to do so.

Institutional Arrangements forProviding Life Annuities

The existing market for life annuities in theUnited States is relatively small. Life annuitiesare offered by insurance companies, which areregulated by states.

Life Annuity MarketMany financial products are called annuities,but are not life annuities. Life annuities are con-tractual obligations to pay the annuitant for therest of his or her life.7 Deferred annuities aretax-favored investment products that do notguarantee payment for life. More common thanlife annuities, deferred annuities are used mainlyto defer taxes on fund accumulations.

Life annuities represent about 15 percent ofannual new product sales of insurance compa-nies. Some experts believe that life annuities are


a growth area as pension plans shift to lump-sum payouts. But such growth has not yetoccurred, perhaps because of limited interestfrom both customers and financial advisors.Two drawbacks from advisors’ perspectives arethat life annuities generally pay smaller commis-sions than deferred annuities and life annuitiesend the opportunity to do further business withthe funds because the money is turned over toan insurance company.

Whether insurers would be allowed to chargedifferent prices to women and men is a key poli-cy issue. In the individual life annuity market,insurers charge women more because womenlive longer than men, on average. Yet, in thegroup annuity market, federal policy bans differential pricing in annuities tied to employeebenefits.

Adverse Selection, Uniform Pricing andSelective MarketingIn a voluntary annuity market, if a companyprices its annuities based on average risks, peo-ple with longer life expectancy would be morelikely to buy the annuities while people withshort life expectancies would not. This adverseselection would drive up the cost to the insurerand lead the company to raise its prices. Thehigher prices would further discourage short-lived people from buying annuities. If policy-makers wanted uniform pricing of annuities foreveryone of the same age (regardless of sex,health status, or other risk factors), the simplestway to avoid adverse selection would be toremove participant choice and require everyoneto buy annuities. Uniform pricing in the pres-ence of differential risks can lead to selectivemarketing, whereby annuity sellers target theirsales efforts on population groups with shorterlife expectancy. It is difficult for regulators tostop selective marketing without direct govern-mental oversight of marketing activities.

Insurance Company RegulationInsurance regulation in the United States hasbeen the purview of the states since enactmentof the McCarran-Ferguson Act in 1945. While

the federal government regulates the banking,securities, and defined-benefit pension industries,states regulate insurance companies. Such regu-lations cover the pricing of annuities, financialbacking of annuities, provisions for guaranteeingpayments in the case of insurance company fail-ure, and other issues.

Unlike federal insurance programs, such as theFederal Deposit Insurance Corporation forbanks, state guaranty funds for insurance com-panies are not pre-funded. Instead, states assess(that is, tax) other insurance companies doingbusiness in the state to cover the cost of aninsurance company failure after it occurs.8 Thelargest such failure involved Executive LifeInsurance Company in the early 1990s. Stateguaranty associations have paid about $2.5 bil-lion for that insolvency as of 2004.

Existing arrangements for guaranteeing lifeannuities might suffice for a new system of indi-vidual accounts if the accounts are viewed assupplemental savings and retirees are given widediscretion on how they take the funds at retire-ment. But new institutional arrangements arelikely to be needed if policymakers want tostrongly encourage retirees to buy life annuitiesindexed for inflation and that automatically pro-vide protection for widowed spouses.

Inflation-Indexed AnnuitiesA large market for inflation-indexed annuitiesdoes not yet exist in the United States and creat-ing one is likely to involve the federal govern-ment in some way. The government might issuea large volume of long-dated Treasury InflationProtected Securities (TIPS) to help insurancecompanies hedge inflation risk, it might reinsureprivate insurers or guarantee their solvency, or itmight issue inflation-indexed annuities directlyto retirees.

Some experts thought that a substantial marketin inflation-indexed annuities would evolvewhen TIPS were introduced in 1997. Three con-ditions might explain why that has not hap-pened. First, consumers may not see the value of


inflation-indexed annuities. Retirees simply maynot understand longevity risk and inflation risk.Second, TIPS may not exist in sufficient volume,duration, and predictability to encourage insur-ers to offer inflation-indexed annuities. TheTreasury Department stopped issuing all 30-yearbonds, including TIPS, in 2001. Insurers mightbelieve that only 30-year TIPS are sufficient tocover the life spans of new retirees. Thirty-yearTIPS are about $40 billion (or roughly one per-cent) of the total Treasury securities market of$3.3 trillion, which is about one third of thenation’s economic output, or gross domesticproduct (GDP). Finally, insurers and their regu-lators might be concerned that inflation index-ing would increase insurers’ exposure tomortality risk. Even if inflation risk is hedged byTreasury securities, insurers who underestimatetheir annuitants’ life spans will be exposed tomuch greater losses if the promised annuitieskeep pace with the cost of living.

The volume of reserves required to back wide-spread inflation-indexed annuities would be sub-stantial. Reserves backing annuities funded with2 percent of workers’ earnings could amount toabout 15 percent of GDP when the system isfully mature.9 Those annuity reserves would beequivalent to roughly 7 percent to 8 percent ofthe value of total U.S. financial assets.10

Options for Widespread IndexedAnnuitiesIf insurance companies were to provide annu-ities on a widespread basis, then policymakersmight want the federal government to beinvolved in insuring the solvency of those com-panies. Proposals for the federal government tocharter and regulate life insurance companiesmight gain broader interest in this case.

The government could issue TIPS in sufficientvolume and duration to back privately issuedannuities or it could provide inflation-indexedannuities directly to retirees. In the latter case,the government would take on the longevity riskand the inflation risk. Whether the governmentprovides annuities directly, or provides TIPS to

back privately issued annuities, the governmentcould be holding very large amounts of assetsbacking the annuities. A key question for policy-makers to address is who would manage andinvest the large volume of assets. New arrange-ments might be needed to segregate the fundsfrom other taxing and spending functions of thefederal government and new institutions mightbe needed to provide for prudent and diversifiedinvestment of the funds.

Pre-Retirement Access toIndividual Accounts

The pros and cons of allowing early access toindividual accounts will depend, in large part,on the intended use of the accounts, whetherpeople have any choice about whether to partici-pate, and whether the accounts are viewed aspersonal property. If the accounts are supposedto provide baseline economic security in old age,the case for banning early access is strong. Yet,if the purpose of the system is to expand oppor-tunities for voluntary retirement saving, thenearly access might encourage people to savemore than they otherwise would.

Precedents for Early Access Individual retirement accounts (IRAs) allowunlimited access as long as account holders paytaxes and, in certain cases, a 10 percent taxpenalty on amounts withdrawn. Employer-spon-sored 401(k) plans permit somewhat more limit-ed access, but employees can usually get themoney if they need it—through a loan or hard-ship withdrawal, or by leaving the job and cash-ing out the account. Most U.S. proposals thatenvision individual accounts as a partial replace-ment for Social Security retirement benefitswould totally ban early access to the money.

Tradeoffs Among Goals Early access rules create tensions among threecompeting goals: ease of access, retirement secu-rity, and administrative efficiency. Participantswill want easy access to their money when theyneed it. But the goal of retirement security calls


for minimizing leakage from the accounts bybanning early access. Yet, if access is allowed,the retirement security goal argues for restrictingaccess to only loans and only for hardship. Thecompeting goal of administrative efficiency alsoargues for a total ban on access. As a secondchoice, administrative efficiency points to theopposite policy of allowing unrestricted with-drawals. More administrative resources areneeded to process loans, which involve repay-ments, and to restrict reasons for withdrawals,which requires documentation, decisions, andperhaps a right to review when access is denied.

GatekeepingIf access to individual accounts is allowed butrestricted in some way, a gatekeeper will beneeded to determine whether a particular with-drawal is allowed. When access is denied, proce-dures will be needed to give participants anopportunity to have a denial appealed andreconsidered. Employers who sponsor 401(k)plans are responsible for deciding whetheremployees’ withdrawals or loans comply withrules of the plan and with the Internal RevenueCode. The employer bears the risk of losing tax-favored status for the entire plan in case ofwrongful determination, although the InternalRevenue Service can levy lesser penalties.

A new national system of individual accountswill pose new questions about: what entitywould play the gatekeeper role; what incentiveswould prompt the gatekeeper to prevent wrong-ful withdrawals; what penalty would beimposed for non-compliance; and on whom thepenalties should fall. If the overall purpose ofthe accounts is retirement income security, apenalty on the accountholder for a wrongfulwithdrawal might undermine the ultimate goal.

Third Parties and Means TestsFinally, early access to the accounts can be atwo-edged sword. Account holders’ access totheir own retirement funds may mean that thirdparties can also make a claim on the funds incases of bankruptcy, divorce, or unpaid federaltaxes. Further, some means-tested benefit pro-

grams treat accessible retirement funds as count-able assets for the purposes of determining bene-fit eligibility. In such cases, if the account holderhas access to the money, he or she must spend itto qualify for assistance.

No U.S. precedent yet exists for a total ban onaccess to individually owned retirement savingsaccounts. If policymakers create such a ban, his-tory suggests that they will face pressure to easethe restrictions. Sustaining limits on access toretirement funds that are required for incomesecurity, but that account holders view as theirown money, is an important issue and likely tobe an ongoing challenge.

Spousal Rights

About 14.0 million individuals – 30 percent ofall Social Security beneficiaries – receive benefitsbased at least in part on a spouse’s work record.These beneficiaries are overwhelmingly women.About 6.0 million women are entitled to SocialSecurity as workers and to higher benefits as awidow, wife, or divorced wife. Another 7.8 mil-lion women receive Social Security solely as wid-ows, wives, or divorced wives.

The cost of paying traditional spousal benefits isspread among all participants in Social Security;the benefits for a widow or wife do not lowerpayments to the husband. As personal property,individual accounts represent a finite pool ofassets, so that payments to a spouse wouldreduce funds for the accountholder and vice-versa.

Policy decisions about spousal rights to individ-ual accounts will be influenced by the purposeof the accounts, the level of traditional SocialSecurity benefits that accompany the accounts,and whether participation is mandatory or vol-untary. If participation is voluntary, spousalrights rules will need to take into account thepossibility that only one member of a couplemay elect to participate.

24367 NASI TEXT 1/17/05 11:36 AM Page 15


Federal or State JurisdictionA key question is whether spousal rights to indi-vidual accounts will be decided in federal law orleft to the states. As a national social insuranceprogram, Social Security has uniform benefitentitlement rules throughout the country. Statelaw has historically determined spousal rights toproperty, and states have distinctly differentapproaches. Common law states consider thetitle-holder to be the owner of property,although all such states call for an equitabledivision of property at divorce. The nine com-munity property states, in which 29 percent ofthe population resides,11 view property acquiredduring marriage as community marital propertythat belongs equally to husbands and wives.Holdings acquired before marriage and bequestsreceived during marriage are considered person-al property and outside marital property.

If spousal rights in an individual account systemare to be uniform, Congress will need to definethe rules clearly in federal law. Alternatively,policymakers could explicitly provide that statelaw will determine spousal rights. While thisapproach would increase flexibility, it alsowould produce different results across states,and would likely increase administrative costsand the need for account holders to have legalrepresentation.

Spousal Rights during MarriageDuring marriage, one option would be to divideaccount contributions equally between husbandsand wives, building community property princi-ples into the account system. Another approachwould be to credit each spouse with his or herown personal contributions. A related issue iswhether a married account holder would needspousal consent to take money out of theaccount or borrow it, if such access wereallowed at all. If a spouse has a future claim onthe account funds at widowhood or divorce,then spousal consent to use the funds for otherpurposes might be warranted. If a spouse hadno such claim, the case for spousal consentwould be reduced.

Spousal Rights at DivorceApproaches for allocating spousal rights atdivorce could be based on federal mandates ordefault rules. In addition, there could be a rolefor state courts to allocate, or reallocate, fundsas part of an overall divorce settlement.Questions for policymakers include: whetherfederal law would require equal division ofaccounts, or make equal division a default rule,and if so, whether the property division wouldapply only to new contributions and investmentearnings during the marriage or to the entireaccount balances. In addition, if accountsinvolve worker-specific offsets, how offsets arehandled at divorce becomes a key question.Whatever federal mandates or default rulesapply, a final issue is whether state courts wouldretain authority to allocate (or reallocate) fundsas part of an overall divorce settlement.

Rights at Widowhood before RetirementAnother key set of policy issues is whether wid-ows and widowers will automatically inherittheir deceased spouse’s account, or whetheraccountholders will be free to bequeath theiraccounts to whomever they choose. Some SocialSecurity proposals require that the accountsalways go to the widowed spouse and be heldfor her or his retirement. These rules aim to pro-tect widowed spouses in ways that resembleSocial Security survivor benefits, but could posenew issues in the case of subsequent marriages.For example, if a widowed spouse remarriedand subsequently died, the property interests ofchildren from a first marriage and rights of thesubsequent spouse might be in conflict. Whilefamily law deals with such issues, blendingsocial insurance survivor protections with com-munity property inheritance rights would posenew issues.

Retirement Payouts for Married AccountHoldersMany individual account proposals require mar-ried account holders to buy joint-life annuities inorder to protect widowed spouses. When aretiree has a much younger (or older) spouse,the age disparity will affect the size of joint-life


annuities that a given premium will buy, becausejoint-life annuities are affected by the age ofboth the annuity partner and the annuity buyer.

Changes in marital status after one buys anannuity could pose new issues in allocatingretirement income. In general, life annuities can-not be rewritten after purchase. So, if an individ-ual marries after buying a single-life annuity,there is no easy way to change the contract tocover a spouse.

Implementation IssuesAdministering spousal rights in a new system ofindividual accounts could impose new reporting,verification, and dispute resolution proceduresbeyond those used to determine Social Securitybenefit entitlement. Social Security spousal bene-fits are determined when benefits are claimed –when a worker retires, dies, or becomes dis-abled. Implementing property rights for individ-ual accounts could require new systems to linkhusbands’ and wives’ account records through-out the work life. A spouse’s right to individualaccount funds will likely incur more dispute-res-olution procedures than occurs with traditionalSocial Security payments.

Disabled Workers and theirFamilies

Social Security pays disability as well as retire-ment benefits, and the risk of disability is signifi-cant. Payout policies at disability onset willdepend on the purpose of the individualaccounts. Six options are explored in ChapterSeven. General rules about payouts from indi-vidual accounts may take on new dimensionswhen accountholders are disabled-worker beneficiaries.

Role of Social Security for Disabled-Worker BeneficiariesAbout six million individuals aged 18-64received disabled-worker benefits from SocialSecurity at the beginning of 2004. Those bene-fits account for more than half of total familyincome for about one in two disabled-worker

beneficiaries. When compared to other people ofthe same age, disabled-worker beneficiaries aremore likely to be black or Hispanic, unmarried,without a high school diploma, live alone, andto be poor or near poor.

The Risk of Disability The risk of becoming so disabled that onereceives Social Security disabled-worker benefitsis significant. About three in ten men, and onein four women, will become disability benefici-aries before they reach retirement age. Disabilityis not the last risk to income security that theywill face. The death of disabled workers beforeretirement may leave family members who reliedon their support, while those who live intoretirement will need to consider the resourcesthey will have in old age. In designing an indi-vidual account proposal, it is important to thinkthrough how the accounts, along with anyaccompanying changes in traditional SocialSecurity benefits, will affect disabled workersand their families throughout the rest of theirlives.

Policy Options for Disability Beneficiariesand Purpose Policy issues with regard to payouts from indi-vidual accounts for disabled-worker beneficiar-ies will vary depending on the purpose of theaccounts. If the accounts are intended to be dis-cretionary savings on top of Social Security, thenpayout rules might resemble IRAs and 401(k)s,which make the money available without penal-ty at the onset of disability.

Yet, if individual accounts become an integralpart of Social Security, and scheduled retirementbenefits are reduced in return for the new per-sonal accounts, new issues arise about whetherand how those offsets will apply to the tradi-tional benefits of workers who become disabled.Because disability benefits are based on the sameformula used for retirement benefits, across-the-board changes in the retirement benefit formulawould automatically affect disabled workersunless policymakers specifically address theseissues.12 Other policy questions relate to when


and how funds in the accounts would becomeavailable to disabled workers. Various policyoptions are explained in Chapter Seven.

Adapting General Rules to the Situationof Disability Beneficiaries Many of the issues covered in prior chapterstake on new dimensions when the general rulesapply to people who have experienced career-ending disabilities. Sustaining a ban on access toaccount funds before retirement age – as dis-cussed in Chapter Five – may pose new chal-lenges when disability beneficiaries have apressing need for the money, particularly if theyhave life-threatening conditions and have nofamily members with a survivorship interest inthe accounts. If retirees are required to buyannuities at normal retirement age, will dis-abled-worker beneficiaries be required to buythem on the same terms as other retirees? Ormight a market in “impaired life” annuitiesemerge, as has occurred in the United Kingdom?These products allow individuals who haveshorter life expectancy to buy annuities on amore favorable basis. Mandating joint-life annuities for married retirees could present newissues if one or both members of the coupleentered retirement as disabled-worker beneficiaries.

Children, Life Insurance, andBequests

Social Security proposals that call for individualaccounts to replace part of traditional retirementbenefits also involve questions about how theplan will affect young survivor families andother beneficiary families with young children.Because assets in individual accounts are notexpected to spread risk the way insurance does,it is important to examine how new accountsmight interact with Social Security benefits forchildren.

Children on Social Security

About three million children under the age of 18receive Social Security as survivors and depend-

ents of deceased, disabled, and retired workers.These children account for about 7 percent ofall Social Security beneficiaries and about 4 per-cent of all children in the United States. Abouthalf of the eligible children are survivors ofdeceased workers, while the others have a par-ent who is disabled or retired.

Disabled Adult ChildrenAdults who became disabled before age 22 areeligible for benefits on the same terms as chil-dren under 18. About 750,000 persons age 18and older with childhood onset disabilitiesreceive Social Security, as children of deceased,disabled or retired parents. Mental retardation isthe main diagnosis for most of these beneficiar-ies, while conditions of the nervous system orsensory organs are the next most prevalent.These beneficiaries range in age from youngadults to senior citizens. About six in ten dis-abled adult child beneficiaries are poor or nearpoor and about four out of five receive SocialSecurity through a representative payee becausethey are not able to manage their own funds.

Policy Options for Defined BenefitsMany plans for mandatory individual accountsin Social Security call for across-the-boardreductions (or offsets) in scheduled SocialSecurity retirement benefits that will phase in asthe accounts build up.13 If these changes weremade in the basic benefit formula for retirees,they would affect young survivor families aswell. But young survivor families may not bene-fit from individual accounts in the same waythat retirees do. Chapter Eight considers fourpossible approaches for adapting changes inretirement benefits to the particular situations ofyoung survivor families. It also considers howpolicymakers might approach benefit changesfor minor children and disabled adult childrenwhen the working parent is a disabled-workerbeneficiary or a retiree.

Children’s Rights to Parents’ AccountsWhether a minor child or a disabled adult childwould have any special rights to an accountwhen a parent dies is also an important ques-


tion. Wives and husbands typically have certaininheritance rights under state law. Would policy-makers want to specify any inheritance rightsfor minor children or disabled adult children?Or, should federal policy leave these decisionsabout bequests to working parents and to statelaws that apply when one dies without a will?

Bequests to Heirs other than Spousesand ChildrenIndividual account proposals generally allow theaccount holder to bequeath funds if the workerdies before retirement. At the same time, manysuch proposals limit bequests by requiringaccount holders to buy annuities or by automat-ically transferring accounts to widowed spouses.These limits on bequests are generally motivatedby a desire to preserve types of benefits thatSocial Security now provides, such as paymentsfor life and spousal protections. New bequestsare more likely to occur for unmarried accountholders (widowed, divorced, or single) who diebefore buying annuities. In the eyes of many,these bequests are desirable and consistent withproperty ownership. Yet, from a social insuranceperspective, such bequests could be viewed as“leakage” that is beyond the purpose of thesocial insurance system. To the extent that SocialSecurity funds go to heirs who would not other-wise be eligible for benefits (such as able-bodiedadult children, siblings, relatives, friends or insti-tutions), either more money would be needed topay other eligible beneficiaries, or their benefitswould be lowered in some way. In designingpayouts, policymakers have the opportunity toweigh tradeoffs between property rights andsocial insurance goals.


When workers can choose whether to shift partof their Social Security taxes into a personalaccount, some mechanism is needed to personal-ize the offset of scheduled benefits to equitablydistinguish between those who do and thosewho do not shift Social Security taxes to person-al accounts. These worker-specific offsets can bedesigned in a variety of ways, becoming a key

aspect of payout issues. Possibilities for design-ing offsets are almost limitless and this chapteroutlines some of the choices and questions.

Basic Design Issues In terms of basic design, should the offset reducethe account holder’s scheduled Social Securitybenefits, or should it reduce the size of his or herindividual account? Offsets that reduce sched-uled defined benefits require policymakers todecide which types of benefits would be affected(retirement or disability) and whether benefits offamily members (spouses, widowed spouses, andchildren) would be reduced.

At retirement, what event should trigger the cal-culation and application of a worker-specificoffset? Applying the offset when Social Securitybenefits are first claimed would ensure that noretirement benefits avoid the offset, but raisesthe question of whether contributions to theaccounts should end and instead go to the SocialSecurity trust funds when individuals keepworking after claiming retirement benefits.

Retired Couples When couples retire, a number of questions alsoarise about how the offset would apply in thecase of family benefits. A different sequence ofcalculating offsets and annuities could result indifferent outcomes. Rules for couples would alsoneed to take account of the possibility that onespouse chose to shift taxes to a personal accountwhile the other did not. Ideally, offset ruleswould be equitable to couples in which neither,both, or only one partner shifted taxes to a per-sonal account.

Offsets and DivorceAt divorce, if the proposal mandates (or per-mits) a division of accounts between husbandsand wives, some conforming rules might beneeded for worker-specific offsets. For example,if the personal account is viewed as an “asset’ indivorce proceedings, should the accompanyingoffset be viewed as a “debt?” Would the debttransfer with the asset, or remain with the origi-


nal account holder? A case might be made foreither approach.

Offsets for Disabled-Worker and YoungSurvivor Benefits Worker-specific offsets could be designed toexempt disabled-worker beneficiaries from theoffset until they reach retirement age. Similarly,when a worker dies leaving minor children (ordisabled adult children), policymakers coulddecide to exempt from the offset the benefitspayable to his or her children. A key question iswhether a worker’s decision to shift SocialSecurity taxes to a personal account shouldaffect family life insurance protection otherwiseprovided by the worker’s earnings and contribu-tion history.

The application of worker-specific offsets couldproduce countless outcomes. This chapter is astep toward exploring details of the still largelyuncharted waters of worker-specific offsets andtheir consequences for beneficiaries, taxpayers,and Social Security finances.

Individual Account Taxation

Finally, how might individual accounts betaxed? The tax model selected can have a dra-matic impact on the costs, participation levels,forms of payout, and benefits and burdens asso-ciated with creating individual accounts.

In general, one cannot understand how to taxpayments from individual accounts withoutunderstanding how contributions to them aretaxed. “Tax equivalences” summarize the dis-tinctions among different tax regimes.

Tax EquivalencesIn brief, the government can tax (T) or exempt(E) income at three points in the saving process:it can tax (1) deposits, (2) investment earnings,and/or (3) withdrawals. An income tax generallytaxes deposits and investment earnings, but notwithdrawals (summarized TTE). A consumptiontax can operate in one of two ways: It may taxdeposits and exempt investment earnings and

withdrawals (summarized TEE), or it mayexempt deposits and investment earnings, buttax withdrawals (summarized EET). Under cer-tain assumptions, these two tax regimes are eco-nomically equivalent. Finally, it is possible toexempt deposits, investment earnings, and with-drawals from a savings vehicle (summarizedEEE), but doing so subsidizes savings in thevehicle and can actually allow taxpayers toextract the subsidy without increasing their netsavings at all.

Models for Taxing Individual Accounts

Based on this general situation, four models fortaxing individual accounts under current lawcould be used. The “normal” model for taxingsavings mirrors the income tax regime (TTE).Money that is saved is taxed when initiallyearned, and the income generated by the savingsis then taxed when it is realized. The traditionalmodel for taxing retirement savings mirrors theconsumption tax regimes. Income earned onqualified retirement savings is exempt from taxso that only the contributions made by workersand their employers are subject to tax. This isaccomplished either by way of an upfront taxdeduction for contributions (EET) or a taxexemption for withdrawals (TEE). Certain otherforms of retirement savings are taxed under athird model of deferral, which taxes both contri-butions and income earned on contributions,but taxes contributions immediately while tax-ing income earned on contributions only uponwithdrawal. Finally, Social Security contribu-tions and benefits are taxed under a fourth,entirely different regime. The employee’s half ofcontributions are taxed, and anywhere fromzero to 85 percent of benefits paid are taxed,depending on the beneficiary's income level.

Each of these models can be, and in some casesis, combined with tax credits, preferential rates,and tax penalties, all of which can further affecttax burdens, subsidies, and incentives.


Considerations in Determining the TaxTreatment of AccountsPolicymakers will need to take a variety of fac-tors into account when deciding which of thesemodels to apply to individual accounts, includ-ing the accounts’ purposes and structure, andcertain implementation issues. In particular, thetax treatment of individual accounts is likely tohave important consequences for participationrates, complexity from a participant and govern-mental perspective, the form of payout, and dis-tributional issues. The challenge forpolicymakers in determining the tax treatmentof the accounts will be how to navigate betweenthese frequently conflicting concerns.

An important question for policymakers iswhether distributional concerns should beaddressed through the tax treatment of theaccounts, through the method for allocatingfunds to the accounts, or by adjusting tradition-al Social Security benefits. How the tax treat-ment of the accounts affects savings in othertax-preferred vehicles also merits attention.

With respect to complexity, the traditionalmodel for retirement savings and, in some cases,the Social Security model, are likely the mostsimple. Unlike the other models, they do notrequire workers or the government to track theamount of each worker’s contributions and theportion of investment earnings on which he orshe has paid tax.

If the accounts are voluntary, policymakers mayalso wish to consider how the tax treatment ofthe accounts affects participation rates. In gener-al, if the account system involves offsets, deci-sions about participation are likely to be

influenced by the after-tax value of funds shiftedto the account relative to the after-tax value ofthe traditional Social Security benefits foregone.If the accounts are independent from the SocialSecurity system, participation decisions are likelyto be influenced by the tax treatment of theaccounts relative to other savings vehicles.

How would the tax treatment of individualaccounts affect the taxation of traditional SocialSecurity benefits? If an individual account planis funded out of existing Social Security taxesbut is not funded equally from the employers’and employees’ shares, the creation of individualaccounts may raise the question whether adjust-ments are appropriate to the taxation of tradi-tional Social Security benefits.

Finally, tax incentives and penalties could beused to discourage withdrawals before retire-ment, or to encourage phased withdrawals orannuitization of the accounts.

Concluding Remarks

The Panel believes that the more detailed analy-ses in the following chapters make importantheadway in identifying issues in the design of anew system of individual accounts that blendproperty concepts with social insurance. Ourpurpose has been to provide dispassionateanalysis that will aid policymakers in this impor-tant aspect of public policy. Although panelmembers disagree about a policy of replacingpart of Social Security with individual accounts,all agree that the work presented in the chaptersthat follow is an important contribution toinformed public policy.


Chapter One Endnotes

1 For instance, ownership of land might notinclude mineral rights, and a vested right to apension might not include the right to receivefunds prior to retirement age.

2 Some state and local employees are exempt fromSocial Security coverage. Under historicalarrangements, states and localities could choosewhether to provide Social Security coverage toemployees who are covered under state or localpension plans.

3 The choice to distinguish individual accountplans by funding source was a difficult one forthe Panel, given the potential fungibility of differ-ent types of government revenue. Given the focuson payouts from individual accounts, however,the Panel as a whole agreed that this distinctionproved helpful.

4 Individual development accounts are matchedsavings accounts targeted to low-income workersand typically restricted to first-home purchase,small-business start-up, and post-secondary edu-cation and training.

5 Assumptions underlying the annuity estimatesare consistent with assumptions used in the 2003report of the Social Security Trustees. It isassumed that the purchase of annuities is manda-tory, the federal government would provide theannuities, inflation is assumed to be 3.0 percentper year, and the real interest rate is 3.0 percentper year, such that the nominal interest rate is 6.1percent.

6 This occurs because single life annuities payhigher monthly amounts than a joint-life annuitythat covers two lives. If the widowed partnerwould inherit the deceased partner’s account, asingle life annuity from the combined accounts ofthe deceased and the widowed spouse would bemuch higher than the survivor payments fromjoint-life annuities that both bought before thedeath occurred.

7 The account holder usually has the option tolater use the funds in the deferred annuity to buya life annuity, but relatively few people do so.

8 In general, the guaranty funds provide insurancecoverage for annuities up to a net present valueof $100,000. To the extent that annuitants havepolicies above the limit, the uninsured portionwould represent a claim on the failed insurancecompany and in all likelihood would not be paidin full.

9 Assumptions underlying this estimate are: partic-ipation in the accounts and purchase of annuitieswould be mandatory; during the accumulationphase, accounts would earn a net real return of4.6 percent; annuity reserves would earn a 3.0percent net annual return.

10 Today, total financial asset values are roughlytwice the size of GDP, according to estimates ofthe Office of the Chief Actuary of the SocialSecurity Administration. Assuming that relation-ship remained unchanged, annuity reserveswould be about 7-8 percent of total financialasset values.

11 The nine community property states are Arizona,California, Idaho, Louisiana, Nevada, NewMexico, Texas, Washington, and Wisconsin;population percentage calculated from data fromthe U.S. Census Bureau Statistical Abstract of the

United States 2003, Table 20.

12 Chapter Nine examines worker-specific offsets inplans that permit workers to shift part of theirSocial Security taxes to individual accounts.

13 Chapter Nine examines worker-specific offsets inplans that permit workers to shift part of theirSocial Security taxes to individual accounts.Chapter Eight considers payment options whenoffsets are mandatory and apply to all retirees.

2Chapter

Financial Demographics

23

Understanding how Americans save—or do notsave—is critical groundwork for any new initia-tive on workers’ retirement income. Millions ofpeople (most of them low-income and/or minor-ity) have no relationship with a mainstreamfinancial institution. Many more, including mid-dle-income families, have saved little for retire-ment and can be expected to depend almostentirely on Social Security for retirement income.

This chapter examines the components ofincome for Americans age 65 and older—includ-ing Social Security, employer-sponsored pen-sions, individual savings, and earnings fromwork—and the relative role these sources ofincome play for individuals at different incomelevels. Several indicators of economic well being,including homeownership and other assets notearmarked for retirement, offer further insightinto how Americans save, as do measures offinancial stress such as poverty, household debt,and personal bankruptcies. The potential size ofindividual accounts is also explored, based ongiven assumptions about the amount and dura-tion of contributions to the accounts and invest-ment earnings.

Components of RetirementIncome

Retirement income in the United States is oftencharacterized as a “three-legged stool” made upof Social Security, employer-sponsored pensions,and individual savings. But some Americans age65 or older who are still working, or who havea working spouse, receive employment incomeas well. Means-tested payments fromSupplemental Security Income (SSI) can helpindividuals age 65 and older who have very lim-ited assets and income.

In 2002, Social Security provided nearly 40 per-cent of the income of Americans age 65 andolder (Figure 2-1), more than from any othersingle source. The next largest source of income,earnings from work, accounted for one-quarterof aggregate income, followed by employer-pro-vided pensions (about evenly divided betweenprivate and public pensions) at 20 percent, andincome from individually owned assets at 14percent. Occupying a smaller portion of theaggregate were SSI, unemployment insurance,workers’ compensation, veteran’s benefits, cash


assistance from state or local programs, andalimony and other contributions from individu-als outside the household, which together com-prised 3 percent of aggregate income for olderAmericans.

Almost all older Americans receive SocialSecurity—about nine in ten Americans age 65and older collect benefits (Figure 2-2). Pensionscan provide an important supplement to SocialSecurity, but about half of married couples andtwo-thirds of unmarried men and women lackpension income. More people have some assetincome, but most of the elderly receive onlysmall amounts. Indeed, just under half of cou-ples and about one in three unmarried olderAmericans get as much as $1,000 per year inasset income. Fewer than one in four elderlyreceive earnings from work, although some indi-viduals earn substantial amounts.

Social Security is a major income source forretired workers through the middle of theincome distribution. Low-income Americans aremost reliant on Social Security because they areless likely to have other sources of support.Figure 2-3 illustrates what funds olderAmericans relied on in 2002, by income quintiles.

Those in the bottom two-fifths of the incomedistribution drew more than 80 percent of theirtotal income from Social Security in 2002—fourtimes Social Security’s share for the top incomegroup. Those in the middle fifth of the distribu-tion, with incomes between about $15,000 and$24,000, counted on Social Security for 64 per-

Figure 2-1. Shares of Income from SpecifiedSources, 2002 Married Couples and UnmarriedPersons Age 65 and Older

Source: U.S. Social Security Administration, Forthcoming. Incomeof the Population 55 or Older, 2002

Other 3%

Asset Income14%

Earnings25%

PrivatePensions

or Annuities10%

PublicPensions

9%

Social Security39%

Figure 2-2. Percent Receiving Specified Sources of Income, 2002Married Couples and Unmarried Persons Age 65 and Older

Type of Income Total Married Couples Unmarried Men Unmarried Women

Percent receiving

Social Security 90 91 87 89Pensions – total 41 51 39 32

Public employee pensions* 15 19 13 13Private pensions 29 37 28 21

Income from assets 55 67 47 48More than $1,000 a year 36 45 30 29

Earnings from work 22 36 18 12Supplemental Security Income 4 2 5 6

*Includes government employee pensions – federal military and civilian and state and local.Source: U.S. Social Security Administration, Forthcoming. Income of the Population 55 or Older, 2002

Chapter Two: Financial Demographics 25

Figure 2-3. Shares of Income from Specified Sources by Income Level, 2002Married Couples and Unmarried Persons Age 65 and Older

83%

3%1%

3%

10%

82%

7%

5%3%

4%

Other4%

Income from Assets

7%

Earnings7%

Pension15%

Social Security64%

3%

14%

13%

24%

46%

2%

24%

36%

19%

19%

Source: U.S. Social Security Administration, Forthcoming. Income of the Population 55 or Older, 2002

Lowestfifth $9,720

Second fifth $9,720 -$15,180

Next to highest$23,880 - $40,980

Top fifth$40,980+

Middle fifth$15,180 - $23,880


cent of total income, with pensions contributingthe second-largest share (15 percent). Those inthe “next to highest” income quintile hadincomes between $22,000 and $41,000, ofwhich Social Security comprised nearly half (46percent) and pensions about one quarter. Finally,the top income group consisted of many benefi-ciaries not yet retired or with a working spouse.Earnings were the largest single income sourcefor these individuals, at 36 percent of the total,followed by asset income, at 24 percent, andSocial Security, at about 19 percent.

Social Security benefits replace part of theincome needed to maintain the worker’s stan-dard of living when he or she retires. But theprogram’s progressive benefit formula, as seen inFigure 2-4, replaces a larger monthly share ofpast earnings for low-wage workers. Althoughhigher earners receive larger benefit checks,those checks represent a smaller fraction of pre-vious earnings. For example, a 65-year-old whoretired in 2004 with a lifetime of “medium”earnings ($34,600 in 2003) would receive$14,500 a year, a 40 percent replacement rate.But someone with a lifetime of low earnings

($15,600 in 2003) would receive 56 percent ofprior earnings from Social Security. A retireewho always earned the maximum amount thatis taxed and counted toward Social Security($87,000 in 2003) will see about a quarter ofthose past earnings in a benefit check. Thesecomparisons are for single individuals who sur-vive to retirement. When evaluated on a lifetime,family basis, the extent of redistribution throughretirement benefits is diminished due to differ-ences in life expectancy, spousal benefits, andother differences between high and low earners(U.S. GAO, 2004).

Figure 2-4 applies only to workers who claimSocial Security at age 65. Although this has longbeen called the normal retirement age, themajority of Americans claim benefits before theyturn 65. Workers who opt to start receivingSocial Security at 62 (the earliest eligibility age)trigger a permanent benefit reduction of about20 percent. Thus, most Americans receive lessthan the full benefits shown in Figure 2-4, butcan receive these benefits for more years.

Figure 2-4. Social Security Benefits Compared to Past Earnings by Earnings Level, 2004Retired Workers Age 65

Source: Board of Trustees, 2004. Annual Report of the Board of Trustees

$0

$20,000

$40,000

$60,000

$80,000

$100,000

Benefits

Past Wages

"maximum""high""medium""low"

56%40%

35%25%

Earnings Amount

$15,600

$34,600

$54,300

$87,000

$8,800 $14,500$19,100

$21,400


In January 2004, the average Social Securitybenefit for a retired worker was $922 a month,or about $11,060 a year. For a retired couple,the average combined benefit was about $1,520a month, or $18,300 a year. Nearly two in threeelderly beneficiaries (64 percent) rely on thesebenefits for more than half of their total income.One in five beneficiaries have no other source offunds.

Role of Social Security—Pastand Future

The role of Social Security in retirees’ totalincomes has been fairly stable for the past 25years. Changes already enacted will cause futurebenefits to grow somewhat slower than wages.It remains to be seen what other policy changesin benefit levels or revenues will be made tobring the system into long-range balance.

The Past 25 YearsFor more than 25 years, Social Security has beenthe main source of income for older Americans.In 1976, as in 2002, about two-thirds of benefi-

ciaries drew more than half of their total incomefrom Social Security—and about half of thosepeople relied on the program almost exclusively(Figure 2-5). Unmarried women are the mostreliant on Social Security, with three in four ofthese beneficiaries drawing at least half theirincome from Social Security. About one in threeolder unmarried women on Social Security haveno other source of income. Today, more olderbeneficiaries rely on Social Security as their solesource of income than was the case in the1970s.

The Next 25 YearsThe prognosis is mixed for upcoming genera-tions of retirees. Future workers are projected toenjoy earnings that grow somewhat more thanthe cost of living. Because Social Security bene-fits for new retirees are indexed to earningsgrowth, future retirement benefits will reflectthese real earnings gains.

At the same time, legislation enacted in 1983called for gradually raising the age at which fullretirement benefits would be paid from age 65

Figure 2-5. Role of Social Security in Total Income, 1976 and 2002Married Couples and Unmarried Beneficiaries Age 65 and Older

Year Percent for whom Social Security is:

50 percent or more 90 percent or more 100 percent of incomea

of total income of total income

Total Married Couples and Unmarried Beneficiaries

2002 66 34 221976 66 28 16

Married Couples

2002 54 21 121976 56 18 8

Unmarried Men

2002 65 35 251976 70 28 21

Unmarried Women

2002 77 44 291976 74 36 21

a The earlier data are for 1978, the earliest year available.

Sources: U.S. Social Security Administration, Forthcoming. Income of the Population 55 or Older, 2002; U.S. Social SecurityAdministration, 1979. Income of the Population 55 or Older, 1976


to 67. These changes are phasing in over thenext two decades.1 While early benefits will stillbe payable at age 62, recipients face a largerreduction for early retirement. When the fullbenefit age reaches 67, benefits claimed at age62 will be reduced by 30 percent and benefitsclaimed at age 65 will be reduced by 13.3 per-cent. Benefits for future age-65 retirees willreplace a smaller portion of their past earningsthan has been the case for retirees in the past.

This change is illustrated in Figure 2-6. A medi-um earner who retired at age 65 in 2004 saw amonthly benefit of $1,184, which replaces about42 percent of his or her prior earnings. By 2030,a similar medium earner who retired at 65(when the full benefit age is 67) would reap abenefit that replaced just 36 percent of his orher prior earnings. Because of real earningsgrowth, the dollar benefit would be higher,about $1,385 in 2004 dollars. One who workedlonger and delayed claiming benefits until 67 in2030 would have a higher dollar benefit of near-ly $1,600 and a replacement rate similar to thatof an age-65 retiree today, about 41 percent ofprior earnings.

Other changes also will modulate future SocialSecurity benefits. First, rising premiums for PartB of Medicare, which are deducted from mostretirees’ Social Security checks, will take a biggerbite because those premiums are projected torise faster than benefits. Second, because theincome threshold for taxing Social Security ben-efits is not indexed to rise as income rises, more

future beneficiaries will have part of their SocialSecurity benefits subject to federal income taxes.Using projections of the Social Security andMedicare Trustees, Figure 2-7 illustrates howthese developments together with the increase inthe full benefit age will affect the level at whichbenefits replace earnings for a medium earnerretiring at age 65. By 2030, benefits afterMedicare premiums and new income taxeswould represent about 30 percent of prior earn-ings, compared with about 39 percent today.

Social Security faces a long-run imbalancebetween revenue coming in and payments goingout. This imbalance could be remedied by low-ering benefits, which would further reduce thereplacement rate (i.e. the ratio of benefits to for-mer earnings), raising revenues, or a combina-tion of both. At this point, the Social Securitysystem is on track to produce benefits that arehigher in real terms, but that replace a smallershare of workers’ prior earnings, than has beenthe case for retirees today or at any time duringthe past three decades.

Tax Favored RetirementSavings

Tax incentives are the federal government’s prin-cipal policy tool for encouraging both workersand employers to set aside funds for retirement.Contributions to pension plans are a tax-deductible business expense for employers;employees, for their part, do not have to pay

Figure 2-6. Social Security Benefit and Replacement Rate, 2004 and 2030Scaled Medium Earner at Age 65 and at Normal Retirement Age

Year Retire at Age 65 Retire at Normal Retirement Age Attains Monthly Replacement Normal Monthly ReplacementAge 65 Benefit* rate (percent) Retirement Age Benefit* rate (percent)

2004 $1,184 41.9 65:4 $1,209 42.52030 $1,385 36.3 67 $1,599 41.1

*Constant 2004 dollars



taxes on pension accruals until they actuallyreceive pension income. These policies areexpensive: lost federal income tax revenue asso-ciated with pension and 401(k) plans are esti-mated to reach $123 billion in fiscal year 2004.Incentives for retirement savings are one of themost significant losses in tax revenue for the fed-eral government, rivaling tax expenditures forhealth care insurance and home ownership (U.S.Office of Management and Budget, 2004).2

Yet, tax breaks for retirement savings are notsignificant for large segments of the population,and most of these tax benefits accrue to high-income tax filers.3 Only about half of the work-ing population is covered by a retirement plan atany given time. College-educated, higher-incomeworkers are more likely to have an employer-sponsored retirement plan than those with lesseducation and lower earnings. As nearly all pen-sion plans tie contributions (and benefits) toearnings level, lower-wage workers who do havea pension participate on a smaller scale thantheir higher earning counterparts.

Pension Coverage Is Stable; The Form isChangingRates of pension coverage have changed littlesince the mid-1970s, at just under half of pri-vate-sector employees and a little over half of allworkers (Munnell et al., 2003; Copeland, 2001).The kind of pensions that Americans hold, how-ever, has undergone a major transformation.

Defined-benefit plans were the most commontype in the 1970s and 1980s; today, defined-contribution plans dominate.

This shift has implications for the payout rulesfor new individual account proposals. In adefined-benefit plan, participants are promised aspecified benefit level at retirement. Defined ben-efits have historically been paid as monthlyamounts and last for the life of the retiree—andusually for the life of a widowed spouse. Inrecent years, however, defined-benefit plans havebeen adopting lump-sum cash out features forworkers who leave their jobs before retirement.

The defined-benefit amount is usually calculatedbased on the worker’s length of service and cov-ered wages. For example, a plan might pay 1.5percent of the employee’s final salary for eachyear of participation in the plan, giving anemployee who retired after 20 years a pensionequal to 30 percent of his or her final pay. Theemployer is responsible for ensuring that its con-tributions into the plan, plus investment earn-ings, will suffice to pay promised benefits.Participation in a defined-benefit pension is usually automatic and does not depend onemployees’ out-of-pocket contributions.Employees have few (if any) choices to makebefore retirement.

With defined-contribution plans, however, retire-ment benefits vary with annual contributions

Figure 2-7. Social Security Net Replacement Rate, 2003 and 2030Medium Earner Age 65

Year and Reason Replacement Rate for Change (benefit as a percent of prior earnings)

Age 65 retiree in 2000 41.2Net, after deducting Medicare Part B premium 38.7

Age 65 retiree in 2030:After raising normal retirement age 36.5Net, after deducting Medicare Part B premium 33.2Net, after new personal income tax 30.5

Percentage change in net replacement rate (-21)

Source: Munnell, 2003. “The Declining Role of Social Security”


made to each participant’s account, plus invest-ment returns. In most cases, employees candecide each year whether and how much theywill contribute. Many defined-contributionplans give employees primary responsibility forchoosing how their accounts will be invested.Employees generally bear the risk that these pen-sions may fall short of producing an adequatesupplement to Social Security and any otherretirement income. Defined-contribution planssometimes offer payments as monthly incomes(annuities); more commonly, they pay lumpsums when employees leave the plan. Workerswho take lump sums before retirement can rollover the funds into another tax-deferred retire-ment plan to avoid paying taxes on the moneyuntil retirement. Spousal rights are more limitedin individual retirement accounts than indefined-benefit plans (see Chapter Six).

The most popular type of defined-contributionplan today is a 401(k) plan. This plan givesemployees the opportunity to contribute part oftheir salary to the plan on a tax-deferred basis.Employers often match the contributions theiremployees make to the plan, up to a set percent.Participation in a 401(k) plan usually requires a

conscious decision by the worker to enroll andmake contributions, although automatic enroll-ment is becoming more widespread. Lump sumsare the typical form of payout. Workers usuallyhave a role in determining investment choices.

These 401(k) plans have both advantages anddisadvantages relative to traditional defined-ben-efit pension plans. They shift more of theresponsibility and financial market risk to work-ers. But, 401(k)s give workers more choice inhow their portfolio is allocated and more waysto access funds; plus, 401(k) balances are fullyportable between jobs.

Between 1979 and 1998, the percentage of pri-vate sector workers participating in defined-ben-efit plans fell from 37 to 21 percent, whereasworkers with only a defined-contribution plangrew from 7 percent to 27 percent (Figure 2-8).Providing workers with a supplemental defined-contribution plan (usually a 401(k) plan in addi-tion to a defined-benefit plan) also became morecommon: the share of private sector employeeswith this option grew from 9 percent to 15 percent.

3734

29

24 21

44 45 45 4548

0

10

20

30

40

50

60

1979 1984 1989 1994 1998

Year

Defined Contribution OnlyBothDefined Benefit Only

Perc

enta

ge

28

19

15

107

Figure 2-8. Pension Coverage Trends, 1979-1998Private Wage and Salary Workers

Source: U.S. Department of Labor, Pension and Welfare Benefits Administration, 2001-2002. Abstract of 1998 Form 5550 Annual Reports

24367 NASI TEXT 1/17/05 11:36 AM Page 30


Occupation and Pension Coverage A gap in pension coverage between white-collarand blue-collar workers grows with this trendtoward defined-contribution plans. While 65percent of white collar, professional, and techni-cal employees participate in some sort of retire-ment plan, only 39 percent of blue collar andservice employees do. In past decades, defined-benefit plans were associated with large, union-ized companies where blue-collar jobspredominate – such as in automobile manufac-turing, steel production and mining industries.U.S. employment in these sectors is a decliningshare of the workforce. Blue-collar jobs todayinfrequently offer pension plan coverage andwhite-collar workers are more likely than blue-collar workers to have defined-benefit pensions.About 26 percent of white-collar workers partic-ipate in traditional defined-benefit plans, com-pared with 17 percent of blue-collar workers.

Figure 2-9 shows that both white-collar andblue-collar workers are now more likely to havedefined-contribution pension plans than tradi-tional defined-benefit plans. Indeed, while nearlyhalf (48 percent) of private employees had sometype of retirement plan in 2000, fewer than one

in five (19 percent) participated in a traditionaldefined-benefit pension plan.

Tax-Favored Retirement SavingsAccounts Today, about half of American families havesome sort of tax-favored retirement savings planor account other than a defined-benefit pensionplan (Figure 2-10). These plans include 401(k)s,other defined-contribution plans sponsored bytheir employers, and individual retirementaccounts (IRAs). Ownership of IRAs increasedduring the 1990s, as job-switching workersopened these retirement savings accounts torollover 401(k)s and other pension funds.Between 1992 and 2001, families with an IRAincreased from one in four to nearly one in three(Copeland, 2003a).

Figure 2-10 divides American families into fiveincome groups to illustrate the relationshipbetween income and the median value of tax-favored retirement savings. The median valueamounted to less than one year’s income forthose families, at all income levels, with savings.The median balance was about $29,000 in2001. Families in the lowest two-fifths of the

9

13121718

26

19

39

49

65

48

0

10

20

30

40

50

60

70

All Employees Professional,technical and

related employees

Clerical and salesemployees

Blue-collar andservice employees

Defined Contribution OnlyBothDefined Benefit Only

Perc

enta

ge

13

Figure 2-9. Pension Coverage by Occupation, 2000Private Employees

Source: U.S. Department of Labor, Bureau of Labor Statistics (BLS), 2003. National Compensation Survey: Employee Benefits in PrivateIndustry in the United States, 2000

24367 NASI TEXT 1/17/05 11:36 AM Page 31


income distribution were much less likely tohave such savings, and their balances weresmaller. Indeed, the median value of retirementsavings in these two income brackets was zero,because most of these households did not have atax-favored retirement account. Among house-holds in the middle quintile, by contrast, abouthalf (53 percent) had savings with a medianvalue of $13,500. But when households withoutaccounts are included, the median value falls to$800.

Age differences also matter. Younger workershave more of their working lives ahead of themand can be expected to save additional fundsbefore retirement, while older families havemost of their retirement savings years behindthem. Although younger families are about aslikely as older ones to have retirement savings,the median value of the accounts rises with age.About 6 in 10 families headed by someone aged55-64 owned tax-favored retirement accounts;the median value of those accounts was $55,000(Figure 2-11). By contrast, median account values totaled only $28,000 for families aged35-44.

Gaps among racial groups exist as well. AfricanAmerican and Hispanic American families areless likely than white families to have some typeof tax-favored, retirement savings and theiraccounts tend to be smaller. The median account

value in 2001 was $8,800 for the 39 percent ofAfrican American families that held accounts,and $8,700 among the 31 percent of Hispanicfamilies holding accounts. By contrast, 57 per-cent of white families had tax-favored retire-ment savings, with a median value of $35,000.

While about half (52 percent) of all Americanfamilies have some form of individual retirementsavings, others have defined-benefit pensionrights from their current or former employment.Taking all these resources together, 59 percent offamilies have either defined-benefit pensions orsome type of retirement savings, or both, for atleast one member of the family.

Implications for the Design of IndividualAccount PayoutsWorkers today have more responsibility for con-tributing to and managing their accounts, pre-serving funds until retirement, and deciding howand when to convert their accounts into retire-ment income. The shift from defined-benefitpensions to defined-contribution plans has madeemployees more responsible for generating theirown retirement income. This shift has implica-tions for the design of payout rules for individ-ual accounts. Workers’ experience with defined-contribution plans, such as 401(k) plans, mayshape their expectations about any new savingsprogram. Yet, the replacement of pensions that

Figure 2-10. Ownership of Tax-Favored Retirement Accounts by Family Income, 2001

Accounts include 401(k) plans, other defined-contribution plans, IRAs and Keogh plans

Number of Percent Percent ofIncome families Median owning total accountGroup (millions) income accounts All families Account holders assets

Total 106.5 $39,900 52 $600 $29,000 100Lowest Fifth 21.3 $10,300 13 $0 $4,500 1Second fifth 21.3 $24,400 33 $0 $8,000 4Middle fifth 21.3 $39,900 53 $800 $13,500 9Next to highest fifth 21.3 $64,800 74 $16,000 $31,000 19Next to highest tenth 10.6 $98,700 85 $36,000 $52,000 17Top ten percent 10.7 $169,600 88 $102,000 $130,000 50

Source: Orszag, 2003. Tabulations using the 2001 Survey of Consumer Finances

Median account value in 2001 dollars


Figure 2-11. Ownership of Tax-Favored Retirement Accounts, by Age, Race and Ethnicity, 2001

Accounts include 401(k) plans, other defined-contribution plans, IRAs, Keogh plans

Number of Percent owning Median account valueFamily Characteristics families accounts (for owners)

(in millions) 2001 dollars

All families 106.5 52 $29,000

Age of Family Head

Less then 35 24.2 45 9,30035-44 23.8 61 28,20045-54 22.0 64 48,00055-64 14.1 59 55,00065-74 11.4 44 60,00075 and older 11.0 26 46,000

Race and Ethnicity

White, non-Hispanic 81.2 57 35,000Black 13.9 39 8,800Hispanic 8.5 31 8,700Other 3.0 48 27,000

Source: Copeland, 2003b. Tabulations using the 2001 Survey of Consumer Finances

Figure 2-12. Asset Ownership by Age, Race and Ethnicity, 2001

Personal MutualFamily Automobile Health Life Retirement Funds, Stocks, BusinessesCharacteristics Insurance Insurancea Accountsb Bonds

All Families 84 77 69 52 41 12White Non-Hispanic Families

All ages 89 81 73 57 47 14Under 35 85 81 64 52 40 935-64 93 85 77 52 51 1865 or older 81 73 69 66 43 8Non-White and Hispanic Familiesc

All ages 71 64 58 37 21 5Under 35 64 58 46 31 17 235-64 78 70 62 47 24 765 and older 52 47 70 10 15 4

a Life insurance includes both whole life and term life insurance.

b Personal retirement accounts include 401(k) plans and other employment-based defined-contribution plans, IRAs and Keogh plans.Defined-benefit pension plans are not included.

c Non-white and Hispanic families are combined because the survey does not provide statistically reliable estimates for each group separately by age.



offer defined benefits for the life of a retiredworker with plans that shift more responsibilityand risk to workers may increase the importanceof providing financial-risk protection in any newsavings program (see Chapter Three).

Other Measures of FinancialSecurity and Stress

While the focus of this report is federal retire-ment policy, economic security in old age reflectsfinancial circumstances and choices madethroughout the working years. This sectionbriefly reviews elements of financial well being,including homeownership and financial assets,and indicators of financial stress such as poverty,personal debt, and bankruptcy filings.

Assets of a Typical American A typical American family owns a car, theirhome, health and life insurance, and some kindof retirement savings account (Figure 2-12).About 84 percent of families own a car, and 68percent own, or are buying, their homes. Healthand life insurance are widespread, although byno means universal; about 77 percent of familieshave some type of health insurance and 69 per-cent have some kind of life insurance. Cashvalue life insurance policies are also used as sav-ings vehicles. About half of all families havesome kind of personal retirement account andabout 41 percent own mutual funds, stocks, orbonds outside of a retirement account.

Significant differences in asset ownership amongraces and ethnicities exist today. Families wherethe head of the household described him or her-self as Hispanic, African American, or a memberof another minority racial group, are much lesslikely to have health insurance or life insuranceother than Social Security, or to own other keyassets. In fact, more than one-third of minorityfamilies are without any kind of health insur-ance, and three in ten such families do not owna car. While Hispanic and African Americanfamilies tend to be somewhat younger thanother families, and their relative youth may

account for smaller asset accumulations, racialand ethnic disparities persist within age groups.

Homeownership A home is often the largest single assetAmerican families own. Individuals whoapproach retirement in homes they own, withmortgages nearly paid off, have an importantsource of financial security that their counter-parts in rental housing lack. About two in threefamilies own, or are buying, their primary resi-dence. But homeownership is less commonamong non-white families and younger families(Figure 2-13). Moreover, non-whites who doown homes typically have less equity in theirproperties, and their homes have a lower value.To illustrate, about 74 percent of non-Hispanicwhite families owned their homes in 2001; theirmedian home equity was $76,000. But only 42percent of Hispanic and non-white families werehomeowners in 2001; their median home equitywas $42,000.

Older families are more likely to own or be buy-ing their homes and, because they have hadmore time to pay off mortgages, they have builtup more home equity. Within each age group,minority families are less likely to own theirhomes and, when they do, they have less homeequity.

Economic Insecurity Economic insecurity is a fact of life for manyAmerican households. The poverty line serves asthe basic measure of economic hardship.4

By this measure, nearly one in five AfricanAmericans and Hispanic American families arepoor (19 percent), compared with about 7 per-cent of white families in 2000 (U.S. CensusBureau, 2002).

Debt is another indicator of financial insecurity,and Americans owe more today, relative to theirincomes, than at any time since World War II(Mishel et al., 2003). Between 1979 and 2001,aggregate household debt, as a percentage ofdisposable income, grew from 73 percent to 109percent. Mortgage debt accounted for much of


this increase, growing from 46 percent to 62percent of income over the period. The introduc-tion of home equity loans in the late 1980s alsocontributed; these loans rose to 10 percent ofdisposable income by 2001. Consumer credit, asa share of disposable income, rose from 20 per-cent to 23 percent of income between 1979 and2001. Debt as a share of assets grew more slow-ly over this period, from 14 percent in 1979 to17 percent in 2001. Mortgages as a share of realestate assets were fairly stable during the 1980sbut, between 1989 and 2001, the ratio of mort-

gages to real estate assets rose from 31 percentto 41 percent.

Not all debt is created equal. Credit card debtcan be particularly problematic because it typi-cally costs more than other kinds of loans.Three-quarters (76 percent) of American familieshave a credit card (Figure 2-14). Despite highinterest rates and large penalty fees for late pay-ments, just over four in ten families carry a cred-it card debt balance from month to month. Offamilies with incomes between $10,000 and

Figure 2-14. Credit Card Debt by Family Income, 2001

Percent of all Percent of card Percent of all Family families with at holders carrying families carrying Average creditincome least one credit card a balance a balance card debt

All families 76 55 42 $4,126

under $10,000 35 67 23 1,837$10,000-24,999 59 59 35 2,245$25,000-$49,999 80 62 50 3,565$50,000-$99,999 90 56 50 5,031$100,000 or more 98 37 36 7,136

Source: Draut and Silva, 2003. Calculations based on the 2001 Survey of Consumer Finances.

Figure 2-13. Homeownership, Market Value and Home Equity by Age, Race and Ethnicity, 2001

Family Percent Owning Median Amountattributes or Buying Home Market Value Home Equity

All Families 68 $121,000 $70,000White Non-Hispanic

All white non-Hispanic 74 $129,000 $76,000Under 35 45 $95,000 $26,00035-64 81 $135,000 $75,00065 or older 83 $125,000 $113,000

Non-white and Hispanica

All non-white and Hispanic 47 $90,000 $42,000Under 35 29 $87,000 $22,00035-64 54 $93,000 $44,00065 and older 59 $92,000 $59,000

a Non-white and Hispanic families are placed in a single category because the survey does not permit statistically reliable information forthe groups separately by age.



$25,000, about one in three carries credit carddebt from month to month, as do about half offamilies with incomes between $25,000 and$100,000. The average credit card debt was justover $4,100 in 2001, according to tabulations ofthe Survey of Consumer Finances. These surveydata may reflect considerable under-reporting.Economists at the Federal Reserve estimate theaverage credit card debt per household at about$12,000 in 2001 (Draut and Silva, 2003).

An individual’s level of indebtedness may wellaffect his or her willingness to participate in afederal savings program. For example, if an indi-vidual account program were voluntary, workerswith high mortgage payments, credit card debt,or home equity loans might choose not to par-ticipate. Indeed, paying off high-interest debtrather than saving for retirement may be theright choice. Debtors may also want to gainaccess to account balances before retirement,particularly if the accounts offer 401(k)-typesavings, against which account holders can takeout personal loans. About a fifth of 401(k) hold-ers have taken out loans against their savings.Indebtedness is a central factor in the financialdemographics of Americans.

Asset PovertyAnother measure of economic vulnerability isthe capacity of families to withstand setbackssuch as joblessness or prolonged illness.Haveman and Wolff (2001) define householdsas being asset poor if their access to wealth-typeresources is insufficient to meet their basic needsfor some limited period of time. They use ameasure of poverty to define basic needs5 andconsider the amount of assets that would beneeded to meet those basic needs if they had noincome for three months. If total net worth isconsidered available to meet basic needs (includ-ing home equity and the value of fungible retire-ment assets as well as liquid financial assets),then about one in four families is asset poor. Ifonly liquid assets are considered (not countinghome equity and retirement assets), then four inten families is asset poor (Haveman and Wolff,2001).

Personal BankruptciesPersonal bankruptcies increased substantially inthe 1990s. In 2001, 7 out of every 1,000 adultsdeclared personal bankruptcy—three times therate in 1980. If such a high rate were sustainedfor seven years, about 5 percent of the adultpopulation would have declared bankruptcyduring that period.6 Changes in bankruptcy law,access to credit, and economic behavior overtime mean that it is not possible to accuratelyestimate the proportion of the population whohave ever filed for bankruptcy. It is also impor-tant to note that many individuals do not paydebts without formally declaring bankruptcy.One study based on data from a large creditcard issuer reported that over 60 percent of itsbad debt was “charged off” for reasons otherthan bankruptcy (Dawsey and Ausubel, 2002).In other words, more people simply didn’t paytheir credit card bill than filed for formal bankruptcy.

Illness, divorce, and job loss are common rea-sons for bankruptcy. One study found that near-ly half of the one million Americans who filedfor bankruptcy protection in 1999 did so, atleast in part, because they could not cope withmedical bills or the income loss associated withan illness or injury. The fact that nearly one infour American families in 2001 did not havehealth insurance (Figure 2-12) suggests thatmany families continue to be at risk of financialdistress if serious illness strikes. Lack of medicalinsurance was a key factor in only a minority ofbankruptcy filings. More typically, the problemwas due to “underinsurance” (i.e., insurancethat does not cover a significant amount ofhealth care costs), lack of income to replace lostwages, or both (Jocoby et al., 2000). Aboutthree in ten private-sector workers lack any typeof paid sick leave or short-term disability bene-fits that would continue their income duringtemporary periods of illness or recuperation ofinjuries (Williams et al., 2003).

Generally, creditors do not have access to adebtor’s retirement savings. To the extent thatthe demarcation between retirement savings and


other forms of savings blurs, however, the argu-ment for creditors’ rights to accounts becomesmore credible.

How Big Might IndividualAccounts Be?

If contributions to individual accounts arepegged to Social Security taxable earnings, thenmany accounts would be small. The data includ-ed in Figure 2-15 cover all wage and salaryworkers, including those who work part time oronly part of the year (excluding self-employedworkers). Workers’ median earnings – wherehalf of the workers earn more and half earn less– are $21,600. Average earnings are consider-ably higher, at just under $32,000. The averageis higher than the median because the relativelysmall group of top earners brings up the aver-age. In 2001, 55 percent of men and 73 percentof women earned less than $30,000. At the highend, about 11 percent of workers earned$60,000 or more, including 6 percent whoearned the maximum amount that was taxedand counted toward Social Security benefits($80,400) in 2001.

To illustrate the potential size of individualaccounts, the Office of the Chief Actuary of theSocial Security Administration depicts earningspatterns for workers at four different lifetimeearnings levels: a “scaled medium-earner” mak-ing a career-average wage of about $34,700 (in2003 dollars); a “scaled low-earner” making acareer-average of about $15,600; a “scaled high-earner” making a career-average of about$55,500; and a “steady maximum-earner” who,each year, makes at least the maximum amountthat is taxed and counted for Social Securitybenefits, or $87,000 in 2003. The illustrativeworkers are shown in Figure 2-16, along withhow much of their average career earningswould be replaced by current law Social Securityretirement benefits.

Balances expected to accumulate in these illus-trative workers’ individual accounts are estimat-ed by taking into account various assumptionsabout: (a) the portion of wages put into theaccounts each year; (b) the number of years theworker contributes before reaching age 65; and(c) the investment returns on the account funds.Figure 2-17 shows results for new individualaccount systems with contributions of 2 percent,4 percent, or 6 percent of earnings for a personreaching age 65 after 10, 20, 30, or 40 years ofcontributing to the new individual account sys-tem. The figure also shows the annuity incomethe account would produce if the 65-year-oldbought a single-life annuity indexed for infla-tion. Investment returns on the accounts areassumed to be 4.7 percent.7

As would be the case with any new savings pro-gram based on career earnings, younger workerswould have more working years to contribute tothe program and be better able to take advan-tage of higher compounding over their careersthan older workers. The years of contributionsin Figure 2-17 are not intended to representworkers with only 10 or 20 years of work intheir entire careers, but rather to represent possi-ble account accumulations and potential annu-ities during the phase-in of an individualaccount plan.

Figure 2-15. Annual Earnings in SocialSecurity Covered Employment,2001: Wage and Salary Workers

Total number of workers (thousands) 144,803Total percent 100Less than $5,000 18$5,000 - $9,999 11$10,000 - $19,999 ($14,800 = “low”) 19$20,000 - $29,999 16$30,000 – $39,999 ($32,900 = “medium”) 12$40,000 - $59,999 ($52,700 = “high”) 13$60,000 - $80,399 (maximum taxable) 5$80,400 or more 6

Median $21,516Average $32,062Percent earning less than the average 68

Source: U.S. Social Security Administration, 2004a. AnnualStatistical Supplement to the Social Security Bulletin, 2003


In a system where workers shifted 2 percent oftheir earnings into individual accounts, a scaledmedium-earner who contributed to his or herindividual account for 20 years would have anaccount balance equal to about 60 percent of hisor her career-average annual earnings, assumingthe investment returns and administrative costindicated in Figure 2-17. If the worker con-tributed to an individual account for 40 years,the account balance would equal about 171 per-cent of career-average earnings. The youngerworker’s account balance would benefit frommore years of contributions plus more years ofcompounded investment earnings. As the per-cent of Social Security taxable earnings con-tributed to individual accounts increases, thebalance in the accounts and the replacementrates of the resultant annuities increase as well.

The worker who contributed 2 percent to anindividual account for 20 years would realize asingle-life annuity that would replace about 4.2percent of his or her annual average earningslevel, each year, assuming the investment return,administrative costs, and inflation rates assumedin Figure 2-17. The worker’s individual accountbased on 40 years of contributions would pro-duce a single-life annuity that would replaceabout 11.6 percent of his or her annual averageearnings. These relationships can illustrate thepotential size of accounts for scaled workers at

different earnings levels.8 Expressed in terms of2003 wage levels:

A scaled medium-earner, averaging about$34,700 over his or her lifetime and retiringat 65 after 20 years in the 2 percent accountsystem, would have a balance of about$21,000 and single-life annuity income atage 65 of about $1,500 a year, or about$125 a month. A scaled medium-earner whoparticipated in the account system for 40years would accumulate approximately$59,300, yielding a single-life annuity ofabout $4,000 a year, or $333 a month.

A scaled low-earner, making about $15,600per year, would have a balance of about$9,400 at age 65 after 20 years. At age 65,the account would produce a single-lifeannuity of about $660 a year, or $55 amonth. A scaled low-earner who contributedto the account program for 40 years wouldaccumulate approximately $26,700, yieldinga single-life annuity of about $1,800 peryear, or $150 per month.

A scaled high-earner, making about $55,500per year (career average), would have a bal-ance of about $33,300 after 20 years. His orher individual account would produce a single-life annuity at age 65 of about $2,330

Figure 2-16. Wage Levels and Age-65 Replacement Rates, 2003 and 2030Scaled Illustrative Earners

Social Security benefit at age 65 Illustrative earner Career-Average as a percent of career-average earnings

Earnings, 2003 2003 2030

“Low” $15,600 55.6 48.9“Medium” $34,700 41.3 36.2“High” $55,500 34.8 30.0“Maximum” $87,000 29.6 24.0

Career-average medium earnings are equal to the average wage of covered workers; low earnings are 45 percent of that amount, whilehigher earnings are 160 percent of the average wage. For scaled illustrative earners, wages vary by age to resemble typical lifetime wagepatterns.



a year, or $195 a month. A scaled high-earner who contributed to the account pro-gram for 40 years would have an individualaccount worth approximately $94,900, pro-ducing a single-life annuity worth $6,440per year, or $536 per month.

Balances and accounts would grow larger thelonger that workers were covered and contribut-ing into the system. After 30 years of participa-tion in a 2 percent account, workers areestimated to have a balance that is larger thantheir career-average annual earnings. Whenannuitized over their remaining lives after age65, the payments would be between 7 percentand 8 percent of their annual earnings. Accountswould grow even more after 40 years of partici-pation, and the life annuities would be commen-surately higher.

Are Americans Linked toPotential Administrators ofAccounts?

Entities that might administer individualaccounts or facilitate the exchange of informa-tion between account holders and accountadministrators include banks or other financialinstitutions, the Internal Revenue Service, the

Social Security Administration, and employerswho have experience administering retirementsavings plans for their employees and whoreport workers’ wages to the Social SecurityAdministration. This section considers the extentto which American adults do or do not have aconnection with these institutions. How manyAmericans do not have a relationship with afinancial institution or with the InternalRevenue Service? How many Americans are self-employed and could face new administrativeburdens?

“Unbanked” AmericansThe number of people who do not have a check-ing or savings account with a bank or creditunion is an important indicator of the popula-tion’s financial literacy. Individuals who do nothave accounts will be less familiar with proce-dures for transactions and they will be more dif-ficult to reach if banks are used asintermediaries.

Studies based on the Survey of ConsumerFinance find that between 9 percent and 10 per-cent of Americans families do not have a bankaccount of any kind (Barr, 2004; Stegman,2003). Studies based on data from the Survey ofIncome and Program Participation find that

Figure 2-17. Size of Account Balance and of Life Annuity at Age 65 as a Percent of Annual Earningsby Contribution Rate and Duration of Contributions: Scaled Medium Earner

Year age 65 Total account balance as a percent Single-life annuity as a percent of (duration of of career-average annual earnings career-average annual earnings contributions) by contribution rate by contribution rate

2% 4% 6% 2% 4% 6%

2044 (40) 171 342 513 11.6 23.2 34.82034 (30) 119 224 336 7.8 15.6 23.42024 (20) 60 117 176 4.2 8.4 12.62014 (10) 22 44 65 1.6 3.2 4.8

Assumptions: Illustrative workers with investment returns during accumulation of 4.7 percent in excess of inflation. Inflation is assumedto be 3 percent per year. Single life annuity is indexed for inflation and based on investment return of 3 percent in excess of inflation.

Source: U.S. Social Security Administration, 2003a. Unpublished calculations from the Office of the Chief Actuary


about one-fifth of all households do not have abank account (U.S. GAO, 2002; Carney andGale, 1999). Further research is needed to deter-mine why these data sets yield such differentestimates. For the purposes of this report, it suf-fices to say that at least 9-10 percent of familiesdo not have bank account of any kind. Groupsmost likely to have neither a checking or savingsaccount with a bank or credit union areHispanic Americans and African Americans,those under age 35 or over age 65, and familieswith incomes below $20,000 (Figure 2-18).

There are many reasons why individuals do nothave bank accounts. Many people cannot or donot want to pay the cost of owning an account.In addition to monthly fees for accounts withsmall balances, overdraft fees for bouncedchecks are often cited as a reason to avoid banks(Dunham and Bates, 2003; Caskey, 2001). It isalso possible that banks do not advertise lowcost accounts aggressively because they believethat other accounts are more profitable (U.S.GAO, 2002). Finally, individuals without bankaccounts may simply find check-cashing outletsmore convenient.

Unbanked families could be difficult to accom-modate in the design of individual retirementsavings accounts. A recent government effort toreach out to this population tells a cautionarytale. In 1996, Congress passed legislation requir-ing that all federal payments (other than taxrefunds) would be paid electronically. The lawalso required that individuals receiving federalpayments have access to a financial institutionaccount “at a reasonable cost.” The TreasuryDepartment created the federal alternativeaccount program—Electronic TransferAccounts—to fulfill this mandate. The programhas had only modest success. Since 1999,36,000 Electronic Transfer Accounts have beenopened, representing less than 1 percent of“unbanked” beneficiaries (U.S. GAO, 2002).This experience illustrates the challenge of usingthe existing financial infrastructure for manag-ing and making distributions from privateaccounts. Yet, an individual account system

might encourage low-wage workers to openbank accounts, if that is one of the goals of thepolicymakers.

Participation in the Income Tax SystemThe administration of an individual account sys-tem would need some way to receive informa-tion about workers, their contributions, andinvestment choices, and perhaps their maritaland family status. Policymakers might seek tobuild on the wage-reporting system sent to theSocial Security Administration. Alternatively, orin addition, policymakers might want to rely onreports directly from households to the InternalRevenue Service for tax-filing purposes.

Using regular filings with the IRS as a means ofcollecting and updating account-holder informa-

Figure 2-18. Percent of Families without a Checking or Savings Account by Race and Ethnicity, Age, and Income, 2001

Family Characteristics Percent All families 9

Race and Ethnicity

White, non-Hispanic 5Black 19Hispanic 30

Age of Family Head

Under 25 2225-34 1235-44 945-54 855-64 765 and over 15

Family Income

Less than $10,000 40$10,000-$19,999 29$20,000-$29,999 10$30,000-4,999 5$50,000 or more 1

Source: Copeland, 2003b. Tabulations using the 2001 Surveyof Consumer Finances; analysis of data by Leslie Parrish, NewAmerica Foundation


tion has some limitations. First, not everyoneregularly files tax returns. Indeed, an estimated18 million households (tax units), or about 12percent of all American tax units, do not file anannual tax return (Figure 2-19). Most of theseindividuals are not evading the tax collector;rather, they do not owe taxes and they may notbe eligible for any tax credits. For example, 97percent of non-filers have incomes below$10,000 a year, and the remaining 3 percent ofnon-filers have incomes below $20,000 a year.Fully 90 percent have no dependent children, sothey would not be eligible for the earned incometax credit available to low-income working fam-ilies with children. Most are single and abouthalf are age 65 and older.

If the IRS tax filing system is used to conveyaccount holder information to the administratorof an individual account system, how can com-munication occur between account administra-tors and households that do not file taxes? Thisquestion might arise if account holders need tochange ownership of an account, for example,following a marriage or divorce. Elderly or dis-abled individuals who may be withdrawingmoney from their accounts would also need ameans of correspondence.

Another issue with the use of the IRS as a wayto communicate with account holders is the neg-ative perception that many people have of taxcollectors. Encouraging voluntary participation

Figure 2-19. All Tax Units and Non-Filers of Personal Income Tax, by Income, Family Status,and Age, 2002

Income, family status and age Percent distribution Non-filers

All tax units Non-filers as percent of all

Total number (in thousands) 151,256 18,131 12Total percent 100 100 12

Adjusted Gross Income (2002 dollars)

Negative 1 0 0$0 - $9,999 29 97 40$10,000 - $19,999 16 3 2$20,000 or more 54 0 0

Number of Dependent Children

None 69 90 16One 14 5 4Two 12 4 4Three or more 5 1 3

Family Status

Single filers With dependents 14 9 7No dependents 47 63 16

Married filersWith dependents 18 3 2No dependents 21 26 15

Elderly Status

Non-elderly 92 49 7Age 65 and older 8 51 36

Source: Urban-Brookings Tax Policy Center Microsimulation Model (version 0503-1)


in a savings program administered by the IRScould be challenging.

Employers as IntermediariesEmployers play a central role in administeringSocial Security and pensions. In 1999, 14.6 mil-lion Americans, comprising between 9 percentand 10 percent of all earners covered by SocialSecurity, were self-employed and responsible foradministering their own tax payments (U.S. SSA,2002a; Table 4.B3). Any administrative tasksdelegated to employers should take into accountthe large number of individuals who are theirown employers. In 2001, over 6 million smallbusinesses employed 20 or fewer employees(U.S. Census Bureau, 2003; Table 744); thesesmall employers may object to any additionaladministrative burden.

Individual Development Accounts Recent scholarship and experimentation havebegun to address how to increase financialaccess and savings by low-income persons.Experts acknowledge that the problem extendsbeyond simply individual behavior, preferences,and information gaps. Solutions would requirenew financial products and public policies thatcan change the mix of financial instruments andopportunities available to Americans (Dunhamand Bates, 2003). In this vein, experiments suchas Individual Development Account (IDA) pro-grams, and linked Earned Income Tax Creditprograms (which encourage people to openaccounts to receive federal benefits), may pro-vide useful models.

Recent experimental research on IDAs – whichare matched savings accounts targeted to low-income workers and typically restricted to first-home purchase, small-business start-up, andpost-secondary education and training – haveshown that low-income and disadvantaged persons can save and build assets. MichaelSherraden and other researchers believe that wemust also consider a novel “institutional” viewof savings to fully explain savings patterns notjust by the poor, but also by persons at all levelsof income (Sherraden and Barr, 2004).

According to Sherraden, institutional determi-nants of savings include: (1) access (is a 401(k)or IDA program available?); (2) information(do employees or residents know about it?); (3) incentives (is there a meaningful tax deduc-tion or matching deposit available to encouragesaving?); (4) facilitation (is there an automaticpayroll deduction to make saving easy?); plusother mechanisms that, when present in policyand program design, will foster savings.

This research and experience could informefforts to expand individual savings accounts.That is, to help ensure maximum participationin and better utilization of individual accounts,policymakers should consider the role that insti-tutional mechanisms play in determining whodoes and does not save.

Finally, designing retirement savings instrumentsfor low- and moderate-income households isparticularly difficult because many face financialhardships during their working lives, and do notcurrently participate in pension plans or ownretirement savings accounts. These households’financial vulnerability and lack of alternativesources of funds will likely lead them to seekgreater pre-retirement access to individualaccount funds than would wealthier workerswho now have pensions and/or retirement savings.

Summary

Retirement income in the United States is oftencharacterized as a “three-legged stool” made upof Social Security, employer-sponsored pensions,and individual savings. But many Americans donot have employer-sponsored pensions andmany have saved very little for retirement. Forthis reason, Social Security has been the mainsource of income for older Americans for severaldecades. The next 25 years will see the SocialSecurity benefits for 65-year-old retirees growmore slowly than wages, though still faster thaninflation, meaning that Social Security willreplace a smaller portion of their previous earn-ings. Because the system is not in long-range


balance, changes in benefits, revenues, or bothare being contemplated. The future level of thedefined-benefit part of Social Security will be animportant consideration in the design of payoutrules for individual accounts.

About half of private-sector workers are notcovered by a pension plan at any given time.Increasingly, employers are changing defined-benefit pension plans to defined-contributionplans, where employees assume more risk andresponsibility, and have more choices concerningtheir retirement savings.

About half of American families do not haveretirement savings accounts such as 401(k)s orother defined-contribution pensions, individualretirement accounts (IRA), or Keogh plans.Many workers, including some from middle-income households, are not building up tax-favored savings to supplement Social Security inretirement.

Other assets and liabilities also affect retirementsecurity. Home equity is the largest asset formost American families, yet more than half ofblack and Hispanic families (as well as one infour other families) lack this important resource.“Asset poverty” – that is, assets insufficient tosustain a family through three months withoutincome – is a risk for many middle-incomeAmericans.

About one in ten families does not have a bankaccount and a similar number do not file taxeswith the IRS. Special considerations might berequired to help these families both save andparticipate in the financial mainstream.

Between 9 and 10 percent of workers coveredby Social Security are self-employed.Presumably, these individuals would be responsi-ble for managing any administrative tasks dele-gated to employers in a new individual accountsystem.


Chapter Two Endnotes

1 The full benefit age will gradually rise from 65 to66 for those born between 1938 and 1943 andwill later rise from age 66 to 67 for those bornbetween 1955 and 1960.

2 The size of tax expenditures associated withhealth care insurance, home ownership andretirement plans can be calculated differentlydepending on the definition of these categories.For example, there are separate estimates for theexclusion of employer contributions for medicalinsurance premiums and medical care, for self-employed medical insurance premiums, for med-ical savings accounts, for workers’ compensationand other health related categories. Likewise,there are separate estimates of the tax loss associ-ated with employer plans, 401(k) plans, individ-ual retirement accounts, Keogh plans, and otherretirement related deductions. It should be notedthat tax expenditures do not necessarily equalthe increase in federal revenue that would resultfrom the repeal of provisions because tax expen-ditures are interactive and because economicbehavior could change.

3 About 70 percent of tax benefits from new con-tributions to defined-contribution plans accrue tothe highest income level (20 percent of tax filingunits in 2004); almost 60 percent of IRA taxbenefits accrue to the top 20 percent of house-holds (Burman et al., 2004).

4 In 2003, the federal poverty line for an individ-ual was $8,980 for a year and for a family offour it was $18,400 a year. The poverty line hasbeen widely criticized for not ensuring even aminimum standard of living and for not includ-ing all income sources available to families.Nevertheless, it remains the common standardfor measuring poverty and economic hardship.

5 The measure of basic needs is based on work byCitro and Michael (1995), which examined alter-native measures of poverty for families with chil-dren. That threshold for a family of four (twoadults and two children) in 1997 was just under$16,000 a year.

6 Individuals who have declared bankruptcy maynot file again for seven years.

7 It is assumed that the funds are invested half instocks, with an average real return of 6.5 per-cent, and half in corporate bonds, with an aver-age real return of 3.5 percent. The portfolio’saverage real return of 5.0 percent becomes 4.7percent after deducting administrative costs of0.3 percent.

8 The relationship of balances and annuities towages will be similar across wage levels as longas wage patterns over the lifetime are the sameacross wage levels and assumptions about invest-ment returns are the same. The assumptions aredescribed in Figure 2-17.

3Chapter

Payments at Retirement

45

Retirement creates a new set of financial risksfor any individual. Policymakers designing anindividual account system must decide how, andto what extent, to protect retirees against vul-nerabilities. Some protections might require con-straining individual choice through rules thatlimit or regulate a retiree’s access to accountfunds. In this regard, when and how paymentswould be made to account holders during retire-ment are key policy questions in individualaccount design.

This chapter describes the financial risks thatretirees face. In light of these risks, the chaptersets out a framework policymakers might use inanalyzing individual account proposals. A shortprimer on annuities offers clarity to the manyproposals that include some form of life annuity.

Four broad options are presented for retirementpayouts. The first option provides forUnconstrained Access. It is based on privatelymanaged individual retirement accounts (IRAs)and, to some extent, the federal employees’Thrift Savings Plan (TSP), which is often sug-gested as a precedent for individual accounts.The second option, Compulsory Annuities with

Special Protections, falls at the other end of thespectrum in terms of protection and individualchoice. This option includes some of the protec-tions currently provided by Social Security, andit gives retirees minimal discretion over access totheir account funds. The third option, DefaultAnnuities with Special Protections, transformscompulsory annuities into a default choice whileoffering several other options that retirees couldtake instead. The last option, CompulsoryMinimum Annuities, calls for annuities similarto those in the second option, but requires themonly up to a given level.

Many questions arise about the design andimplementation of retirement payouts. Thischapter examines how mandatory joint-and-sur-vivor annuities affect couples under different cir-cumstances, for instance one-earner couplescompared to dual-earner couples, what happenswhen spouses are different ages or when onlyone spouse participates in an individual accountplan, and how widowhood could affect the tim-ing of an annuity purchase. Whether or not theinterest of heirs should be considered whendesigning annuity rules is examined, as is howto provide individual account holders with accu-


rate and complete information concerning retire-ment payout choices.

Financial Risks in Retirement

No matter how diligently an individual hassaved and planned over the years, a new set offinancial risks loom once the working life ends.Social Security is public recognition that retireesshould be protected from particular financialvulnerabilities. Although the risks of retirementmay be inevitable, retirement income can makea substantial difference in a retiree’s economicwell being.

Retirees face potential income insecurity in fourkey areas: uncertainty about how long they willlive (longevity risk), loss of purchasing powerover time (inflation risk), unpredictable invest-ment returns (investment risk), and uncertaintyabout the longevity of one’s spouse (spousal sur-vivorship risk). In addition, retirees face the possibility of unpredictable changes in theirfinancial needs due to prolonged illness, long-term care, family expenses, and a host of otherreasons.

Longevity Risk No one knows how long he or she will live. Theaverage American woman at her 65th birthdayhas a life expectancy of about 20 years; her male

counterpart has about 17 years (Figure 3-1). Noone knows if he or she will be “average.” Thevaried life spans in Figure 3-1 show the difficul-ty of trying to estimate one’s remaining yearsfrom an average. For example, about 11 percentof men and 7 percent of women reaching age 65will die before they reach age 70. Yet, about 6percent of men and 14 percent of women willlive more than 30 years, to beyond their 95thbirthdays.

This uncertainty makes it difficult to allocatemoney wisely throughout retirement. If retireeslive longer than they planned for, they risk run-ning out of funds. If they try to avoid that prob-lem by spending conservatively, they may have alower standard of living than they could other-wise afford.

Inflation Risk

Even a modest rate of inflation can significantlyerode the long-term purchasing power of a fixedincome. Annual inflation of just 3 percent—wellwithin the range experienced in the UnitedStates in recent years—will make $100 todayworth only about $74 in 10 years. After 25years, the value would drop by more than half,to about $45. With Americans living longer thanever before, inflation protection is becomingeven more urgent. As shown in Figure 3-1, near-

Figure 3-1. Cohort Life Expectancy of Americans Age 65 in 2005

Total Men Women

Average life expectancy at age 65 18.3 16.8 19.8Percent who will be alive at age

70 90.7 88.8 92.675 78.9 74.9 82.680 63.9 58.0 69.485 45.6 38.4 52.390 25.7 19.0 31.895 10.2 6.3 13.7

100 2.6 1.3 3.7

Source: U.S. Social Security Administration, 2003a. Unpublished calculations from the Office of the Chief Actuary; Board of Trustees,2003. Annual Report of the Board of Trustees

Chapter Three: Payments at Retirement 47

ly one woman in three and one man in five atage 65 is projected to live at least 25 more years.

The volatility of inflation over long time periodsis a large part of inflation risk. In the threedecades since Social Security implemented cost-of-living adjustments, the annual adjustment inSocial Security benefits has ranged from a lowof 1.3 percent to a high of 14.3 percent (Figure3-2). Over a single five-year period (1977-1982),benefits rose by 60 percent just to preserve theirpurchasing power. Although inflation has hov-ered between 2 percent and 4 percent since theearly 1990s, these examples from the not-so-distant past warn against assuming that recenttrends will hold.

Investment Risk Some retirees invest in financial markets to pro-tect themselves against loss of purchasing power.But such investments yield uncertain gains orlosses, producing both year-to-year incomevolatility and an unknown income stream overthe entire retirement period. In general, the high-er-risk investments—such as corporate stocks—offer a higher expected return, but they alsopose greater risk of financial loss.

Spousal Survivorship Risk Married retirees could also suddenly have tofend for themselves if they divorce or are wid-owed. Concerns about surviving alone are par-ticularly acute for women, who have longer lifeexpectancy and tend to be younger than theirhusbands. Also, retired widows are more likelythan men to remain unmarried, in part becauseunmarried women outnumber unmarried menafter age 65 by about three to one.

To maintain a comparable standard of livingafter widowhood, the newly single spouse needsabout two-thirds to three-quarters of the cou-ple’s prior income. The official poverty thresholdpegs the needs of a single elderly person atabout 79 percent of the needs of a couple.2

Uncertain Spending Needs Retirees also face uncertainty about how muchit will cost to meet their needs in old age. Highout-of-pocket medical expenses or the costs oflong-term care could wreck even the best-laidplans. Retirees face limited options for adjustingto financial setbacks. During the work life, indi-viduals can adapt by working more, delaying

0

2

4

6

8

10

12

14

16

1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005

Perc

ent

Figure 3.2. Cost-of-Living Adjustments in Social Security Benefits, 1975-20041

Source: U.S. Social Security Administration, 2004c. Automatic Increases, Cost-of-Living-Adjustments


retirement, or spending less. In advanced oldage, spending less might be the only option.

Framework for AnalyzingRetirement Payout Options

Policymakers might consider a range of factorsin designing individual account payout rules,with one crucial factor being the system’s rela-tionship to Social Security.

Intended Use of Account Funds If the main purpose of the accounts is to providebasic income security, then retirement payoutrules might aim to resemble features of SocialSecurity. If the main purpose of the accounts isto expand ways to save in addition to SocialSecurity, then payout rules might resemble thoseof 401(k) plans, the federal employees’ ThriftSavings Plan, or individual retirement accounts(IRAs). If the purpose of the accounts is toexpand wealth creation during the work life –through earmarked savings for homeownership,higher education, or a business enterprise, forexample – then the accounts would normallyhave been used long before retirement andretirement payout rules would not be a majorissue. This chapter focuses on accounts that arefor retirement.

Level of Social Security Benefits The lower the level of remaining Social Securitybenefits, the stronger the case would be forrequiring that individual account payouts pro-vide the kinds of protections found in SocialSecurity, such as inflation indexing, lifetime pay-ments, and spousal protections. On the otherhand, if Social Security is thought to provide anadequate baseline of retirement income, moreflexible payment options might be favored.

Voluntary or Mandatory Participation If workers have a choice whether or not to par-ticipate in the individual accounts, unpopularpayout rules could deter workers from partici-pating. Workers might choose other forms of

retirement savings or decide to save less to avoidthe rules.

If participation were required, then unpopularpayout rules would not reduce participation.Highly unpopular payout rules could, however,weaken public support for the system altogether,or the rules could be overturned through the leg-islative process.

Tax Treatment of Account Funds Subsidies or favorable tax treatment create acase, at least arguably, for structuring payoutrules to ensure qualifying for the tax preference.Minimum distribution rules in tax-favoredretirement plans, for example, require that atleast part of tax favored retirement savings areused and taxed in retirement rather than usedsolely for bequests.

Private or Public Management ofPayouts Finally, the payout rules might vary dependingon whether account payouts are provided by thegovernment or by private companies—and inthe case of annuities, whether the governmentserves as annuity provider or as annuity regulator.

The interplay of these factors can be seen in theexisting proposals for individual accounts,which vary greatly in their aims and, conse-quently, in their retirement payouts rules (Figure3-3). Plans that envision individual accounts as apartial replacement for Social Security often callfor compulsory inflation-indexed annuities—asin the plan offered by the chair of the 1994-1996 Advisory Council on Social Security and invarious Social Security proposals since then.Other plans would require annuities only up toa specified monthly income, including proposalsby Kolbe-Stenholm, DeMint-Armey, Gramm-Hagel, and the 2001 President’s Commission toStrengthen Social Security. Finally, proposalsthat envision the retirement accounts as separatefrom Social Security do not require annuities, asin the Social Security Plus plan proposed byRobert M. Ball, former Commissioner of Social


Figure 3-3. Retirement Payout Rules in Selected Individual Account Proposals

Proposal Retirement Payout Provisions

Mandatory Accounts Funded with New Contributions

ACSS (Gramlich): • Automatic CPI-indexed annuitization of full account balance at retirement, withIndividual Account Plan, 1996 a 1-year certain annuity guarantee.

• Default joint-and-survivor annuity for married retiree, unless spouse declines.

Committee on Economic • Committee “favors ensuring that funds are withdrawn gradually over the life ofDevelopment, 1997 the participant, (as would occur if account were annuitized at retirement).”

Mandatory Accounts Funded with Scheduled Social Security Taxes

Reps. Kolbe-Stenholm: Bipartisan • No withdrawal until retirement or until account balance is sufficient to purchaseRetirement Security Act of 2004 annuity (or phased withdrawal) equal to at least 185% of poverty level. Solvency(H.R. 3821 in 108th Congress) estimates assume CPI-indexed annuities.

• Requires annuitization of the account balance such that when the account annuity is combined with the Social Security benefit, it guarantees a payment equal to 185% of the poverty level. Any remaining account balance may be taken as a lump sum.

Voluntary Accounts Funded with New Contributions from Workers

Clinton Retirement Savings • No annuitization requirement. Accounts, 2000

Social Security Plus (proposed • No annuitization requirement. Payouts would follow IRA rules. by R.M. Ball, 11/2003)

Voluntary Accounts Funded with Scheduled Social Security Taxes

President’s Commission (PCSSS) • Joint-and-survivor annuity required for married retirees. Models 1,2, 3 (2001) • Annuity or gradual withdrawal required for all others.

• Can take as a lump sum the portion of the balance that exceeds the amount needed to provide a monthly payment (Social Security + the annuity) above the poverty line. Solvency estimates assume CPI-indexed annuities or variable annuities.

Reps. DeMint-Armey: Social • Compulsory CPI-indexed annuitization of a minimum of 40% of the accountSecurity Ownership and balance with additional amounts annuitized as needed to guarantee a combinedGuarantee Act of 2001 Social Security benefit (reduced by an offset) plus annuity equal to 100% of(H.R. 3535 in 107th Congress) the poverty level.

• Any remaining funds could be withdrawn as a lump sum.

Unspecified General Revenues for Accounts

Rep. Shaw: Social Security • Compulsory scheduled withdrawals would be transferred to Social SecurityGuarantee Plus Act of 2003 Trust Funds to pay defined benefits.(H.R. 75 in 108th Congress) • At retirement or disability, 5 percent of account balance paid as lump sum.

Sources: U.S. Social Security Administration, 2004b, Selected solvency memoranda; National Academy of Social Insurance, December1996, Social Insurance Update; Robert M. Ball, 2003, Social Security Plus, and November 2002 communication; Committee on EconomicDevelopment, 1997, Fixing Social Security


Figure 3-4. Five Strategies to Make Income Last a LifetimePurchasing Power of Income Stream from Age 65 to 115

Strategy #1. If a 65-year-old bought a fixed life annuity with $100,000, it would pay about $9,700 a year,assuming unisex pricing. Given annual inflation of 3 percent, after 25 years the annuity would be worth justover $4,630. If the annuitant lives to 110, the purchasing power would fall to about $2,565.

Strategy #2. If the same person bought an inflation-indexed (real) life annuity, the payments would startout lower, at about $7,450 a year, but they would rise with inflation. The annuity would maintain its pur-chasing power for as long as the annuitant lived.

Strategy #3. If the annuitant followed a fixed withdrawal strategy, each year he or she would simply with-draw about $9,700 from the account—the amount available from a fixed annuity. This person saves theupfront purchase price of an annuity, but would not be able to provide $9,700 annually for life, since thisapproach lacks the annuity provider’s risk pooling advantage. If this individual continued to withdraw$9,700 a year, the money would run out after about 16 years and the individual would have nothing left.

Strategy #4. A one-over-life-expectancy approach might make the money last longer. Each year the individ-ual would divide the total account balance by remaining life expectancy and withdraw the resulting amount.For example, if at 65 the individual expected to live 18.3 more years, he would withdraw (1/18.3 x$100,000) from his savings and leave the rest to earn investment returns. The next year, he would divide hisremaining resources by 17.3, and withdraw accordingly. While this strategy would make his money lastlonger than Strategy #3, withdrawals would decline to negligible amounts if this person lives much beyondage 90.

Strategy #5. Finally, this individual could amortize-to-age-100 and try to make the money last until he orshe reaches 100 years of age. One would choose an investment with a secure return and then determinehow much he could withdraw each year to make his money last until his 100th birthday. If he lived beyondthat age, he would have no income.

Source: Brown, 2003. “Redistribution and Insurance: Mandatory Annuitization with Mortality Heterogeneity”

2000

4000

6000

8000

10000

12000

Age

Strategy #1 Strategy #2 Strategy #3 Strategy #4 Strategy #5

Co

nst

ant

year

200

0 d

olla

rs

65 69 73 77 81 85 89 93 97 101 105 109 113 115

24367 NASI TEXT 1/17/05 11:36 AM Page 50


Security, and President Clinton’s RetirementSavings Accounts.

A Primer on Life Annuities

A life annuity is a financial product that canguarantee money will last for the rest of aretiree’s life.3 When an individual buys a fixedlife annuity from an insurance company, theinsurer assumes a contractual obligation to paythe annuitant a guaranteed income for life, ineffect, transferring the mortality risk and invest-ment risk to the insurance company. The insurerpools the mortality risk among a large group ofannuitants, with the extra funds from those whodie early used to cover the annuity costs of thosewho live a long time.

Comparing Annuities to other Strategies Economic analyses indicate that a life annuitywould be a rational choice for a person whowanted to ensure income for life (Davidoff et al.,2003; Diamond, 2004). Consider a healthy 65-year-old who has a savings account of $100,000and who wants to produce income for life.Figure 3-4 compares two annuity options andthree “do-it-yourself” strategies that this individ-ual could employ. As the figure indicates, anannuity would be an advantageous choice.

The lesson from Figure 3-4 is that only the twoannuity options provide the certainty of incomefor life and, of the two annuities, the inflation-indexed payments start out somewhat lower butsoon surpass the purchasing power of the fixedannuity.

But guaranteed lifelong income comes withtradeoffs. An annuity requires the buyer to paythe full price up front (in this example,$100,000), and the purchase is irrevocable.Strategies 3, 4, and 5, by contrast, allow theretiree to keep the money not consumed. Theunused money remains available for emergenciesor other unforeseen spending, it can be invested,and it can be set aside for bequest.

Annuity Features A basic life annuity covers the risk of outlivingone’s income, but additional annuity featurescan help mitigate other retirement risks such asloss of purchasing power and the loss of aspouse’s income. Below, we focus on three annu-ity features that insure against inflation, widow-hood, and loss of bequests if the annuitant diesearly.

Fixed, Rising, Inflation-Indexed, or VariableLife Annuities

Fixed annuities pay a flat dollar amount for thelife of the annuitant; as such, they lose purchas-ing power over time, as illustrated in Figure 3-4.Some insurance companies offer rising or gradedannuities with payments that increase at a speci-fied rate, such as 3 percent a year. These rela-tively easy-to-design annuities provide partialprotection against the erosion of purchasingpower. Neither of these two options guaranteesagainst purchasing power loss due to unexpect-ed inflation.

Inflation-indexed annuities, by contrast, areadjusted each year by the recent rate of infla-tion, much as Social Security benefits are nowadjusted. Currently, the U.S. market for infla-tion-indexed annuities is almost non-existent;few insurers offer them and few consumers, ifany, actually buy them. Inflation-indexed annu-ities are somewhat more common in the UnitedKingdom, but still comprise only about 10 per-cent of the life annuity market (Finkelstein andPoterba, 2000).

Finally, a variable life annuity is tied to the per-formance of a particular investment portfolio,such as corporate stocks or bonds. Dependingon the portfolio chosen, variable annuities havethe potential to pay higher average returns (par-ticularly over longer time periods), but paymentscould decline if the underlying assets lose value.While other types of life annuities shift invest-ment risk to the insurance company, individualannuitants bear the investment risks and returnswith variable life annuities.


Single-Life or Joint-Life Annuities

A single-life annuity pays monthly income forthe life of one person. A joint-and-survivorannuity pays a guaranteed stream of income fortwo lives: the primary annuitant (the individualbuying the annuity) and an annuity partner orsecondary annuitant (usually a spouse). Joint-lifeannuities can take several forms, including thefollowing typical examples:

Full Benefit to Survivor. The death of either theprimary annuitant or the partner produces nobenefit reduction to the survivor. Whoever livesthe longest will continue to receive the full annuity.

Two-thirds Benefit to Survivor. The death ofeither the primary annuitant or annuity partnerreduces the income of the survivor to two-thirdsof the original amount. Under this option, theprimary annuitant foregoes the guarantee of fullbenefits for life in exchange for spousal protection.

Half Benefit to Annuity Partner. If the annuitypartner is widowed, the payment is reduced byhalf. If the primary annuitant is widowed, theoriginal annuity amount continues. Here, unlikein the cases above, the annuity partner is treatedless generously than the primary annuitant whenwidowed. This option is the default spousal pro-

tection in private pension plans under theEmployee Retirement Income Security Act(ERISA).

Guarantee Payments when Annuitants DieEarly

Some annuity contracts guarantee a payment tothe annuitant’s named beneficiary if the annui-tant dies within a prescribed period after thepurchase. For example, under a 10-year certainannuity, the annuitant’s death within 10 yearsafter the initial purchase triggers payments forthe remaining period to the annuitant’s namedbeneficiary. Alternatively, a refund of premiumannuity guarantees that the annuity will pay outat least the nominal purchase price. If the annui-tant dies before the total nominal paymentsreceived total the initial purchase price, thenamed beneficiary receives a lump sum equal tothe difference.

Price Comparisons

Each additional layer of inflation or survivorprotection lowers the initial monthly annuitypayment a given premium will purchase relativeto a fixed, single-life annuity. A 65-year-old withan account worth $10,000 could buy a flat, sin-gle-life, monthly annuity of about $80, as shownin Figure 3-5. If he wanted inflation protection,an annuity set to rise 3 percent each year wouldstart out lower, at about $62 a month, but

Figure 3-5. Effect of Inflation Indexing and Survivor Protection on Initial Monthly Annuity for$10,000 Premium: 65-Year-Old with 65-Year-Old Spouse

Type of Annuity Initial Monthly Payment

Single-life, flat monthly payment $80.46Single-life, inflation indexed $61.78

Joint-Life, Inflation-Indexed Annuities

100 percent survivor benefit $50.18Two-thirds survivor benefit $57.39

Payment to survivor $38.28

Annuity estimates are based on assumption that: purchase of annuities is mandatory (reflecting total population life tables for individualsage 65 in 2005); annuities are priced the same for men and women; the annual inflation rate is 3.0 percent; and the real annual interestrate is 3.0 percent. Joint-life annuities are symmetric, in that they pay the same amount to whichever spouse lives longer.



would maintain purchasing power over time. Ifthat indexed annuity were to cover both theannuitant and his 65-year-old wife at full value,the monthly payment would start out lower still,about $50 a month. Reducing survivor benefitsto two-thirds would raise the initial payment toabout $57 a month, but would lower the wid-owed partner’s benefit to about $38 a month.Assumptions underlying these estimates aredescribed in Box 3-1.

Guarantees also lower the monthly annuity thatwould otherwise be paid. A 10-year certain fea-ture would lower the indexed, single-life annuityfor this 65-year-old from about $62 to $58, or 6percent; a refund of premium feature would bemore expensive, reducing the payment by about11 percent, from $62 to about $55 (Figure 3-6).Many economists believe guarantees are a rela-tively expensive way of providing protection forone’s heirs in the case of an early death, as dis-cussed in more detail below.

Figure 3-6. Effect of Inflation Indexing and Guarantee Features on Initial Monthly Annuity for$10,000 Premium: 65-Year-Old

Type of Annuity Initial Monthly Payment

Fixed, single life, flat monthly payment $80.46Indexed 3 percent – no guarantee $61.84Indexed 3 percent – 10-year certain $58.43Indexed 3 percent – refund of premium $54.78

Annuity estimates are based on assumptions that: purchase of annuities is mandatory (reflecting total population life tables for individualsage 65 in 2005); annuities are priced the same for men and women; the annual inflation rate is 3.0 percent; and the real annual interestrate is 3.0 percent.


Box 3-1. Annuity Estimates and Assumptions

To price an annuity, the provider must make assumptions about: (a) the interest rate it will receive on theinvested annuity premiums; (b) expenses for administration and profit; and (c) the pattern of remaininglife expectancy of the pool of annuitants. Inflation-indexed annuities also require making separateassumptions about the inflation rate and the real (net of inflation) interest rate.

The Office of the Chief Actuary of the Social Security Administration provided the annuity examples usedin this chapter unless otherwise noted. The estimates are consistent with long-range intermediate, or“best estimate” assumptions in the 2003 report of the Social Security Trustees. The annual inflation rateis assumed to be 3.0 percent. The interest rate on special issue Treasury securities (in which Social Securitytrust funds are invested) is assumed to be 3.0 percent in excess of inflation, or 6.1 percent total (1.03 x1.03 = 1.0609 or 6.1 percent, rounded). This 6.1 percent interest rate is used for most of the annuityexamples in this report.

These estimates assume that the purchase of annuities is mandatory by normal retirement age and thatannuities are priced the same for men and women. The calculations are theoretical in that they reflect noexplicit expense for administrative fees, risk, or profit. Other issues affecting the pricing of annuities in theprivate market are discussed in Chapter Four – Institutional Arrangements.


Few People Buy AnnuitiesAlthough annuities help people offset some ofthe financial risks of retirement, relatively fewpeople actually buy them. As said in ChapterTwo, the U.S. private retirement system is mov-ing away from pension plans that pay setmonthly benefits (defined-benefit plans).

While Social Security and some defined-benefitpensions continue to pay benefits only in theform of monthly benefits, other retirement plansare shifting to lump-sum distributions. Whengiven a choice between taking a lump sum orbuying an annuity, people usually take the lumpsum. These trends are evident in private pen-sions and 401(k) plans, in the Thrift SavingsPlan for federal employees, and in TIAA-CREF.

A growing number of defined-benefit pensionsare offering the choice of taking a lump suminstead of monthly benefits for life. A study ofover 1,500 participants in defined-benefit plansthat offered lump sums as well as immediate ordeferred annuities found that 88 percent ofworkers who left the plans took lump sums(Watson Wyatt, 1998). Participants might haverolled the money over into other tax-favoredretirement savings plans, or used it for otherpurposes.

Private 401(k) plans rarely pay annuities. Arecent survey of several hundred 401(k) plansponsors found that all offered lump-sum distri-butions while a declining number offered annu-ities. Between 1999 and 2003, plans with anannuity option declined from 31 percent to 17percent and only 2 percent of participants choseannuities (Hewitt Associates, 2003). Some indi-viduals retiring with 401(k) accounts use theiraccounts as supplements to both defined-benefitpensions and Social Security. Thus, these indi-viduals might feel they have enough wealth inthe form of monthly income and prefer, instead,to hold the 401(k) funds in a form that can bebequeathed to heirs.

The Thrift Savings Plan for federal employeesoffers annuities as one of several forms of pay-

out, but even its favorable annuity rates attractfew takers. In 2002, about 54,000 civilianemployees retired on immediate pensions fromfederal jobs, but only 870 used their TSP fundsto buy annuities (FRTIB, 2002).4

Finally, the Teachers’ Insurance AnnuityAssociation and College Retirement EquityFunds (TIAA-CREF) is the largest privateprovider of defined-contribution retirementplans in the United States. As its name implies,life annuities have been a key feature since thesystem began in 1918. In fact, until 1989, lifeannuities were the only form of retirement pay-ments the plan offered. Since then, however, par-ticipants have been allowed to choose amongnon-annuity payouts, including an “interestonly” benefit, a minimum distribution option,or systematic withdrawals. In all of these casesthe individual retains ownership of the funds notyet withdrawn. After these non-annuity pay-ments became available, the portion of retireeschoosing annuities declined from 94 percent in1990 to 43 percent in 2003 (Ameriks, 2002).These findings refer only to the participant’s firstdecision about the form of distribution – someretirees may later decide to buy an annuity.

Many explanations have been offered for whypeople do not choose annuities (Brown andWarshawsky, 2001; Mackenzie, 2002). First,many retirees with funds available to annuitizemay already have sufficient monthly income. Infact, TSP participants receive monthly incomesfrom both Social Security and defined-benefitpensions in addition to their TSP accounts. Asindicated in Chapter Two, many private employ-ees with 401(k) plans have both Social Securityand defined-benefit pensions. For a retiree whocan count on receiving two monthly checks forlife, the risk of outliving one’s assets might seemdistant—and the marginal value of further annu-itization small.

In addition, when facing the tradeoff betweenthe upfront loss of the purchase price in returnfor a fixed future income, many people may pre-fer to retain ownership of their money and


invest it themselves. Retirees might want savingsin case of uninsured health expenses or long-term care, or they might want to bequeath themoney to an heir. An unwillingness to bind one’sincome in an annuity may explain the popularityof TIAA-CREF’s recently introduced non-annu-ity options; a retiree could always decide to buyan annuity later, but that road can only becrossed once.

Related to this desire to keep one’s options openis the concept of wealth illusion. People tend toput a much higher value on a pot of money inhand than on a stream of future income of equalvalue (Kahneman, 1999). Studies also indicatethat people tend to be shortsighted and are par-ticularly myopic when choosing between imme-diate gratification and long-term gains(Fetherstonhaugh and Ross, 1999). In principle,people want their lives to improve and be secureover time; in practice, they are often driven byshort-term temptations or concerns. Further,people are more distressed by prospective lossesthan they are pleased by the prospect of equal oreven larger gains. When buying an annuity, thesacrifice is immediate and tangible while thegains are uncertain and distant (Loewenstein,1999).

Finally, on the supply side, insurers do notactively market life annuities. Insurance compa-nies largely emphasize life insurance and vari-able deferred annuity products over lifeannuities. Most variable annuities sold in theUnited States today are used solely as investmentproducts and are rarely converted into life annuities.

Tradeoffs between Optional andCompulsory AnnuitiesThe designers of retirement payouts from indi-vidual accounts face basic tradeoffs in the prosand cons of compulsory or optional annuities.The tradeoffs include adverse selection, whichaffects annuity costs, the treatment of peoplewith different life expectancies, the risk of out-living one’s money, and how individual accountsinteract with means-tested programs.

Compulsory Annuities Assure that PeopleWill Not Outlive Their Money

Compulsory life annuities would eliminate therisk that people would outlive the money intheir accounts. Making annuities optional wouldremove that guarantee from people who choosenot to purchase annuities.

Optional Annuities Cost More

Optional annuities generally pay less for anygiven premium because of what insurers call“adverse selection.” People with short lifeexpectancies tend to not buy annuities, whilepeople who expect to live a long time are morelikely to annuitize. It is estimated that the size ofprivate individual annuity payments in theUnited States is reduced by 6 to 12 percentbecause of adverse selection (Mitchell et al.,1999; Brown et al., 2000; Liebman, 2002). Ifannuities were compulsory, payouts for anygiven premium would be somewhat higher,although the extent of such an improvement isnot clear.

Optional Annuities Are More Acceptable tothe Sick and Short-Lived

Higher payouts from compulsory annuitiesoccur precisely because people who are sick,dying, or who otherwise have short lifeexpectancies are forced to pay full premium,despite the likelihood of fewer payments.Optional annuities break up the collective riskpool by allowing those who are ill, dying, orshort-lived to opt out of the purchase. This fea-ture of annuities may be of particular concernfor certain subgroups, as life expectancy differsamong socioeconomic groups (Brown, 2003).

Compulsory Annuities May Reduce Relianceon Means-Tested Benefits

Some policymakers want to avoid creatingopportunities or incentives for individuals toqualify for means-tested programs, such asSupplemental Security Income (SSI) or Medicaid.This concern could interact with annuity policy.If retirees’ entire monthly incomes (includingtheir potential annuities) would be about equalto the SSI threshold, they could improve their


quality of life by spending the funds and thenrelying on SSI for monthly income. Optionalannuities would permit this; compulsory annu-ities would not.5 Low-income individuals wouldbe required to use their accounts to buy annu-ities so as to reduce their reliance on SSI.

Retirement Payout Options

This section describes four options for retire-ment payouts. Option One: UnconstrainedAccess offers broad account holder discretionabout the form and timing of payouts. OptionTwo: Compulsory Annuities with SpecialProtections goes to the other end of the spec-trum: it is designed to resemble aspects of SocialSecurity, including mandatory, inflation-adjustedlife annuities and spousal protections. Theseprotections also appear in Option Three:Default Annuities with Special Protections, butas a default rather than a mandate. Similarannuities are again compulsory in Option Four:Compulsory Minimum Annuities, but only up toa specified threshold. Many variations on thesebroad options are possible.

Option One: Unconstrained AccessThis option would give retirees a great deal ofdiscretion about the form and timing of retire-ment withdrawals. It is based on privately man-aged individual retirement accounts (IRAs) and,in some respects, the Thrift Savings Plan (TSP)for federal workers.

The TSP is designed to resemble a 401(k) planfor federal employees, with optional participa-tion and government (employer) matching con-tributions. The TSP provides a third tier ofretirement income in addition to each federalemployee’s defined-benefit pension and SocialSecurity. (For married participants, the TSP con-strains access in some important ways; for moreon spousal rights, see Chapter Six.)

Four payout choices would be made available toretirees under this option: leaving the money inthe account; withdrawing the entire amount as a

lump sum; buying an annuity; or taking phasedwithdrawals.

Leaving Money in the Account

Payout rules might impose certain limits onretirees’ choice to leave small amounts of moneyin their accounts indefinitely. First, the TSP’sminimum balance rule provides that any TSPaccount containing less than $200 will be auto-matically cashed out to the employee when heor she leaves federal employment. This mini-mum balance rule is designed to eliminateadministrative costs for very small accounts.

Second, in return for granting tax-favored statusto retirement accounts such as the TSP or IRAs,the Internal Revenue Code (IRC) requires thatretirees take minimum distributions after a cer-tain age. The purpose of this rule is to ensurethat at least part of the funds set aside for retire-ment are, in fact, used and taxed in retirement,instead of being bequeathed to heirs. So mini-mum distribution rules could apply to any indi-vidual account system with favorable taxtreatment.

Under the IRC, minimum distribution rulesapply the year after a retiree reaches age 701/2. Ifa person over the statutory age fails to take theminimum amount and count it as taxableincome, he or she would owe a tax penaltyequal to half the mandated distribution. TheTSP helps its retirees avoid tax penalties by cal-culating and disbursing the account in accor-dance with minimum distribution rules.Presumably, individual account administratorscould perform this same function.

Taking a Lump Sum

Unmarried TSP retirees can withdraw theiraccount funds or transfer the funds to other tax-favored retirement plans.6 The withdrawn fundsare subject to income tax, and an additional 10percent penalty if the TSP retiree is not at leastage 55. Individual retirement account (IRA)withdrawals are also taxable and subject to a 10percent penalty if the withdrawal is taken before


age 591/2. Similar rules could apply to individualaccounts.

Buying Annuities

TSP participants with at least $3,500 in theiraccounts can buy annuities. Through competi-tive bidding, the TSP Board selects an insurancecompany that provides annuities to participantsunder terms set by the TSP, with a variety ofannuity choices offered (Box 3-2). The currentprovider is Metropolitan Life InsuranceCompany (MetLife). The TSP helps participantsbuy annuities but then drops out of the process.Once the annuity is purchased, the annuitant isno longer in the TSP and he or she deals directlywith MetLife on all further arrangements withthe annuity.

Phased Withdrawals

Finally, instead of buying an annuity, an unmar-ried retiree (and a married retiree who obtainsspousal consent, if applicable) can leave his orher money in the TSP and choose one of the fol-lowing options for receiving substantially equalmonthly payments:

(a) Monthly payments computed by the TSPbased on IRS life expectancy tables. Thepayment is calculated according to the per-son’s age and account balance and is updat-ed annually. (This method is similar toStrategy #4 illustrated in Figure 3-4.)

(b) Specified dollar payments. The retiree canchoose a monthly payment of at least $25.The TSP will pay the specified amount untilthe balance is gone.

This Unconstrained Access option offers retireesa great deal of flexibility, possibly subject to cer-tain spousal rights. (For further discussion ofspousal rights, see Chapter Six.) It reflects thepurpose of the TSP as a system for optional,tax-favored saving on top of Social Security anddefined-benefit pensions. The unconstrainedaccess approach would have drawbacks if thepurpose of individual accounts were to producebasic security for retirees and their families whomight otherwise lack adequate defined-benefitincome.

Box 3-2. TSP Annuity Choices

Single life with level payments or increasing payments. The latter rises with the cost of living, but notmore than 3 percent per year.

Joint-and-Survivor with a spouse with level payments or increasing payments. The annuitant canchoose either a 100 percent survivor annuity or a 50 percent survivor annuity.

Joint-and-Survivor with someone other than a spouse with level payments only. The same two sur-vivor options are available as for joint annuities with a spouse. In addition, an annuitant can choose thefollowing features to provide guaranteed payments if the annuitant dies early:

Cash refund. This option ensures that the total annuity amount paid is at least equal to the purchaseprice of the annuity. If the annuitant dies before total payments equal the purchase price, the differencewill be paid to a named death beneficiary or to the annuitant’s estate.

Ten-year certain. Payments will continue for ten years even if the annuitant dies before that time. If theannuitant dies after eight years, for example, the named death beneficiary would receive payments fortwo more years. This option is not available under a joint-life annuity.


Option Two: Compulsory Annuities withSpecial ProtectionsOption Two resembles, by design, aspects ofSocial Security for both retirees and their wid-owed spouses. Current spousal benefits arebriefly described in Box 3-3.

This option has three key features: (1) retireeswould be required to buy annuities with theiraccount funds; (2) the annuities would have tobe indexed for inflation; and (3) married retireeswould be required to purchase joint-life annu-ities with their spouses. The goal is to produce astream of income that resembles three aspects ofSocial Security: monthly payments that last forlife, that maintain their purchasing power, andthat provide continuing income to widowedspouses.

Social Security provides supplemental benefits tospouses, minor children, and/or adult childrendisabled since childhood. These benefits arefinanced by Social Security taxes on workers’

earnings. Consequently, the system shiftsresources from households without eligible fami-ly members to those with eligible family mem-bers. Such explicit transfers are less likely to befeasible in individual accounts, as they wouldprobably be viewed more as a property-rightsbased system with benefits based on accountcontributions. Individual accounts might be con-ceptualized as personal assets rather than gov-ernment entitlements derived from tax revenue.Individual account issues relating to children,including disabled adult children, are discussedin Chapter Eight, while spousal rights underindividual accounts are discussed in more detailin Chapter Six.

Compulsory Annuities

As noted earlier in this chapter, the pros andcons of compulsory annuitization depend on theindividual account system’s overall goals andstructure. Compulsory annuities prevent retireesfrom either outliving their money or spendingdown their accounts and turning to SSI for

Box 3-3. Social Security for Married Retirees

Social Security benefits are paid only as monthly amounts that continue for life and that keep pace withinflation.

A spouse and dependent children are eligible for benefits in addition to the retiree’s benefit. The spousebenefit is equal to 50 percent of the retiree’s full benefit and is paid only to the extent that it exceeds thespouse’s benefit from his or her own work record. A 50-percent benefit is also paid to dependent children (ifunmarried and under age 18, or under age 19 and still in high school, or any age and disabled since child-hood). A maximum amount caps the total amount payable to the family when two or more dependents areeligible.

Social Security also pays a widowed spouse after the retiree dies. A widow can receive up to 100 percentof the deceased worker’s benefit, but only to the extent that it exceeds the widow’s own benefit as aretired worker. If the husband and wife had been receiving 100 percent and 50 percent of the husband’sbenefit, respectively, the surviving spouse’s 100 percent benefit would be a one-third reduction from theprior 150 percent benefit previously received by the couple.

Benefits are available on the same terms to wives and husbands and to widows and widowers. Becausemost husbands earn more than their wives, relatively few men receive benefits as spouses or survivingspouses based on their wives’ work records.


means-tested benefits. Also, by eliminatingadverse selection, compulsory annuities providehigher monthly payouts to the average retiree.Yet, the higher payouts are achieved by requir-ing the sick, dying, and other short-lived work-ers to pay more than the product is worth tothem. Indeed, annuities, by their very nature,shift resources from short-lived to long-livedworkers and making them compulsory wouldmaximize this transfer.

Predictably, some individuals—the terminally ill,for example—would want an exemption fromthe annuity requirement. Should exceptions begranted? A strong case could be made forexempting the dying because forcing them tobuy annuities might be viewed as harsh andunfair. Yet, to exempt the ill or dying raises newadministrative and equity problems (Mackenzie,2002). In terms of equity, the terminally ill arenot the only group who could make a case forexiting the annuity pool. Anyone facing largeunexpected expenditures—such as high medicalbills, or a sick family member—could argue forexemption. Even defining what constitutes a ter-minal illness could be a matter of dispute. Fromthe viewpoint of account administration, granti-ng exceptions on a case-by-case basis wouldrequire new rules and adjudicative processes(including appeal rights), which could signifi-cantly raise administrative costs. It could evenbe argued that to allow any exceptions wouldultimately unravel the mandate and forsake anyadvantage of universal annuities. Chapter Sevenfurther discusses possible exceptions for workerswho become disabled.

Compulsory Inflation Protection

Inflation poses a significant risk to retirees’ eco-nomic well-being. Yet, when given the option,few people who annuitize actually choose infla-tion protection—both in the United States andabroad (Liebman, 2002). Possible widespreadmisunderstandings about inflation risk supportthe argument for compulsory inflation indexing.Institutional arrangements for providing infla-tion-indexed annuities are discussed in ChapterFour.

Compulsory Joint-Life Annuities for MarriedRetirees

An individual account system modeled on SocialSecurity would require all married retirees tobuy spousal protection in the form of joint-and-survivor life annuities. If a married accountholder died before annuitizing, the accountwould go to the widowed spouse to be used forthe spouse’s annuity. Key design questions aboutjoint-life annuities involve the size of the sur-vivor benefit and whether spousal protectionwould be symmetric.

Symmetric Treatment of Husbands and Wives.Symmetric joint-life annuities provide identicalprotection for the primary annuitant (the hus-band, for example) and the annuity partner (thewife, for example). That is, a two-thirds sur-vivor annuity would pay two-thirds of the hus-band’s original amount to the wife if she becamewidowed and would also reduce the husband’sannuity to two-thirds of the original amount ifhe became widowed.

“Contingent” joint-life annuities are not sym-metric as they pay a reduced amount only if theannuity partner (the wife, for example) is wid-owed. If the primary annuitant is widowed, theoriginal amount continues. The ERISA mini-mum requirement for spousal protection is con-tingent. It requires that a widowed spousereceive at least 50 percent of the retiree’s pay-ment, but it does not require that the retiree’spayment be reduced if he is widowed. Whilecontingent joint-life annuities are the default inERISA pension policy, most joint-life annuitiesin the private individual annuity market providesymmetric treatment (Sondergeld, 1998).

We examine three levels of symmetric survivorannuities: a full (100 percent) survivor payment;a two-thirds payment, and a three-fourths payment.

Full Survivor Payment. These annuities continuethe original annuity amount when a spouse dies.A full survivor payment is one of three joint-lifeoptions offered to TIAA-CREF participants and


it is the most popular by far. About three-fourths of male annuitants chose joint-life annu-ities in 1994. Of those, nearly three in four ofthem (73 percent) took full survivor payments, 6percent chose the ERISA default (half paymentto the annuity partner), and 21 percent took atwo-thirds survivor payment (King, 1996). Thepopularity of the full benefit to survivors sug-gests that male annuitants wanted to avoid anyreduction in monthly income when either theyor their wives became widowed.

Two-thirds Survivor Payment. Many individualaccount plans that aim to resemble SocialSecurity survivor protection call for symmetricjoint-life annuities that pay two-thirds of theoriginal amount to the widowed spouse. For aone-earner couple, the current Social Securitybenefit for a widowed retiree or spouse is two-thirds of the prior benefit received by the couple.The two-thirds survivor annuity aims to repli-cate this feature.

Three-fourths Survivor Payment. Some propos-als would change Social Security’s defined bene-fits so that a widowed spouse would receive 75percent of the prior benefit received by the cou-ple. The goal is to reduce poverty among elderlywidows (Liebman, 2002).7 Annuities could alsobe designed to pay a 75 percent survivor benefit.

Option Three: Default Annuities withSpecial ProtectionsOption Three would turn the compulsory annu-ity of Option Two into a default option: theaccount would be automatically annuitized uponretirement, unless the account holder (and, inthe case of a married account holder, the spouse)chose differently. The default life annuity wouldbe inflation indexed and, in the case of marriedretirees, would provide a two-thirds symmetricsurvivor benefit.

The other payout options that retirees couldchoose under this system might include thoseelaborated in Option One. That is, the retireecould take a lump sum, receive phased with-drawals, or leave money in the account subject

to minimum balance and minimum distributionrules.

Although spousal protection would not bemandatory, the system could require spousalconsent for the account holder to take any pay-ment that provides less than the default two-thirds survivor benefit to the widowed spouse.ERISA imposes a spousal consent requirement.The couple could also choose greater spousalprotection through full survivor payments or 75percent survivor benefits.

Many studies show that default options have asignificant impact on behavior. When decisionsare difficult or confusing, consumers mightavoid them by procrastinating or by acceptingthe default (Loewenstein, 1999). A naturalexperiment illustrated this point. Buyers of autoinsurance in New Jersey and Pennsylvania weregiven a choice to pay lower insurance rates inreturn for a reduced right to sue for pain andsuffering. In Pennsylvania, the default was thefull right to sue, with a rebate for acceptingreduced rights. In New Jersey, the default was alimited right to sue with a surcharge to get thefull rights. In both states, about 75-80 percent ofdrivers took the default option (Johnson et al.,1993; Loewenstein, 1999).

Kahneman (1999) further observes the appeal ofdefaults, noting that:

“Staying with the default option is unde-manding; it is what we do when we are notthinking very hard…. The default is alsofavored by a desire to conform and to dowhat most other people do. Finally, in thecontext of choices that are offered by abenevolent agency, the default is likely to beperceived as an implicit recommendationabout what is best for most people.”

Defaults also influence behavior in retirementplans. For example, when enrollment in a401(k) plan is the default, more employees par-ticipate (Choi et al., 2002; Madrian and Shea,2001). Making spousal survivor benefits thedefault in private pension plans under ERISA


increased the number of wives who obtainedthat protection from their husband’s pension(Holden and Zick, 1998). Defaults also influ-ence how workers invest retirement funds.When Sweden introduced its new, supplemental,defined-contribution pension plan in 2000, one-third of Swedish workers did not make aninvestment decision, ending up in the defaultfund (Cronqvist and Thaler, 2004). Since then,92 percent of new enrollees did not make achoice and were added to the default fund(Krueger, 2004). The Swedish experience high-lights the importance of well-designed defaultsas many people end up in them (Sunden, 2004).

Option Four: Compulsory Annuities Upto a Specified Level This option would require the annuities pro-posed in Option Two only up to a certainthreshold. It poses a series of new issues aboutsetting and implementing the desired thresholdfor compulsory annuities.

Policymakers could look to compulsory partialannuitization to serve several objectives. Thelimited mandate could aim to produce an ade-quate baseline of retirement income. But whatlevel of income is deemed “adequate?” The offi-cial poverty threshold is one common measure;another is the replacement of a stated fraction ofprior earnings. A related approach might set theannuity threshold to approximate currentlypayable or currently scheduled Social Securitybenefit levels. Finally, those concerned aboutincreased dependence on means-tested benefitsmight set an annuity threshold at a level highenough to preclude eligibility for SSI.

Poverty Line Thresholds

Using the poverty threshold to set the level forcompulsory annuitization raises several issues.First, would the threshold take into account the retiree’s family living situation? Officialpoverty measures count the incomes of all fami-ly members who live together, with a higherpoverty threshold for larger families. Usingfamily income to set a compulsory annuitiza-

tion threshold for an individual could be compli-

cated because living arrangements and familymembers’ incomes can change significantly frommonth to month. A simpler approach could setthe annuity requirement at the poverty thresholdfor a one-person household, regardless of theretiree’s living situation.

How would the retiree’s other sources of incomebe counted toward meeting the mandated annu-ity threshold? Presumably, Social Securitydefined benefits would be considered. For exam-ple, if the poverty threshold were $800 a monthand the retiree had $750 in Social Security, he orshe would be required to annuitize only thatportion of the account needed to produce aninflation-indexed annuity of $50. Would othersources of monthly income – such as pensions,interest on savings accounts, or earnings fromwork – also count toward the annuity require-ment? Presumably, only inflation-indexed andhighly predictable income sources would begood candidates to count toward the mandatedannuity level, although counting only thesewould likely raise administrative costs.Otherwise, the once-in-a-lifetime compulsoryannuity purchase could be affected by tempo-rary changes in one’s income.

Replacement Rate Thresholds

A second approach to basic retirement incomeadequacy would set the compulsory annuitythreshold to achieve a target rate of earningsreplacement. This threshold is more similar tothe existing Social Security system, where thebenefit formula is structured to produce particu-lar replacement rates. If individual accountannuities were structured similarly, the replace-ment rate requirements would be lower at high-er levels of pre-retirement earnings. The annuityrequirement could be set with the aim of equal-ing present-law payable or scheduled benefitswhen combined with remaining Social Securitydefined benefits. Payable benefits are lower thanscheduled benefits, starting in about 2042 whenthe trust funds are projected to be exhausted.

´


Means-Tested Program Thresholds

A minimum annuity mandate could reduce eligi-bility for means-tested programs by requiringretirees to annuitize as much of their account asnecessary to produce monthly income that,together with Social Security, reaches the limitfor means-tested programs such as SupplementalSecurity Income (SSI). The SSI limit of countableretirement income for an elderly person livingalone in 2004 is $564 a month.

Some may find this partial annuitization optionappealing as a compromise between the compet-ing goals of income security and individualchoice. Some see an advantage in that high earn-ers would not be required to use their entireaccount balances to buy annuities. This wouldreduce the amount of annuitization by highearners and would permit lower annuity prices.It would reduce transfers from low earners (whoon average have lower life expectancy) to highearners (with higher life expectancy) that a uni-versal full annuity pool would entail.

Yet, the annuity threshold requirement wouldforce low-income account holders to buy annu-ities with their entire account balances, whilehigher earners would be allowed more choices.

This could create what might seem like a regula-tory class system. To the extent that compulsoryannuities might not be popular, this potentialdisadvantage for low-income individuals andtheir families could enhance the perceivedunfairness.

Design and ImplementationIssues

Regardless of payout option, policymakersdesigning the payout rules will face major issuesabout the institutional arrangements for provid-ing annuities and the role of government or pri-vate insurers in bearing the inflation, mortality,and investment risk inherent in annuities. Theseissues are discussed in Chapter Four. Otherissues include the implications of joint-life annu-ities for couples in different circumstances; thetiming of annuity purchase; rules about guaran-tees; and, how to ensure that retirees with deci-sions to make have enough information to makeinformed choices.

Joint-Life Annuities for Couples inDifferent CircumstancesPlans that require joint-life annuities for marriedindividuals are generally designed to resemble

Figure 3-7. Joint-Life Annuity* and Traditional Social Security for One-Earner Couple Age 65:Payments as a Percent of Amount for a Single Worker

Benefit as percent of amount forMarital status and situation a single worker

Traditional Social Security Joint-Life Annuity*

Single retiree – Bob 100 100Married retire – John 100 93John’s non-working wife spouse – Mary 50 0Total for John and Mary 150 93When John or Mary is widowed 100 62

* Symmetric life annuity that pays two-thirds of the original amount to whichever spouse lives longer. Annuity estimates are based onassumptions that: purchase of annuities is mandatory (reflecting total population life tables for individuals age 65 in 2005); annuities arepriced the same for men and women; the annual inflation rate is 3.0 percent; and the real annual interest rate is 3.0 percent.

Source: Author’s calculations based on present law and annuity examples provided by the U.S. Social Security Administration’s Office ofthe Chief Actuary, 2003


features of Social Security. How such plans actu-ally compare with traditional Social Securitywould be different for one-earner couples thanfor two-earner couples. Results may also differdepending on the ages of the spouses, the choic-es they make (when they have choices), the tim-ing of widowhood, and treatment of marriageafter retirement. Further issues about spousalrights are discussed in Chapter Six.

One-Earner Couples

In traditional Social Security, a one-earner cou-ple receives 150 percent of the benefit of a singleretiree (the spouse receives a separate checkequal to 50 percent of the worker’s benefit). Thewidowed spouse receives two-thirds of thatamount, or 100 percent of the single retiree’sbenefits (Figure 3-7). The cost of spousal bene-fits is borne by all workers paying into theSocial Security system.

With annuities, in contrast, the retiree takes acut in his or her monthly annuity to provide survivor benefits for a spouse. The size of thecut would depend on the respective ages of thehusband and wife. If John and Mary were bothage 65, a two-thirds survivor annuity wouldlower John’s initial benefit to about 93 percentof what a single annuitant would receive. Nosupplemental annuity is available for John’swife. When one of them dies, the survivorreceives two-thirds of the “93-percent annuity,”or about 62 percent as much as a single retireewould have from an account like John’s.

In brief, a symmetric joint and two-thirds sur-vivor annuity would ensure that a widowedspouse receives two-thirds of the amount paidbefore widowhood. But a one-earner couple’sincome from such an annuity is smaller than theincome a single person would receive from thesame size account for two reasons. First, joint-life annuities start out lower than single-lifeannuities because of the added cost of coveringtwo lives without subsidy. Second, only two-thirds of the reduced amount is paid to the sur-vivor after a spouse dies.

Dual-Earner Couples

In dual-earner couples, the husband and wifewould each have their own annuities. With com-pulsory survivor protection, each would berequired to buy joint-life annuities. When onespouse died, the survivor would receive bothsurvivor annuities. This is somewhat differentfrom traditional Social Security.8

Survivor annuities for dual-earner couples wouldgenerally be added together. As illustrated inFigure 3-8, if both John and Mary purchasedjoint-life annuities with their accounts of$20,000 and $5,000, respectively, their com-bined monthly annuities would be about $144while both are alive; John would receive $115and Mary would receive $29. When one of themdied, the widowed person would shift to receiv-ing two-thirds of each annuity, for a total of$96.

If members of dual-earner couples compare theirannuities to those of single persons with similarearnings, they sometimes have more and some-times have less. For example, in the dual-earnercouple with unequal earnings, John’s annuity of$115 is less than the $124 annuity that his sin-gle counterpart, Jack, receives. If John becamewidowed, his annuity would drop to $96, fallingfurther behind his counterpart Jack. But Mary,the low-earner in the couple, has survivor pro-tection not available to her single counterpart.Mary’s survivor annuity of $96 is more than shewould have if single.

In the dual-earner couple with equal earnings,both Bob and Sue take a modest cut in theirannuities to provide survivor protection. If wid-owed, either one would have higher combinedannuities ($96) than if they were single likeDiane, who has a single-life annuity of $77, buta smaller combined annuity than Richard,who—with the same earnings as the couple—has a single-life annuity of $155.

In brief, a great deal of variation exists in howannuities might compare to traditional benefitsfor dual-earner couples if symmetric joint and


two-thirds survivor annuities were required. Thehigher earner in the couple would often have asmaller annuity than his single counterpart. Thisdoes not occur with traditional Social Securitybecause survivor protection is subsidized. Yet,couples with nearly equal earnings could havehigher annuities as widowed spouses than theywould have had if single, because they can com-bine the survivor payments from their ownaccounts and their deceased spouses’ accounts.

If an individual account plan is a part of SocialSecurity, the ultimate impact on combinedincome from the annuity and traditional benefitswill depend on the size of traditional benefitsand the accounts, and how, if at all, they arecoordinated or offset against each other.

When Spouses Are Different Ages

How would compulsory survivor protectionwork when one spouse is considerably olderthan the other? For example, if the husband wasage 65 and his wife age 53, he would buy a

joint-life annuity. Twelve years later, when thewife reached age 65 and the husband age 77,she would buy a joint-life annuity. Then, whoev-er lived the longest would receive two-thirds ofboth annuities.

Using examples for 65-year-olds, Figure 3-9shows that an individual with $10,000 couldbuy a single-life annuity, indexed to rise by 3percent per year, of $62 a month. If he had a53-year-old wife, buying a joint and two-thirdssurvivor annuity would reduce the initial pay-ment to about $48 a month, a drop of 22 per-cent. When the wife later reached age 65 andused her $10,000 to buy a joint and two-thirdsannuity with her 77-year-old husband, herimmediate annuity would be about 10 percenthigher than the single-life annuity she couldbuy.9

The key point is that a young spouse reduces amarried retiree’s own annuity as well as the sur-vivor protection, while a much older spouse willincrease one’s own joint-life annuity. In brief,

Figure 3-8. Joint-Life*, Inflation-Indexed Annuities: Dual-Earner Couples Both Age 65

Individual Account balance Monthly payment Payment to survivor*

Dual-Earner Couple – Unequal Earnings

John $20,000 $115 (67%) $77Mary $5,000 $29 (67%) $19Total $144 (67%) $96

Dual-Earner Couple – Equal Earnings

Bob $12,500 $72 (67%) $48Sue $12,500 $72 (67%) $48Total $144 (67%) $96

Single Individuals

Jack $20,000 $124 $0Martha $5,000 $31 $0Richard $25,000 $155 $0Diane $12,500 $77 $0

* Symmetric life annuity that pays two-thirds of the original amount to whichever spouse lives the longer. Annuity estimates are basedon assumptions that: purchase of annuities is mandatory (reflecting total population life tables for individuals age 65 in 2005); annuitiesare priced the same for men and women; the annual inflation rate is 3.0 percent; and the real annual interest rate is 3.0 percent.



while joint and survivor annuities can bedesigned to resemble some features of SocialSecurity, annuities for couples of different ages,or in different circumstances, produce monthlyincomes quite different from what SocialSecurity provides in defined benefits. WhileSocial Security benefits are adjusted for the ageof the retiree when benefits are claimed, the ageof the retiree’s spouse does not affect the retiree’sbenefit amount.

When Spouses Make Different Choices toParticipate

In a voluntary account system, a husband andwife could make different choices about partici-pation. These different choices might underminethe goals of symmetric treatment and adequatespousal protection if joint-and-survivor annuitieswere compulsory for married couples. Forexample, if the wife—but not the husband—decided to join the individual account plan, shepresumably would be required to buy a joint-and-survivor annuity that provided for him. He,however, would not have an individual account

or an annuity to provide survivor protection forher. Should policymakers require that husbandsand wives make similar choices about participa-tion in the accounts, or at least require that thehigher earner participate to provide protectionfor a more dependent spouse?

In theory, rules could require that husbands andwives make the same choice and, if they cannotagree, provide a default rule to determine theirstatus. But new marriages and remarriageswould create mismatches between spousal participation. In the end, it appears that theintended outcomes of compulsory joint-lifeannuities for married retirees would be achievedonly if participation in the accounts was alsomandatory.

Timing of Annuity Purchase and Widowhood

Whether a married person died before or rightafter buying a joint-life annuity would impactthe widowed spouse’s financial security. Forexample, consider John and Mary who haveaccounts of $20,000 and $5,000, respectively, as

Figure 3-9. Effect of Spouse’s Age on Initial Joint-Life, Inflation-Indexed Annuity for $10,000Premium: Retiree with Spouse Age 53 or Age 77

Attribute Initial Monthly Annuity

Retiree only, age 65

Single life annuity $61.84Retiree age 65 with spouse age 53

Joint and two-thirds survivor annuity $48.54Size of survivor benefit 32.36

Joint and 100% survivor annuity 42.79Size of survivor benefit 42.79

Retiree age 65 with spouse age 77

Joint and two-thirds survivor annuity $69.12Size of survivor benefit 46.08

Joint and 100% survivor annuity 58.23Size of survivor benefit 58.23

* Symmetric joint-life annuities that pay the same amount to whichever spouse lives the longer. Annuity estimates are based on assump-tions that: purchase of annuities is mandatory (reflecting total population life tables for individuals age 65 in 2005); annuities are pricedthe same for men and women of the same age; the annual inflation rate is 3.0 percent; and the real annual interest rate is 3.0 percent.



shown in Figure 3-10. If they both purchasedjoint and two-thirds survivor annuities, theircombined income while both are alive would be$144. If either John or Mary died shortly there-after, the survivor’s monthly income would dropto two-thirds of the original amount, or $96.

Yet, if one died before they bought annuities, thewidowed person would have much higherincome. This would occur because the widowedspouse, for example Mary, would inherit John’saccount which, when combined with her ownaccount, would purchase a single life annuity of$155 a month. The same would be true if Johnwere widowed before they annuitized. He, too,would have a life annuity of $155. If they haddelayed buying annuities, the income for thewidow or widower would be more than 60 per-cent higher than the joint and two-thirds sur-vivor annuities from their two accounts. Thisdisparity in survivor protection appears to beunavoidable. A single life annuity purchased bya widow or widower with the proceeds of twoaccounts will always be larger than the sum ofthe survivor benefits from two joint-life annu-ities purchased with the same premium.

This discontinuity in survivor benefits does notoccur in traditional Social Security because sur-vivor benefits are not affected by whether one iswidowed just before or just after taking retire-ment benefits.

Annuities, Guarantees, and the Interestsof HeirsAnnuitization uses funds that would have beeninheritable and the purchase is irrevocable, soheirs have some interest in the account holder’sdecision to buy an annuity. The prospect thatthe annuitant will die shortly after buying anannuity prompts an interest in guarantee features.

Would Guarantees Be Offered or Allowed?

Guarantees provide that a life annuity will pay acertain amount even if the annuitant dies withina specified period after the purchase, but guar-antees come at a price: they lower the monthlyannuity that a given premium will buy. For a65-year-old individual, adding a 10-year certainfeature would lower the annuity by about 6 per-cent, while a refund of premium would lowerthe monthly amount by about 11 percent.10

Nonetheless, guarantee features are popular. OfTIAA-CREF participants who bought single-life

Figure 3-10. Timing of Annuity Purchase and Widowhood: Joint-Life Annuities or Delaying Annuity Purchase Beyond Widowhood

John Mary Total

John and Mary each buy joint and two-thirds, inflation-indexed life annuities

Account balance $20,000 $5,000 $25,000Joint and two-thirds life annuity $115 29 $144

Two-thirds payment to survivor $77 $19 $96John and Mary delay annuity purchase until after one is widowed

Account balance $20,000 $5,000 $25,000Single-life, inflation-indexed annuity $155

Symmetric joint-life annuities that pay two-thirds of the original amount to whichever spouse lives longer. Annuity estimates are based onassumptions that: purchase of annuities is mandatory (reflecting total population life tables for individuals age 65 in 2005); annuities arepriced the same for men and women of the same age; the annual inflation rate is 3.0 percent; and the real annual interest rate is 3.0percent.


24367 NASI TEXT 1/17/05 11:36 AM Page 66


annuities in 1994, two out of three chose annu-ities with guarantee periods of 10, 15, or 20years (King, 1996).

Some of the appeal of guarantees is psychologi-cal and emotional. Annuitants might want toavoid the serious disappointment for their heirsand survivors if they paid a large amount for anannuity and died shortly thereafter. The largeoutlay with nothing to show for it might feellike a big mistake and a very bad deal.

Many experts believe guarantees are not a wisepurchase. If the annuitant wishes to leave abequest to heirs, these experts suggest other lessexpensive and more predictable ways to do so.One could buy a smaller annuity and keep therest of the funds as liquid assets to spend orbequeath. This strategy would produce a pre-dictable bequest instead of a random amountthat would depend on when the annuitant died(Diamond, 2004). For example, instead of buy-ing a refund of premium annuity (which wouldlower the annuity by 11 percent), a 65-year-oldcould keep 11 percent of the premium to spendor bequeath and buy the same size annuity with89 percent of the original premium.

Given these mixed verdicts on annuity guaran-tees, would guarantees be allowed in an individ-ual account system with compulsory annuities?Aside from the expert criticisms mentionedabove, the main drawback of guarantees forcompulsory annuities is that they reintroducesome of the longevity risk and adverse selectionthat universal, required annuitization is sup-posed to eliminate. Yet, policymakers might findit difficult to ban guarantees because they easethe serious regret and financial losses for peoplewho end up on the short end of the annuitywager.

Some have suggested that it might be possible toban guarantees if retirees were required to annu-itize only a portion of their accounts (as in someversions of Option Four). A partial mandatemight reduce the desire for guarantees if liquidi-ty in case of early death is their main attraction.

But this policy would not address the psycholo-gy of the annuity tradeoff. Buyers might stillwant a guarantee against the risk of losing theentire purchase if they die soon after annuitizingonly part of their accounts.

Would Guarantees or Bequests Pay a LumpSum or Periodic Payments?

If guarantees are offered, the pros and cons ofdifferent forms of guarantee payments meritsome attention. At the election of the annuitant,guarantees can be designed to either pay a lumpsum or to make periodic payments to a namedbeneficiary. The periodic payments can be in theform of life annuities for the death beneficiariesor installment payouts. Standard life insurancesettlements offer policyholders a choice of theseoptions. If the policyholder did not specify, thebeneficiary can choose the payment form orchoose to leave the funds as an interest-bearingdeposit with the insurance company. Thesechoices for the form of guarantee payments orbequests could be built into an individualaccount program, or the plan could limit thechoices to only a few. A lump sum paymentwould pose a lesser administrative burden onthe annuity provider.

How Might Guarantees Ease the TensionBetween Annuities and Heirs’ Interests?

There is an inevitable tension between buying alife annuity and the interests of heirs who wouldget the money if it had not been spent on anannuity. Guarantee features or partial annuitiza-tion could ease some of this tension.

Figure 3-11 illustrates how partial annuitizationor guarantees could affect outcomes for a single,65-year-old retiree with a $20,000 account whohas named her nephew as her death beneficiary.If she dies before buying an annuity, the entire$20,000 account goes to her nephew but, if shebuys a single-life annuity with the entireaccount, she is guaranteed $124 a month for lifewhile her nephew will inherit nothing. To avoidcompletely wiping out his inheritance, she couldbuy a $15,000 annuity and reserve $5,000 for apotential bequest. Her monthly annuity is now


reduced to $93, but her nephew stands to inher-it $5,000 (plus any interest earned if she investsthe money).

If the system allowed guarantees, the retiringaunt could buy a refund of premium annuity,which would reduce her monthly income byabout 11 percent, to $110 a month. Upon herdeath, her nephew would receive the differencebetween the full $20,000 premium and the sumof benefits she received. If she lived only threemonths, for example, her $20,000 premiumwould have paid her $330, so her nephewwould inherit $19,670. Yet, if she were still aliveafter 16 years there would be nothing left forher nephew.

In brief, the purchase of an annuity is a perma-nent loss to heirs. A retiree who wishes tobequeath part of an account would have to pur-chase a smaller annuity. If she decides to leave aportion of her account for a beneficiary, she isaccepting lower income for herself. The tradeoffof partial annuitization may be appealing to her:it leaves funds available for her own unexpectedspending needs or as a bequest if she does notneed the balance.

By contrast, the retiree could choose to providefor her nephew by purchasing a guarantee fea-ture. In this case, her nephew’s inheritance isexpressly contingent upon how long she lives.The only thing certain about her “guarantee” isthat she will receive a lower income than shecould otherwise have afforded. In return for thissacrifice, her nephew could end up with a largebequest, a small bequest, or nothing at all.

The question about whether to buy guaranteesor partial annuities could also arise for marriedor widowed retirees who wish to leave a bequestto their adult children.

Timing of Annuity PurchaseConcerns about the interests of heirs can addcomplexity to the timing of the annuity pur-chase. From a strictly selfish perspective, namedbeneficiaries might prefer that their benefactorswould never buy an annuity so as to keep theaccount in a form that can be inherited. If askedfor advice, an heir’s self interest might encouragedelay. Retirees, too, might want to delay; theymight simply be hoping for a more favorableinterest rate. More importantly, delay mightreveal new insights about the wisdom of buyingan annuity at all. As noted earlier, a life annuitywould not be the preferred choice for one who

Figure 3-11. Full and Partial Annuities, Bequests, and Guarantees: Single Individual

Single Individual Aged 65 with a $20,000 Individual Account Monthly Payment toIn all cases, annuity purchased is single-life, 3% inflation indexed Annuity Beneficiary

#1 Dies before buying an annuity 0 $20,000#2 Buys $20,000 annuity with no guarantee feature $124 0#3 Buys $15,000 annuity with no guarantee feature, holds $5,000 as bequest

(a) Dies 3 months later $93 $5,000(b) Lives 20 more years and still holds $5,000 bequest $93 $5,000

#4 Buys $20,000 annuity with a refund of premium guarantee. (a) Dies 3 months later $110 $19,670(b) Lives at least 16 more years $110 0

Annuity estimates are based on assumptions that: purchase of annuities is mandatory (reflecting total population life tables for individualsage 65 in 2005); annuities are priced the same for men and women; the annual inflation rate is 3.0 percent; and the real annual interestrate is 3.0 percent.



has a life-threatening illness, nor would joint-lifeannuities be the best choice for a couple inwhich one partner is expected to die soon. Thetiming of annuitization is tied to market risk,longevity risk, and the interests of widowedspouses and other heirs.

How Would Delaying Annuity PurchaseAffect Potential Income?

All other things being equal, delaying the pur-chase of an annuity results in a smaller futureincome stream (Ameriks, 1999). Future changesin one’s own status (such as becoming widowed,divorced, or married) or changes in annuityterms (such as a rise or fall in interest rates)would also alter the income stream that couldbe purchased with a given sum of money.

As shown in Figure 3-12, a retiree hesitantabout buying an annuity could take systematicwithdrawals for several years. This arrangementseems attractive on its face: the retiree wouldhave “annuity-like” monthly income, wouldretain the liquidity of the remaining account bal-ance, and would retain the option to buy anannuity down the road.

The delay comes with a price, however. Thesteadily depleting account balance will buy anincreasingly smaller annuity as time goes by. Ifthe 65-year-old retiree shown in Figure 3-12took systematic withdrawals for ten years andthen purchased a standard annuity, her annuity

income would be 15 percent lower than whatshe could have received if she had made the pur-chase at age 65. If she waited 20 years to buy anannuity at age 85, it would be only 13 percentthe size of what was originally available to her.

Would Timing of Purchase Affect theAnnuity Amount?

The timing of annuity purchase is also impor-tant because the annuitant faces two financialrisks that could change the annuity terms overtime. Fluctuations in asset value or interest ratescan alter both the account balance available forannuitization and the income stream that can bebought with a given sum of money.

All account holders would face investment risk,or the possibility of short-term changes in thevalue of assets in the account. An account couldbe invested in equities, for example, and thestock market could decline just before theaccount holder is due to buy an annuity. In thiscase, the unfortunate timing of the marketslump could significantly reduce the accountholder’s retirement income for life. This invest-ment risk is always present, but it could bereduced if people were required to invest theiraccounts in bonds, money market funds, orother relatively safe products as they approachretirement. After annuitization, the risk is trans-ferred to the annuity provider (except under avariable annuity, which shifts the risk to the

Figure 3-12. Relative Size of Fixed, Single-Life Annuity by Age Purchased

Age Annuity is Purchased Monthly Income from Annuity

Age 65 $100Age 70 $95Age 75 $85Age 80 $62Age 85 $13

Assumptions: Account accumulation ends at age 65. If annuity purchase is delayed, the withdrawal each year is the size a fixed annuitywould be if purchased at age 65. Annuity is single-life standard annuity based on 6.5 percent interest rate and mortality based onAnnuity 2000 merged unisex tables with ages set back two years.

Source: Ameriks, 1999. “The Retirement Patterns and Annuity Decisions of a Cohort of TIAA-CREF Participants”


annuitant or spreads the investment riskbetween the insurer and the annuitant).

Fluctuations in interest rates will also affect thesize of the annuity purchased with any givenlump sum. For a given price, a lower interestrate will produce a smaller annuity. Gradualannuity purchases could reduce the interest raterisk. For example, a retiree could be given theoption of buying annuities gradually over a 10-year period between the ages of 62 and 71.During that time, the retiree would receiveincome from a small but growing annuity aswell as investment returns on the balance notyet annuitized. In the first year, for example, theretiree would have an annuity based on 10 per-cent of the account balance and would receiveinvestment returns on the other 90 percent.

Gradual annuitization provides only limited pro-tection against interest rate fluctuations.Gradual purchase may lessen the impact onretirement income from year-to-year interest ratechanges, because it spreads the annuitant’s inter-est rate risk over a longer period. Still, one whogradually bought annuities over a decade wheninterest rates averaged 3 percent would get farsmaller income than one who bought annuitiesover a decade of 8 percent average rates.

Gradual annuitization also has its downsides.Retirees would not know how much incometheir accounts would provide until almost all theannuities had been purchased. In addition,administrative costs would be higher.

Would Purchase of Annuities Coincide withClaiming Social Security?

If the accounts are considered an integral part ofSocial Security, there may be a case for requiringthat retirees buy annuities when they claimSocial Security benefits. Yet, there may be rea-sons why it is not in a retiree’s best interest tobuy an annuity then. For example, if joint-lifeannuities were required of married retirees, acouple with a terminally ill spouse would be bet-ter off to delay annuitizing even though they areready to claim Social Security. As noted earlier, a

single-life annuity purchased by a widow orwidower with the proceeds of two accounts willalways be larger than the sum of the survivorbenefits from two joint-life annuities from thesame accounts. Likewise, a single retiree withshort life expectancy may prefer to delay buyingan annuity even though he or she is ready toclaim Social Security benefits. If annuities weremandatory, or joint-life annuities were requiredfor married retirees, some flexibility in timingmight reduce the extent to which such policiesare perceived as unfair because they requireretirees to buy an undesirable product.

Providing for Informed ChoiceIf retirees in an individual account system are tohave any choice about whether, when, or how toannuitize, who will advise them and answertheir questions? And to what degree would theeducator or adviser be held responsible for theconsequences if the advice turns out badly? Asbackground for answering these questions, it isuseful to consider how the federal employees’Thrift Savings Plan and Social Security providefor informed choice among participants.

Informed Choice in the Thrift Savings Plan:Employer Assistance

Retirees in the TSP face a broad range of choicesabout the form and timing of payouts.Responsibility for ensuring that retirees under-stand their choices rests largely with theiremployers. Federal agencies’ personnel officesprovide retirement planning seminars to helpemployees understand their retirement benefits,including health coverage, life insurance,defined-benefit pensions, and payout optionsunder the TSP. Participants can also use the TSPwebsite, which provides information and annu-ity calculators to help retirees choose payoutoptions.

In the private sector, many large employers alsoprovide help with retirement planning, alongwith information about retiree health, pension,and life insurance benefits. Some employers alsoprovide information about Social Security bene-fits to their retirees, but they have been reluctant


to provide retirement investment advice for fearof fiduciary liability under ERISA.

An individual account system could not relysolely on employers to educate retirees. Further,a universal individual account system wouldinclude more low-income workers who may beless educated. Such workers might need moreextensive help understanding individual accountplans than those workers with employer-spon-sored pension plans. An individual account system would also include more part-time orcontingent workers and casual laborers whomight not have a regular employer to help themwith their retirement choices. Not all employerscould provide the necessary services; small companies, or companies with high employeeturnover, lack capacity to provide employee ben-efits and do not offer retirement advice.

If federal policy introduced individual accountson a universal or widespread basis, some largeemployers might add information about payoutoptions to their retirement planning seminars.But it is not reasonable to expect that smalleremployers would be equipped to educate andadvise employees about a new system of federal-ly sponsored accounts.

Limited Choice in Social Security

Social Security’s answer to the problem ofinformed choice is to give retirees almost nochoice. All beneficiaries receive monthly infla-tion-indexed benefits that last for life. Retireescan choose only whether to take their entitledbenefits and when to take them.

The first decision—whether or not to take SocialSecurity benefits—is straightforward. Noexchange is required to get benefits or to pro-vide family protection. With no downside toaccepting, very few people actually decline benefits.11

The second decision—when to take benefits—involves some financial tradeoffs, but SocialSecurity policies on retirement age, benefit eligi-

bility, and returning to work after retirementavoid sharp cliffs in a retiree’s income.

Under current policy, a retiree can take benefitsany time from age 62 onward. Benefits arereduced for each month they are received beforethe statutory full-benefit age, which is set to risefrom 65 to 67 over the next two decades.Monthly benefits are increased for each monththey are not received between that age and age70. After age 70, there is no financial advantageto delaying the benefit claim.

Early retirees who decide to go back to workcan stop receiving reduced benefits. When theyreclaim their retirement benefits, the benefitswill be adjusted to reflect an early retirementreduction only for the months that early benefitswere actually received. (An early retiree can evenundo any early retirement reduction by payingback the early benefits received.)

The Social Security Administration (SSA) keepsits administrative costs low (less than 1 percentof benefit outlays), in part because the law offersalmost no benefit choices to retirees. If the SSAwere called upon to administer an individualaccount program, any rules that offered morechoice would increase administrative burdensand costs.

Recap of Choices

Policymakers designing individual account pay-out rules will confront an inevitable tensionbetween offering choice and guaranteeing ade-quate retirement income for life. Hard-and-fastrules, mandates, or defaults might ensure thatthe system meets certain high-priority goals, butthey might also create pressure for exceptions.The following list summarizes the issues thatarise as individuals approach retirement. Howmany of these issues would actually be left up tothe retiree’s choice, and the kind of options thatwould be available, would depend upon the ulti-mate design of the payout rules:

(a) Whether to buy an annuity at all;


(b) How much of one’s account to spend on anannuity;

(c) Whether the annuity would be indexed forinflation;

(d) When to buy an annuity;

(d) Whether to buy a guarantee feature;

(e) If a guarantee is desired, whether it is peri-od-certain (and for how long) or a refund ofpremium, and whether it would go to anamed beneficiary or to the estate;

(f) If joint-life annuities were optional forunmarried individuals, whether to buy oneand with whom;

(g) If joint-life annuities were offered orrequired for married individuals, whether tobuy symmetric or contingent products;

(h) If joint-life annuities were offered orrequired, what size survivor protection toprovide;

(i) Whether to buy a guarantee in addition to ajoint-life annuity;

(j) Whether to coordinate claiming SocialSecurity with the purchase of an annuity.

Summary

Life annuities are financial products that canhelp solve the problem of making a sum ofmoney last for the rest of one’s life. Yet, fewpeople choose to buy life annuities and insurersdo not seem to actively market life annuities.Investment and tax-deferral products that arecalled annuities are more actively marketed, butdo not have the positive features of life annu-ities. Genuine life annuities provide varyingdegrees of protection for annuitants and theirfamilies and each of the various features involvesome tradeoffs. Policymakers might want todecide what baseline protections would berequired depending on the goal of the individual

account system and its relationship to SocialSecurity.

If provision of guaranteed lifelong income werethe goal of annuitization, then an inflation-indexed annuity would be important to ensurethat the income would maintain its purchasingpower even at advanced age. If annuities are notinflation-indexed, retirees could face substantialloss of purchasing power, even over relativelyshort time periods. Institutional arrangementsfor providing inflation-indexed annuities are dis-cussed in Chapter Four.

Individual account annuities could be compulso-ry, partially compulsory, or optional with a widerange of features. In designing the annuitizationrules, policymakers will face tradeoffs.Compulsory annuities would result in higheraverage payouts for any given premium but thehigher payouts would be achieved by requiringthe sick, dying, or other short-lived workers tobuy a product that might be disadvantageous.

While life annuities can make lump sum retire-ment savings resemble defined benefits, lifeannuities and defined benefits are distinctly dif-ferent. A defined-benefit plan computes aretiree’s benefit based on his or her past workrecord, age, family status, and a variety of otherfactors. When the retiree qualifies, the specifiedpayments are automatic; there is no purchase. Incontrast, an annuity arises from a financialtransaction between a seller and a buyer, withthe buyer making an immediate sacrifice inexchange for the life annuity contract.Mandating life annuities means requiring peopleto buy a product they may not want.

Symmetric joint and two-thirds survivor annu-ities would ensure that widowed spouses receivetwo-thirds of the amount paid before widow-hood. A one-earner couple’s monthly incomefrom such an annuity is smaller than the incomea single person would receive from the same sizeaccount for two reasons. First, joint-life annu-ities start out lower than single-life annuitiesbecause of the added cost of covering two lives


without subsidy. Second, only two-thirds of thereduced amount is paid to the survivor after thespouse dies.

A great deal of variation remains in how annu-ities might compare to traditional benefits fordual-earner couples if symmetric joint and two-thirds survivor annuities were required. Thehigher earner in the couple would often have asmaller annuity than his single counterpart. Thisdoes not occur with traditional Social Securitybecause survivor protection is subsidized. Yet,couples with more nearly equal earnings couldhave higher annuities as widowed spouses thanthey would have had if they were single becausethey can combine the survivor payments fromtheir own accounts and their deceased spouses’accounts. If an individual account plan is a partof Social Security, the ultimate impact on com-bined income from the annuity and traditionalbenefits will depend on the size of traditionalbenefits, the accounts, and how, if at all, theyare coordinated or offset against each other.

Annuitization also involves tradeoffs betweenthe interests of retirees and the interests of theirheirs. An annuity purchase marks the end of anheir’s possibility of inheriting the funds.Guarantee features could provide the possibilitythat heirs would receive some portion of theprice paid if the annuitant dies early, but theguarantee lowers the annuitant’s lifetimeincome. Partial annuitization—setting aside partof one’s account for bequest purposes—couldalso help accommodate the interests of heirs.

Again, however, this accommodation comes inexchange for a smaller monthly payment for theretiree.

The timing of annuity purchase can make animportant difference in the protection it pro-vides. For example, whether a married persondies right before buying a joint-life annuity, orjust after buying such an annuity, would make asignificant difference in the widowed spouse’ssurvivor benefit. This disparity occurs becausethe sum of the survivor benefits from bothspouses’ joint-life annuities will always be lessthan the single-life annuity the widow could buywith the combined balances from both accounts.These features could affect policies about thetiming of mandated annuitization.

In general, delaying the purchase of an annuityallows time to gather more information to aidthe annuity decision. Changes in marital status,life expectancy, or annuity terms (such as inter-est rates) will alter the payout for a given premi-um. However, taking phased withdrawals or anyother consumption of funds during the delayperiod will generally reduce the size of the annu-ity one could buy later.

The following chapters provide more detail onthe many individual issues that are relevant tothe decisions about payout rules. As this chapterhas made clear, the goal of providing retirementincome security requires acknowledging theinterests of individuals beyond the particularretiree.

Chapter Three Endnotes

1 Cost-of-living adjustments to Social Security ben-efits are based on the increase in the ConsumerPrice Index for Urban Wage Earners and ClericalWorkers (CPI-W), referred to here simply asinflation.

2 Estimates of equivalence scales in the economicsliterature vary. See Citro and Michael (1995), pp.166-182.

3 This chapter focuses on life annuities – that is,products that guarantee payments for the life ofthe annuitant. Other financial products are called“annuities,” but are not life annuities. Theseproducts are also discussed in Chapter Four.“Deferred” annuities, sometimes called “vari-able” annuities, are tax-favored investment prod-ucts that do not guarantee payments for life,although they can be converted to life annuities.These products are essentially mutual fundswrapped in a life insurance contract. While thesedeferred annuity products offer participants anoption to convert the balance into a life annuityat retirement, there is no requirement that theybe converted and only a small fraction of con-tracts are converted in this way.

4 Civilian pensions are paid from the FederalEmployees’ Retirement System (FERS), whichbegan in 1986, and the older Civil ServiceRetirement System (CSRS). Employees in FERSare covered by Social Security and a supplemen-tal pension and the TSP. CSRS employees are notcovered by Social Security, but have a pensionthat is designed to fill the role of both SocialSecurity and a supplemental pension.

5 This discussion assumes that the individualaccount would be a countable asset and annuityincome would be countable income for purposesof the SSI and Medicaid means tests. Whetherthe accounts are or should be are important poli-cy questions that are discussed in Chapter Five.

6 Married TSP retirees in the Federal EmployeesRetirement System can withdraw or transfer theiraccount funds only if the spouse consents; mar-

ried Civil Service Retirement System participantsmay withdraw funds after notice has been givento the spouse.

7 Many such proposals would cap the 75 percentsurvivor benefit for dual-earner couples so as totarget the benefit increases only to low- andmoderate-income couples. Some versions of thisproposal would also lower the spouse benefitfrom 50 to 33 percent of the retiree’s benefit.This achieves a 75 percent relationship betweenthe retiree or widow’s benefit and the couple’sbenefit (100/133), by lowering the couple’s bene-fit rather than by increasing survivor protectionfor those couples.

8 Under the Social Security dual-entitlement rules,a widowed spouse receives an amount equal tothe larger of his or her own benefit or up to 100percent of the deceased partner’s benefit. Thesurvivor does not receive both benefits in full.

9 The joint-life annuity, in this case, is larger thanthe single life annuity because the joint-life annu-ity pays less when the primary annuitant is wid-owed, which is highly probable when the spouseis 12 years older. Illustrative annuity amounts arefor those purchased in 2005. Annuities pur-chased in future years are likely to show similarrelationships among age groups, but will be dif-ferent dollar amounts.

10 As previously noted, a 10-year certain annuityguarantees that the regular payments will contin-ue for at least 10 years; a refund of premiumannuity pays a lump sum equal to the amount bywhich total payments fall short of the initial premium.

11 A rarely used provision of the Social SecurityAct, section 201(i), authorizes the Social Securitytrust funds to accept gifts and bequests.Individuals can decline benefits and designatethem as unconditional gifts to the trust funds, ifthey wish. In 1980, 59 gifts were received. Mostwere less than $100, while one was over $10,000(U.S. SSA, 1982).


4Chapter

Institutional Arrangements forProviding Annuities

75

A life annuity is an insurance product thatpromises payments for as long as the annuitantlives. Chapter Three explored how policymakersmight set rules for workers to withdraw fundsfrom their individual account at retirement. Onepolicy option would grant retirees broad latitudeto withdraw funds as they wish, as is commonin voluntary, supplemental retirement savingsplans in the United States. To date, when retireesare given these choices, few buy life annuities.Other policy options would place much moreemphasis on the purchase of life annuities atretirement. Proposals that view the proceedsfrom the accounts as an integral part of SocialSecurity often require, or strongly encourage,the purchase of a life annuity that is indexed forinflation and that automatically provides incomefor a spouse widowed after retirement.

This chapter considers the institutional arrange-ments for providing life annuities. It begins withbackground on the U.S. annuity market andhow it is regulated by the states. Both the exist-ing private market (model 1) and a modified pri-vate market that reflects the role of the federal

government in designing annuity options forretirees in the federal employees’ Thrift SavingsPlan (model 2) are described. Either of theseexisting models might be compatible with a sys-tem of voluntary retirement accounts on top ofSocial Security that offers retirees wide choiceabout payouts, but neither is likely to be appropriate for individual accounts that aim to replace part of Social Security retirement benefits.

This chapter also examines new arrangementsfor providing inflation-indexed life annuities –products that are rarely found in the U.S. mar-ket today—and describes hybrid models for pro-viding inflation-indexed annuities on a broadscale, assuming that such annuities are mandat-ed or strongly encouraged in a new system.Model 3 emphasizes a private sector role, withthe federal government helping insurers to hedgeinflation risk. In model 4, the federal govern-ment or a quasi-governmental administrativeauthority takes on administrative tasks, butleaves the management of mortality and invest-ment risk with the private sector. In model 5, the


federal government serves as the annuityprovider and contracts out investment manage-ment activities to the private sector.

U.S. Market in Life Annuities

Life annuities are insurance products thatrespond to two contrasting features of humanmortality—while the length of life of any indi-vidual is quite uncertain, the distribution of lifespans of a large group of persons can be predict-ed with a much higher degree of accuracy. Froman insurer’s perspective, life annuity contractscomplement life insurance policies. Both prod-ucts insure the uncertain duration of individuallives, but in opposite directions. Life insuranceprotects the insured against the financial conse-quences of dying early (that is, dying before theinsured’s financial goals for his or her survivorshave been met), while a life annuity protectsagainst living long (that is, against the risk ofoutliving one’s savings).

In the United States, life annuities are providedby life insurance companies and are regulated bythe states. Key aspects of the life annuity marketinvolve the assumption of mortality risk andinvestment risk, the design of annuity products,and issues in pricing and marketing annuities.As regulators, the states are responsible formonitoring the financial soundness of annuityproviders and for providing a degree of protec-tion to customers in the event of insurance com-pany failure.

Mortality and Investment RiskOperating a life annuity pool in the current U.S.market poses at least two kinds of risks to theannuity provider – mortality risk and investmentrisk.1 Mortality risk refers to the danger to theinsurer that the life spans of persons in theannuity pool will exceed the mortality assump-tions on which the annuity contracts werepriced. In this case, the insurer will have tomake payments longer than anticipated, whichwill erode company profits or cause outrightlosses. Insurers price their life annuity productsbased on the expected mortality for the insured

group, projected investment returns, plus a mar-gin for administrative costs, marketing costs,and profit.

Investment risk inheres in the process of invest-ing the assets needed to support the promisedannuity payments. In the case of a life annuity,the insurer usually takes in a large sum ofmoney up front, which it will invest to earnreturns until the annuity obligation has beenfully paid. Because life insurers invest largely incorporate bonds and other fixed-income securi-ties, interest rates have an important effect onannuity pricing. Higher interest rates lead tomore investment income, which allows theinsurer to charge less for a given level of annuitypayment.

Annuity ProductsMany products are called “annuities” but arenot life annuities. It is important to distinguishlife annuities from deferred annuities and termannuities (see Box 4-1).

Only life insurance companies provide life annu-ities, which are contractual obligations requiringthe insurer to make payments to the annuitantfor the rest of the annuitant’s life. The insurerassumes both mortality and investment risk.

Deferred annuities are tax-favored investmentproducts that do not guarantee payments for thelife of the annuitant. The account holder has theoption to use the funds in a deferred annuityaccount to buy a life annuity, but relatively fewpeople do so. The product is used mainly as amechanism for deferring taxes on fund accumu-lations. Fixed term annuities are also differentfrom life annuities. A term annuity is a contractto pay the annuitant a specified amount ofmoney for a specified period of time – such asfive or ten years. The provider of a term annuitybears investment risk, but does not take on mor-tality risk.

Life annuities can be designed and priced to pro-vide a fixed nominal payment over time, togrow by a specified percentage each year, or to

Chapter Four: Institutional Arrangements for Providing Annuities 77

Box 4-1. Types of Annuities

Different ProductsLife annuities are issued by insurance companies and are a contractual obligation to make payments for thelife of the annuitant. Typically, one buys a life annuity by paying a lump sum or “single premium” to theinsurance company.

Deferred annuities are tax-favored investment products that do not provide payments for life, but they can beconverted to life annuities. The account holder has the option to use the proceeds to buy a life annuity, butrelatively few do. The product is used mainly as a mechanism for tax deferral during fund accumulations.

Fixed term annuities are contracts that promise specified payments for a given term, say five or ten years.The annuity provider bears no mortality risk.

How Do Life Annuity Payments Change Over Time?A fixed life annuity pays a flat dollar amount, usually monthly, for the life of the annuitant.

Rising life annuities pay amounts that rise at a prescribed rate, say 3 percent per year, for the life of theannuitant.

Inflation-indexed life annuities pay monthly amounts that are adjusted each year to keep pace with the con-sumer price index.

Variable life annuities pay benefits that vary from year to year depending on investment returns. Paymentscan go down as well as up. The annuitant bears all or part of the investment risk.

Participating variable life annuities are variable life annuities in which the risk of changes in life expectancy isshared between annuitants and the annuity provider. TIAA-CREF provides participating variable life annuities.

How Many Lives Are Covered? What Does the Survivor Get?Single life annuities make payments only for the life of the individual annuitant.

Joint-life annuities make payments for the life of the primary annuitant and a secondary annuitant (typicallythe primary annuitant’s spouse).

Symmetric joint-life annuities pay the same amount to a widowed primary annuitant as would be paid to awidowed secondary annuitant. The payment to the longer-lived person could be 100 percent, 75 percent,67 percent, or any other fraction of the amount paid while both were alive.

Contingent joint-life annuities pay a lower amount to a widowed secondary annuitant than to a widowedprimary annuitant. The primary annuitant’s payment is not reduced if he or she is widowed. If the secondaryannuitant is widowed, the payment could be 75 percent, 67 percent, 50 percent or any other fraction ofthe amount previously paid to the primary annuitant.

Guarantee FeaturesA ten-year certain annuity will make payments for ten years, even if the annuitant dies within ten years.Period-certain annuities can guarantee five, ten, twenty, or other durations of payments.

A refund of premium annuity guarantees to pay until the sum of payments equals the nominal purchaseprice of the annuity.


be inflation-indexed. Life annuities can also bedesigned to shift part of the investment risk andreturns or the mortality risk to the annuitants.For example, variable life annuities guaranteepayments for life, but have the annuitant sharein all or part of the investment risk.Participating variable life annuities also guaran-tee payments for life, but have annuitants sharein both investment risk and changes in lifeexpectancy that occur after the annuity is purchased.

Size of the Life Annuity Market Life annuities are a relatively small share of thetotal business of life insurance companies. Oflife insurance companies’ $300 billion in newproduct sales annually, life annuities representabout 5 percent, or $15 billion, of which $5 bil-lion is annual direct sales of life annuities andabout $10 billion represents the conversion ofdeferred annuities to life annuities (LIMRAInternational, 2004). Other products, includinglife insurance and deferred annuities, constitute95 percent of the volume of insurance companybusiness.

Some experts in the insurance industry see lifeannuities as an area for growth, particularly asprivate pensions and retirement plans shift frommonthly payments to lump-sum payouts.Obstacles to growth in this market appear to beconsumers’ weak demand for life annuities, asdiscussed in Chapter Three, and financial advi-sors’ limited interest in marketing them. Fromthe advisor’s perspective, life annuities have twodrawbacks compared to deferred annuities: lifeannuities generally pay smaller commissions(typically 4 percent) than do deferred annuities(typically 6 percent). Perhaps more important,deferred annuities hold the prospect for futuretransactions and commissions, while the pur-chase of a life annuity ends the advisor’s oppor-tunity to generate future business from the fundsbecause the money is turned over to an insur-ance company in exchange for the annuity con-tract. It remains an open question whether andhow consumer demand for, and the marketingof, life annuities will change as future retirees

receive less of their retirement resources in theform of monthly income. New individualaccounts that required, or strongly encouraged,the purchase of annuities would change theexisting market.

Pricing and Marketing AnnuitiesInsurance companies have considerable latitudein the design of life annuities, as suggested bythe many features described in Box 4-1. Joint-life annuities, which will pay for the life of theprimary annuitant and a secondary annuitant(typically a spouse), offer various choices abouthow much will be paid to the longer-lived annu-itant. Guarantee features ensure continued pay-ments to the annuitant’s heirs if the annuitantdies shortly after buying an annuity. As dis-cussed in Chapter Three, while these guaranteefeatures lower the size of the monthly annuity,they appear to be quite popular among annuitybuyers.

In pricing life annuities, insurance companiesseek characteristics that are good predictors oftheir customers’ future life spans. Insurers puttogether groups of purchasers with similar mor-tality risks to estimate the necessary price. Themost obvious characteristics are age, health sta-tus, and sex. While differentiation by age hasnot been controversial, policy views differ withregard to differentiation by sex or health status.

Price Differences between Men and Women

A key policy issue with regard to individualaccount annuities is the extent to which annuityproviders will be allowed to differentiate pricesbetween men and women. In the individualannuity market, insurers charge different pricesbecause women live longer than men, on aver-age. For example, a 65-year-old who wishes tobuy an annuity with $10,000 is offered $65 amonth, if male, or $62 a month, if female(www.annuityshopper.com). While this differen-tiation is common in the individual annuity mar-ket, federal policy bans sex discrimination ingroup annuities or pensions that are provided aspart of an employment relationship. The U.S.Supreme Court ruled in 1978 that Title VII of


the Civil Rights Act of 1964, as amended, for-bids this differentiation in employee benefits,even if it is justified on actuarial grounds (Cityof Los Angeles Department of Water and Powerv. Manhart, 1978).

If individual accounts require unisex pricing,men might be discouraged from buying annu-ities. From an insurer’s perspective, men wouldbe more profitable customers, on average. Yet,to the extent that individual accounts require orencourage the purchase of joint-life annuities formarried individuals, the sex differences diminishbecause mortality assumptions for husbands andwives are averaged in pricing joint-life annuities(Brown, 2002). Still, a requirement for unisexpricing would lead insurers in the single-lifeannuity market to favor men over women,unless other mechanisms were in place to avoidmarket segmentation (Box 4-2).

Price Differences between Healthy andUnhealthy People

In the U.S. annuity market, price distinctionsbetween healthy and unhealthy annuity buyershave not yet developed. Relatively few peoplebuy life annuities and those who do buy themtend to have above average life expectancy. Butif the purchase of annuities were to becomemuch more widespread, or even mandatory, thequestion of explicit price distinctions betweensick and healthy buyers is likely to becomeimportant. In the United Kingdom, where pur-chase of annuities is required by age 75, a mar-ket for “impaired life” annuities has emerged. In2002, the impaired life market accounted for 9percent of total individual annuity premiums inthe UK. This was roughly a 50 percent increasein volume of impaired life annuity premiumsover the prior year. Industry experts in the UKestimate that the payout rate on these annuitiesis about 5 percent to 20 percent higher thanordinary annuities (Watson Wyatt, 2003).

If a U.S. system of individual accounts called formandatory annuities, access to impaired lifeannuities is likely to become important to indi-viduals with short life expectancies. Impaired life

annuities are discussed briefly in Chapter Sevenwith regard to payouts for disabled-worker ben-eficiaries when they reach retirement age. Adrawback of impaired life annuities is that theyincrease the cost of annuities for healthy retirees.

Other Factors Affecting Annuity Pricing

Annuities purchased from different vendors maybe priced differently because insurers may havedifferent cost structures and profit margins. Onestudy found a difference of about 20 percent inthe annuity payouts between the ten highest andten lowest payout companies in the UnitedStates in 1995 (Mitchell et al., 1999). Estimatesof overhead charges associated with annuitiesrequire a number of assumptions and are diffi-cult to calculate. A study of the U.S. markets inthe late 1990s estimated average overheadcharges to be in the range of 5 percent to 15percent of costs for annuity purchasers andfound roughly comparable overhead charges instudies of foreign annuity markets (Brown et al.,2002).

In addition to differences in administrative costs,annuity providers may have other reasons tooffer different prices to different customers. Forexample, companies may be willing to offer abetter deal on annuity prices to customers withwhom they have (or hope to have) businessinvolving other financial products and services.Wealthy account holders would be attractiveannuity customers on this score.

Companies may also price life annuities differ-ently because they have different views of futuremortality rates or because they follow differentstrategies about how aggressively they pursuenew business.

State Regulation of Life InsurersThe life insurance and annuity business has longbeen a sphere of intense governmental regula-tion. A striking feature of insurance regulationin the United States is that Congress ceded thisregulatory authority to the states under terms ofthe McCarran-Ferguson Act of 1945. Whilethere is a strong component of federal regulation


of the banking, securities, and defined-benefit

pension industries in the United States, regula-

tion of the insurance industry remains under the

jurisdiction of the 50 states. Important areas for

regulation include the design and pricing of

annuities, standards for determining insurers’

financial backing for annuities, and provisions

for guaranteeing payments in case of insurancecompany failure.

Licensing, Approval of Products, and Pricing

States have licensing rules for insurance compa-nies and agents that do business within theirborders. States typically also require companiesto submit new products for approval. In the

Box 4-2. Adverse Selection, Uniform Pricing, and Selective Marketing

The cost of providing insurance varies based on the risk characteristics of the persons being insured.Because it is cheaper to provide life annuities to people with short life expectancy, insurance companieswill tend to charge them less than they charge individuals with longer life expectancy. In a voluntaryannuity market, if a company priced its annuities based on average risks and charged everyone the sameprice, people with longer life expectancy would be more likely to buy the annuities, while people withshort life expectancy would not. This adverse selection would drive up the cost to the insurer and lead thecompany to raise its prices. The higher prices would further discourage short-lived people from buyingannuities.

Adverse selection occurs when the purchase of insurance is voluntary and sellers do not charge differentprices for buyers who pose different risks. If policymakers wanted uniform pricing of annuities for peopleof the same age (regardless of sex, health status, or other risk factors), then the simplest way to avoidadverse selection would be to remove participant choice. Proposals that mandate the purchase of lifeannuities minimize adverse selection, as long as all sellers use uniform pricing. If a dual market were to exist (i.e., where buyers can choose between government-provided annuities with uniform pricing or privately provided annuities differentially priced based on risk), then adverse selection would againoccur.

Even when annuity purchase is required and all annuities use uniform pricing, adverse selection couldoccur if buyers have a choice among types of annuity products. For example, if annuities were mandatory,but retirees could choose whether or not to buy a guaranty that the full purchase price would be paidback to the annuitant’s heirs in the event of an early death, then retirees with short life expectancieswould prefer this feature. This adverse selection in favor of guaranty features among short-lived annui-tants would introduce some increase in the average price of annuities for everyone.

Uniform pricing in the presence of differential risk can lead to selective marketing. If multiple annuity sell-ers are competing for the same pool of buyers, sellers will have an incentive to seek out participants withshorter life expectancies. For example, if policymakers required uniform prices for men and women, theninsurers would prefer to sell annuities to men because they have shorter life expectancy. Regulations couldbe put in place to discourage explicit targeting, but sellers might still find indirect ways to attract malecustomers, such as by advertising in men’s magazines, male-oriented television sports programs, or com-mercials geared to appeal to men. It is difficult for regulators to stop selective marketing without directgovernment oversight of marketing activities.


individual life annuity market, insurance compa-nies are generally permitted to set prices basedon their own underwriting criteria.

Financial Backing for Annuities

Regulations on the financial backing of annu-ities include: (a) limitations on investment inrisky assets, such as common stocks, (b) reserverequirements, and (c) minimum capital stan-dards. Some states limit the extent of insurancecompanies’ investment in stocks explicitly, whileothers do so by requiring additional assets (orcapital requirements) when funds are invested instocks, effectively limiting stock investment.

For every life insurance or annuity contract onits books, a life insurance company must holdassets backing their reserves. Required reservesfor life annuities are the insurer’s legal liabilities,which are calculated according to formulas setby states using specified mortality tables andinterest rates. Each state’s laws and regulationsspecify the minimum amount of reserves for various products. Most state laws closely followthe “standard valuation law” developed by the National Association of InsuranceCommissioners, the national organization ofstate insurance regulators.

Capital requirements come in two forms. First,an insurance company must have a certain mini-mum level of capital to be licensed in a givenstate, and this requirement varies from state tostate. Second, additional capital requirementsdepend on each company’s investments andother risks. If a company invests in Treasurybonds, it does not need to hold additional capi-tal to guard against default risk (or most otherrisks), but it pays a price in terms of lowerreturns. If a company chooses to invest instocks, then state regulators require that addi-tional capital be held as a form of insuranceagainst the various risks associated with invest-ing in stocks. These capital requirements are not“reserves” in that they are not directly related tospecific contractual obligations.

State Guaranty Funds

The 50 states and the District of Columbia haveset up guaranty funds to ensure payment ofannuities and other life insurance products inthe event of an insurance company failure.Unlike federal insurance programs – such as theFederal Deposit Insurance Corporation, whichinsures bank deposits, or the Pension BenefitGuaranty Corporation, which insures defined-benefit pensions – state guaranty funds are notpre-funded. Instead, funds must be found tocover the cost of an insurance company failureafter it occurs, by making assessments on (thatis, taxing) other companies doing business in thestate. In general, these guaranty funds provideinsurance coverage for annuities up to a netpresent value of $100,000.2 To the extent thatan annuitant has a policy or policies above thecoverage limit, the uninsured portion would rep-resent a claim on the receivership of the failedinsurance company and, in all likelihood, wouldnot be paid in full. State laws limit the amountof insurance company assessments that can bemade per year, based on total premium volume.This capacity might be inadequate to meetpotential losses if several large insurers fail.

The National Organization of Life and HealthInsurance Guaranty Associations (NOLHGA)was set up about two decades ago to help statesdeal with insurance company insolvencies thatinvolve three or more states. About 150 suchinsolvencies involving multi-state insurers haveoccurred since NOLHGA began in 1983; mostwere small insurers. Total state guaranty associ-ation assessments since 1988 add up to about$5.8 billion. The largest insolvency involvedExecutive Life Insurance Company of Californiaand its subsidiary, Executive Life of New York.So far, state guaranty associations have paidabout $2.5 billion across the country for theExecutive Life insolvency (www.NOLHGA.org).Ultimate losses to policyholders remain in dis-pute (Box 4-3).

In general, when an insurer licensed to writeinsurance in a particular state becomes insol-vent, that state’s guaranty association (made up


of all insurers who are licensed in the state) isresponsible for providing protection to all resi-dent policyholders or annuitants of the insolventinsurer. Annuitants in states where the insolventinsurer is not licensed are covered, in most cases,by the guaranty association in the state wherethe insolvent insurer is located (www.NOLH-GA.org). Because each state writes its own rulesabout guaranty rules, jurisdictional disputescould arise if other large annuity providers wereto fail, although NOLHGA attempts to mini-mize these disputes.

Jurisdictional boundaries are important becauselife annuities are long-term contracts and anannuitant sometimes moves away from the stateof purchase. Depending on the state law, theguarantee might not cover out-of-state residents.To consider an example, suppose that Johnbought a life annuity from XYZ InsuranceCompany in Michigan and later moved to

Florida. Then XYZ Insurance Company failed.Would other insurance companies in Michiganbe obligated to make good on XYZ’s brokenpromise to John in Florida? If the failed XYZInsurance Company was also licensed in Florida,would the Florida guaranty fund (funded byFlorida insurers) be required to make good onthe promise to John? State rules may differabout exactly which annuitants are protected,what amounts are covered, which insurers’ obli-gations are covered, and which other insurerswill be taxed to meet those obligations. So, con-siderable complexity could ensue if one or morelarge annuity providers were to fail.

A Framework and ExistingAnnuity Arrangements

The various functions involved in providingannuities can be grouped into three broad

Box 4-3. Executive Life Insurance Company

In 1991, the failure of a large insurer active in the pension annuity markets, Executive Life InsuranceCompany (ELIC) of California, resulted in defaults that imposed losses on pension plan participants. In the1980s, Executive Life had been particularly aggressive in soliciting annuity business from terminatingdefined-benefit pension plans. It attracted this business by underpricing its competitors. Executive Lifeobtained its price advantage by pursing a higher-risk investment policy, investing heavily in so-called junkbonds. When the junk bond market declined, Executive Life became insolvent. In receivership, ExecutiveLife was able to pay its pension annuitants about 70 cents on their promised dollars (Langbein and Wolk,2000; U.S. GAO, 1992). State solvency funds covered policyholder losses up to state guaranty limits.

As part of a comprehensive rehabilitation plan adopted in September 1993, policyholders were given achoice to “opt-out” and accept new coverage from another company (Aurora Life Assurance Company)or “opt-in” and surrender their policies for cash. Over the years, as ELIC assets were liquidated, proceedswere distributed to Aurora toward the policies of those who “opted-in” and to individual ELIC policyhold-ers who “opted-out.”

In 1999, the Insurance Commissioner of California sued the entities that bought junk bonds from ELICduring the rehabilitation and the owners of Aurora Life Insurance Company. The suit alleges that theyintentionally deceived the Commissioner in order to gain control of ELIC’s junk bonds and insurance poli-cies. The suit seeks return of all profits gained by the defendants, or alternatively, all damages caused bytheir deceit. Any recoveries from the lawsuit would go to the policyholders (Conservation and LiquidationOffice, 2002). The ultimate size of losses experienced by ELIC policyholders who were not fully covered bystate guaranty funds remains in dispute (Executive Life Action Network, 2004).


categories – risk bearing, management andadministration, and regulation. At least threepotential entities could perform those functions– private insurers, state governments, and thefederal government. A three-by-three frameworkcan show the various combinations of ways tocarry out these functions. Before consideringnew institutional arrangements, this section usessuch a framework to describe two models thatcurrently exist: (1) the individual annuity mar-ket; and (2) arrangements for offering annuitiesto federal employees through the Thrift SavingsPlan. Either of these approaches might beacceptable for a system of voluntary supplemen-tal savings that offers wide choices about retire-ment payouts. Neither approach is suggested asa way to provide payouts from accounts thataim to replace, in part, for Social Security retire-ment benefits. Neither of these models providesinflation-indexed annuities; neither has otherfederal protections that retirees might expect ifthe federal government were to require orstrongly encourage the purchase of privateannuities.

Model 1: The Current Private AnnuityMarketThe current market for individual life annuitiesis quite small. The key players are individualswho wish to buy longevity insurance, insurancecompanies who provide it, and states who regu-late, as shown in Figure 4-1.

Private insurers bear mortality risk and invest-ment risk in offering life annuities. Individualannuitants bear the risk of inflation in that inflation-adjusted annuities are generally notavailable.3 Insurers also design, price, and mar-ket life annuities, manage their invested assets,and administer payments. All regulatory over-sight and solvency guarantees rest with state governments.

Existing institutional arrangements for annuitiesoffered by insurance companies under the super-vision of state regulators might be acceptable topolicymakers if the individual accounts werevoluntary supplemental savings plans in whichretirees had broad choices about the form andtiming of withdrawals. But current arrangementspresent many drawbacks for providing wide-

Figure 4-1. Current Private Annuity Market and State Regulation Actual Division of Responsibilities

Federal Government State Government Private Life Insurers

Risk Bearing

Mortality risk XInvestment risk XInflation risk n.a. n.a. n.a.

Management

Annuity design XPricing XConsumer education & marketing XAdministration XInvestment management X

Regulatory Oversight

Annuity design XConsumer education & marketing XSolvency regulation XSolvency guaranty X


spread annuities to replace part of SocialSecurity. Among the potential drawbacks are:the absence of inflation-indexed annuities in thecurrent market; different prices for men andwomen, and other market segmentation thatmight be viewed as problematic; possible short-comings of the state guaranty system if insur-ance companies should fail; and potentiallyinadequate consumer education.

Model 2: Annuity Arrangements in theThrift Savings Plan The Thrift Savings Plan (TSP) for federalemployees is another existing model for offeringlife annuities to retirees. It allows retirees broadchoices in form and timing of payouts. It alsointroduces a federal role in simplifying annuitychoices and in consumer education. In its role asplan sponsor, the federal government selects arange of annuity products to make available toits retirees and, through competitive bidding, thegovernment contracts with one or more insur-ance companies to offer those annuities toretirees.

Under this approach, the federal governmentacts to simplify the potentially vast array of

annuity choices, negotiates favorable prices, andinforms retirees about annuity choices and otherpayout options. As discussed in Chapter Three(Box 3-2), annuities available in the ThriftSavings Plan include: (a) monthly payments thatare fixed or will rise by 3 percent per year; (b)annuities with unisex pricing; and (c) joint-lifeannuities that are symmetric (such that the pay-ment to the longer-lived annuitant will be thesame whether he or she is the primary or thesecondary annuitant).

In the Thrift Savings Plan, the retiree is under noobligation to accept any of the annuity choices.He or she can take a lump-sum distribution andshop for an annuity in the individual market. Asillustrated in Figure 4-2, private insurers bearmortality and investment risk. The insurer pricesthe annuity and is responsible for selling andadministering it for the TSP participant whobuys it. Once retirees buy annuities, they dealwith insurance companies on all future transac-tions. The insurer is responsible for investing theannuity premiums. Regulatory oversight and sol-vency guarantees remain with the states.

Figure 4-2. Annuity Arrangements in the Thrift Savings Plan Actual Division of Responsibilities


Risk Bearing

Mortality risk XInvestment risk XInflation risk n.a. n.a. n.a.

Management



Annuity design XConsumer education & marketing n.a. n.a.Solvency regulation XSolvency guaranty X


Like the current annuity market model, the TSPapproach might suffice for a new system of indi-vidual accounts if the funds were envisioned asdiscretionary supplemental savings and retireeshad wide discretion in payouts. TSP goesbeyond the purely private market model in thatthe government ensures that participants have aspecified number of annuity choices and requiresunisex pricing.4 Nonetheless, the TSP approachhas shortcomings if payouts are considered to bean integral part of Social Security because theTSP does not provide inflation-indexed annu-ities. Further, reliance on state guaranty fundscould be problematic if the federal governmentwere to require, or strongly encourage, retireesto buy life annuities with their accounts.

Providing Inflation-IndexedAnnuities

Life annuities that are automatically adjusted forinflation would expose the insurer to inflationrisk in addition to mortality and investment risk.Private insurers in the United States rarely offersuch products, yet many proposals for individ-ual accounts call for inflation-indexed annuities.

In theory, there are at least three ways to try tohave annuities keep pace with inflation, but onlyone provides full inflation protection. The firstmethod would be to purchase a variable annuityinvested in a portfolio that might keep pace withinflation. Recent research suggests that variableannuities invested in stocks do not ensure infla-tion protection; in fact, over short time hori-zons, stock returns and the rate of inflation tendto move in opposite directions (Brown et al.,2001).

Another way to seek inflation protection withexisting products would be to buy an annuitythat rises at a pre-determined rate, which mightmatch the inflation rate. This strategy wouldnot, of course, protect against unexpected surgesof inflation. For example, a 3 percent pre-deter-mined increase in annuity payments would fallshort of full protection if the cost of living roseby 10 percent, or if double-digit inflation were

to persist for several years in a row, as occurredin the late 1970s and early 1980s (Figure 3-2).

The third, and only, way to ensure full protec-tion against inflation is to have an annuity beautomatically adjusted each year by the actualchange in the consumer price index experiencedin the recent past. Such products do not yet existin any significant numbers in the private annuitymarket in the United States.

Developing a market for such inflation-indexedannuities would probably involve the federalgovernment in some way. The government couldissue a large volume of inflation-indexed securi-ties, which private annuity providers could useto hedge inflation risk; the government couldreinsure private insurance companies for theinflation risk; or, the government could issueinflation-indexed annuities directly to retirees.Absent any of these developments, inflation-indexed annuities are not likely to evolve (Box4-4).

United States Treasury Inflation-Protected SecuritiesPrivate insurers might sell inflation-indexedannuities if they could back them with securitiesguaranteed to keep pace with inflation. U. S.Treasury Inflation-Protected Securities (TIPS) aresuch a product. The Treasury Departmentadjusts the principal value of TIPS daily, basedon a rolling average of the consumer price indexin the recent past. The fixed rate of interest onthe security is applied to the adjusted principaland is paid every six months. At maturity,Treasury redeems the security at its inflation-adjusted principal amount (Box 4-5).

Some experts expected that the government’sintroduction of TIPS in 1997 would launch amarket in private inflation-indexed annuities.Several explanations have been offered on whythe market has not evolved. Weak consumerdemand is one possible reason. Just as thereseems to be weak demand for life annuities ingeneral, there seems to be even weaker demandfor life annuities that keep pace with the cost of


living. If consumers do not understand the riskof inflation, they might prefer a higher initialpayment without inflation protection instead ofa smaller initial payment that keeps pace withthe cost of living.

A second possible explanation is that the supplyof TIPS is not yet of sufficient size, duration,and predictability to support a market in infla-tion-indexed annuities. When TIPS were firstintroduced in 1997, the Treasury Departmentissued both 30-year and 10-year TIPS. But inOctober 2001 Treasury stopped issuing all 30-year securities. Only 10-year TIPS were issued inthe next two years. Some new five-year and 20-year TIPS were added in 2004 (Bitsberger,2004). Today, TIPS remain a relatively smallshare of the total Treasury securities market; inJune 2004 they totaled about $200 billion, orabout 6 percent of the total Treasury securitiesmarket of about $3.3 trillion. Only about $40billion in TIPS are of 30 years’ duration, a peri-od that insurers might consider necessary tocover life spans of new retirees.

A third possible reason why a market in infla-tion-indexed annuities has not developed is thatinsurance companies and their regulators havenot yet worked out how inflation risk mightinteract with the other risks that insurers bear.Even if inflation risk is fully backed by the gov-ernment through TIPS, inflation-indexed annuitypayments would increase insurers’ exposure to

mortality risk. That is, with fixed annuities,insurers lose money if they underestimate theirannuitants’ life spans, but the losses would bemitigated by the fact that unanticipated pay-ments in the distant years would have declinedin value due to inflation. If annuity payments inthe out years kept pace with inflation, then loss-es to the insurer due to unanticipated increasesin longevity would be much greater.

Federal Reinsurance of Private AnnuitiesInstead of expanding its supply of TreasuryInflation Protected Securities, the governmentcould try to encourage a private market in infla-tion-indexed annuities by serving as reinsurerfor private-sector annuity providers.Reinsurance means that all or part of the origi-nal insurer’s risk is assumed by another entity inreturn for part of the premium paid by theinsured. The federal reinsurance could be limitedto the insurer’s risk of inflation in excess ofsome predicted level, or the government couldreinsure the full set of provider risks—that is,mortality and investment risk, as well as inflation.

Federal Government as Annuity Provider The third possible role for the federal govern-ment in promoting inflation-indexed annuitieswould be to provide annuities directly toretirees. The federal government already hasexperience paying Social Security benefits and

Box 4-4. Private Insurers and Inflation Protection

In theory, private insurers could take on inflation risk if they could charge a sufficiently high premium tocover the risk. A key question is what regulators would require as reserves for such a product. To date,the National Association of Insurance Commissioners has not developed detailed rules for how to com-pute reserves on true inflation-indexed annuities (as distinct from annuities that rise by a pre-determinedpercentage each year). Experts in the insurance industry believe, however, that because of states’ empha-sis on solvency considerations, the required reserves for inflation-indexed annuities would be so large asto make the product essentially impossible to sell. That is, the required capital could be so great as tomake inflation-indexed annuities cost more than customers would be willing to pay for the product.

Source: Correspondence, Bruce Schobel, New York Life Insurance Company, 2004


civilian and military employee pensions indexedfor inflation. New issues that arise with the gov-ernment as an annuity provider are discussedlater in this chapter.

Before examining particular arrangements forproviding inflation-indexed annuities, we exam-ine three cross-cutting questions. First, how bigwould the inflation-adjusted annuity market be?Second, what role might capital markets play inproviding inflation-indexed annuities (Box 4-6)?And third, how binding is the obligation to payautomatic inflation adjustments, whatever theinflation rate may be?

How Big Would the Annuity ReservesBe? If an individual account system required orencouraged retirees to buy inflation-indexedannuities, how big would the annuity reservesbe? The answer would depend on the size of thecontributions to the accounts, the investmentreturns, and the degree to which funds wereannuitized. We use three possible benchmarks toillustrate the potential size of annuity reserves ina mature system. The first benchmark is grossdomestic product (GDP), which is the value ofall goods and services produced by the U.S.

economy in a year. A second benchmark is theaggregate of all publicly held government securi-ties (the federal debt held by the public). Yet athird benchmark is the value of all financialassets in the United States.

As an illustrative plan, we assume that all work-ers contribute 2 percent of their Social Securitytaxable earnings to individual accounts, hold thefunds until retirement, and then buy inflation-indexed annuities. In this illustrative scenario,the reserves backing the annuities would equalabout 15 percent of GDP when the system wasfully mature, according to estimates by theOffice of the Chief Actuary of the SocialSecurity Administration.6 This suggests that ifthe annuities were provided by insurance com-panies and fully hedged by inflation-adjustedsecurities issued by the federal government,those securities would amount to about 15 per-cent of GDP. Similarly, if the government pro-vided inflation-indexed annuities directly toretirees, then reserves held by the government toback the annuities might equal about 15 percentof GDP.7

Today, the aggregate of all government securitiesheld by the public (the publicly-held federal

Box 4-5. U.S. Treasury Inflation-Protected Securities (TIPS)

TIPS are a special class of Treasury securities. Like other securities, TIPS make interest payments every sixmonths to holders and pay back the principal when the security matures. TIPS are unique, however, inthat the interest and redemption payments are tied to inflation.

The Treasury Department adjusts the principal value of TIPS daily, based on the consumer price index(CPI). The fixed rate of interest is then applied to the inflation-adjusted principal. Each interest paymentwill be larger than the previous one if inflation occurs throughout the life of the security. At maturity,Treasury redeems the security at its inflation-adjusted principal amount (or, in the rare event of sustaineddeflation, at the higher par value).

TIPS are generally exempt from state and local income taxes but are subject to federal income tax. Whenthe principal grows, the gain is reported as income for the year, even though the inflation-adjusted princi-pal is not actually paid until the security matures.

Source: U.S. Department of the Treasury, 2003. Bureau of the Public Debt, Treasury Inflation-Indexed Securities (TIPS)


debt) is about one-third of GDP. So, in this sce-

nario, inflation-indexed government securities

that would back private annuities might be

equivalent to about half of today’s federal debt.

Or, if government were the annuity provider, its

annuity reserves in the illustrative individual

account system would be equivalent to about

half the publicly held federal debt today.

Finally, we can compare the estimated size of

annuity reserves to the value of financial assets

in the United States. Today, total financial asset

values are roughly twice the size of GDP.8 This

suggests that reserves backing annuities in the

fully mature individual account system outlined

here might amount to about 7 percent to 8 per-

cent of national financial assets.

Box 4-6. A Potential Role for Capital Markets

For the most part, this chapter proceeds as if the risks associated with underwriting inflation-indexedannuities must be borne by private insurance companies and/or the federal government. It is possible,however, that private capital markets might also play a role in assuming some of the risks of underwritinginflation-indexed annuities. Theoretically, the package of risks embedded in an inflation-indexed life annu-ity could be split into pieces, and capital-market investors could be paid for bearing some portion ofthese risks. The capital markets now share the risks of underwriting mortgage loans, which financialintermediaries once bore exclusively. By gaining access to new sources of capital in this way, the size ofthe market for inflation-indexed annuities could grow and the pricing for consumers could improve.

Inflation-indexed annuities carry three main risks that could be broken out: mortality, inflation, and invest-ment performance. Capital markets routinely take on investment risk. But some investors might find mor-tality risk attractive because it is virtually uncorrelated with other financial investments, increasing thedegree of diversification in their portfolios. The property and casualty insurance industry has already dis-covered that some investors are willing to take on the risk of hurricanes or earthquakes, risks traditionallyborne exclusively by insurance companies. Blake and Burrows (2001) have proposed that governmentsissue “survivor bonds” that would pay out solely on the basis of future mortality rates for the populationas a whole. Dowd (2003) has suggested that private markets could equally well meet this need.

There is also room for the capital markets to complement the government’s role in taking inflation risk offthe backs of the insurers. First, the United Kingdom, which has issued inflation-indexed governmentbonds for some years, has developed a modest market of privately issued inflation-indexed bonds, suchas from utilities. A complementary market could develop in the United States as well. Second, investmentbanks are capable of creating “synthetic” inflation-indexed corporate bonds by combining Treasury infla-tion-protected securities with existing corporate bonds. Availability of inflation-indexed corporate bondsmight allow insurers to provide better annuity pricing by increasing the rate of return on the investmentsthey use to match future inflation.

The degree to which capital market innovations might facilitate the underwriting of inflation-indexedannuities is a matter of speculation. Experiences over the past few decades suggest that such innovationis at least a possibility. If these market innovations occur, it will be important that regulatory oversightkeeps pace in order to handle the greater complexity that can arise when complex instruments, such asmortality bonds or inflation-indexed corporate bonds, are developed to take advantage of the flexibility ofthe capital markets.


A larger or smaller individual account programwould, of course, result in larger or smalleraggregate annuity reserves. For example, if allworkers contributed 4 percent of wages insteadof 2 percent of wages, the reserves would betwice as big in a fully mature system. These esti-mates reflect a fully mature system after almostall retirees would be receiving annuities basedon lifelong accumulations. The annuity reserveswould accumulate gradually as early cohorts ofretirees would have been in the individualaccount system only part of their lives.

How Binding Would the Government’sInflation-Indexing Obligations Be? The nature of the government’s promise to pro-vide inflation-indexed payments in the futurecould be somewhat different depending on howthe inflation protection is provided. Would thegovernment be: (a) promising future cost-of-liv-ing increases to Social Security beneficiaries; (b)issuing large quantities of long-duration, infla-tion indexed Treasury securities to investors; or(c) entering into inflation-indexed annuity con-tracts with individual retirees? In consideringways to provide inflation protection to retirees,policymakers may want to also consider theimplications of modifying the automatic infla-tion adjustment if future circumstances shouldwarrant such action.

Social Security as a Statutory Entitlement

Social Security benefits are a statutory entitle-ment, which Congress can change by amendingthe Social Security Act. The benefits are not acontractual obligation of the federal govern-ment, as the Supreme Court decided in 1960.9

While Congress has generally approachedchanges in future Social Security benefits withcare, it has renegotiated the social insurancecompact a few times in the past. In fact, somehave argued that the very nature of a pay-as-you-go defined-benefit social insurance systemwill require adjustments in benefits or revenuesfrom time to time to keep it in balance(Diamond, 2004). Changes in automatic cost-of-living increases were part of the benefit adjust-ments in 1983. In the late 1970s and early

1980s, inflation (on which benefits are based)rose much faster than wages (on which revenuesare based). This brought a serious near termshortfall in Social Security finances. The solven-cy legislation enacted in 1983 included a provi-sion to delay all future cost-of-living adjustmentsby six months as part of a package of revenueand benefit changes. That provision accountedfor 24 percent of the solution to the near-termfinancial shortfall and about 15 percent of thesolution to the long-run imbalance (Svahn andRoss, 1983).

Treasury Securities as ContractualObligations

In contrast with Social Security benefits,Treasury Inflation-Protected Securities are con-tractual obligations that are backed by “the fullfaith and credit of the United States’ govern-ment.” From the federal government’s perspec-tive, issuing long-duration, inflation-indexedsecurities would represent a more binding com-mitment to make inflation-indexed outlays inthe future than is represented by the promise ofinflation-indexed Social Security benefits. Itwould be a major departure from precedent –with significant implications for domestic andinternational financial markets – if the govern-ment tried to modify the terms of its outstand-ing debt. Inflation has been modest sinceTreasury introduced inflation-indexed securitiesin 1997, so it does not yet have experience withthese products during periods of unexpectedlyhigh or erratic inflation.

Inflation-Indexed Annuities as Contract orPromises

Inflation-adjusted annuities issued by the federalgovernment could, in theory, be designed tohave the same legal status as Social Security ben-efits – that is, a statutory expectation that couldbe changed through the legislative process.Another option is that they could be designed ascontractual obligations backed by the “full faithand credit of the United States’ government,”akin to government bonds. If individuals buy theannuities by relinquishing funds from their per-sonally owned accounts, then the case for treat-


ing the annuities as binding contracts would bestronger. Policymakers may want to address thelegal status of the annuity contract and automat-ic inflation protection in thinking through thedesign of retirement payouts in a new individualaccount system.

The next section of this chapter examines hybridapproaches to institutional arrangements forproviding life annuities indexed to keep pacewith the cost of living.

Hybrid Models to ProvideInflation-Indexed Annuities

If the goal of an individual account plan is tooffer widespread life annuities at retirement thatare indexed for inflation, then what would bethe respective roles of insurers, state govern-ments, and the federal government? This sectionexplores several hybrid arrangements. Model 3relies on private insurers, but involves the feder-al government to help insurers bear inflationrisk. Model 4 sets up a central administrativeauthority for administrative activities, but con-tinues to rely on the private sector to insuremortality risk. Finally, model 5 envisions thegovernment as annuity provider, but with invest-ment management functions contracted out tothe private sector.

Model 3: Private Annuities Backed byTIPSUnder this approach, the federal governmentwould issue TIPS in sufficient volume and dura-tion to back privately insured, inflation-indexedlife annuities. Private insurers would bear mor-tality risk and investment risk, as illustrated inFigure 4-3. The government might play a role inannuity design, for example, by stipulating uni-sex pricing and setting minimum standards forjoint-life annuities. Nonetheless, retirees mightstill have a broad range of choices about guar-antee features, the timing of annuities, level ofsurvivor protection, and whether to buy annu-ities at all. The government might also takesome role in consumer education. Consumereducation responsibilities could be substantial if

retirees have many choices, as discussed inChapter Three.

In this model, private insurance companieswould administer the annuity payments. Theinsurers would be responsible for making timelypayments, adjusting annuities for the cost-of-liv-ing each year, keeping track of annuitants’change of address or change of direct depositinstitutions, reporting annuity income to theInternal Revenue Service (and withholdingincome taxes as appropriate), documentingannuitant deaths, and making the transition tosurvivor payments in joint-life annuity cases.

At least three important policy issues remain inthis model with regard to: whether and how todeal with adverse selection and selective market-ing in the private annuity market; how the fed-eral government would use and manage thefunds it receives when it issues a large volume ofTIPS; and whether the federal governmentwould have a role in regulating and ultimatelyguaranteeing the solvency of private annuityproviders or leave it with the states.

Adverse Selection and Selective Marketing

The more choices retirees have about whether tobuy annuities, when to buy them, and what typeto buy, the greater the likelihood that adverseselection and selective marketing will occur, asdiscussed in Box 4-2. For example, if federalpolicy required unisex pricing of life annuities,then men might delay or avoid buying them. Atthe same time, insurers would want to selective-ly market life annuities to men, because menwould be more profitable customers on average.More generally, if retirees have a choice whetherand when to buy annuities, those who are inpoor health or believe they have a short lifeexpectancy will avoid or delay buying annuities,which will drive up the average price of annu-ities for those who do buy them.

Government Use of Funds from Sale of TIPS

If the government issues a large volume of TIPSthat insurers would use to back inflation-indexed annuities, how would the government


account for and manage the money it receivesfrom issuing TIPS? Would the long-durationTIPS gradually substitute for existing shorterduration, non-indexed debt instruments as theymatured? Or would the long-duration TIPS addto the federal debt? It is possible to envision sev-eral ways in which the government might usethe new funds. In theory, it might spend themoney for current budget obligations, or investit in public infrastructure (which might resemblespending). Alternatively, it might set up institu-tional arrangements to invest the funds in pri-vate financial markets. These new institutionsmight seek to wall off investment decisions fromother federal priorities.

Solvency Regulation and Guarantees

If the federal government required or stronglyencouraged people to buy life annuities, shouldit guarantee those annuities? According toMackenzie, (2002), a moral hazard issue couldarise if the government guaranteed privatelyprovided annuities. Government guaranteescould “create an incentive for the provider tooffer excessively generous terms (i.e., relatively

low premiums) to attract customers since cus-tomers will suffer no loss even if the company’sinvestment experience is unfavorable.”

Nonetheless, if the government required orstrongly encouraged the purchase of privateinflation-indexed annuities, buyers might expectthe government to guarantee those annuities incase of insurance company failure. If the federalgovernment were to be the solvency guarantor,then policymakers might want the federal gov-ernment to have a role in solvency regulation aswell – that is, in setting reserve requirements,investment restrictions, and capital require-ments. If the government were involved in set-ting these standards, jurisdictional issues wouldarise regarding the interaction between federalrules, which (at a minimum) would apply to fed-erally specified life annuities, and state rules,which might still apply to other business of thelife insurance companies. The layering of federaland state regulation could introduce new complexities for both insurance companies andconsumers.

Figure 4-3. Private, Inflation-Indexed Annuities, Backed by TIPS Hypothetical Division of Responsibilities


Risk Bearing

Mortality risk XInvestment risk XInflation risk X

Management

Annuity design XPricing XConsumer education & marketing X XAdministration XInvestment management X


Annuity design XConsumer education & marketing XSolvency regulation ?? ??Solvency guaranty ?? ??


The life insurance industry is currently develop-ing a proposal to give insurance companies whodo business in more than one state the option ofoperating under federal, rather than state, regu-lations (Insure.com, 2004). A proposal put forthby the American Council of Life Insurers wouldcreate an Office of National Insurers within theU.S. Treasury Department. The Office mightresemble the Office of the Comptroller of theCurrency for banks, or the Office of ThriftSupervision for savings institutions. A presiden-tial appointee would head the Office, whichwould be funded by assessments on insurancecompanies. The new federal Office would over-see consumer protection, marketplace practices,and the regulation of sales and premiums rates.Insurance companies could choose whether tooperate under state or federal regulation.Federally regulated insurers would be requiredto join the guaranty association of each state inwhich they do business. The proposal wouldalso set up a national guaranty association ofinsurers operating in states where the guarantyfund does not meet federal standards. To date,the proposal is generally opposed by theNational Association of Insurance Commission-ers and the National Conference of StateLegislatures and has not been acted upon byCongress (Insure.com, 2004).

Model 4: Federal Administration: PrivateRisk Bearing and Fund ManagementThis approach might set up a new centraladministrative authority – either as part of thegovernment or a new semi-private entity toadminister interactions with account holders andannuitants. As in the prior model, the govern-ment would issue TIPS in sufficient volume andduration to back the insurers’ inflation risk.Private insurance companies would underwritethe annuities and bear the mortality and invest-ment risk as bulk annuity providers describedbelow (Figure 4-4).

The design of annuities would be specified infederal law or regulations. The governmentwould be responsible for consumer educationand would sell annuities to individual retirees

and collect their premiums. The governmentwould then package large pools of annuities andcontract with private insurers who, in return forreceiving bulk premiums, would make futurebulk annuity payments back to the governmenteach month to cover promised payments toannuitants. The government would constructpools of annuitants with demographic character-istics that represent the entire population ofannuitants. This pooling could avoid the risk ofselective marketing by insurers and give the gov-ernment the capacity to manage adverse selec-tion issues.

For the insurance companies who serve as con-tractors, the government could set contractualrequirements for creditworthiness, servicingcapability, or other relevant factors. Insurerswho met these requirements would be allowedinto a panel where they would be able to bid onthe pooled annuities.

The federal government would make the actualpayments to annuitants, possibly by adding theannuities to Social Security benefits. The govern-ment would be the recordkeeper and wouldinform each insurance company of the aggregatemonthly annuities payable to annuitants in thatcompany’s pool each month.

In this model, insurance companies would setprices based on pools set up by the federal gov-ernment. As with Model 3, two questionsremain. First, would the existing system of statesolvency regulations and solvency guaranties besufficient if such annuities were mandatory orstrongly encouraged by the federal government?Second, how would the government account forand manage the money it received from the saleof a large volume of long-duration TIPS?

Model 5: Federal Annuity Provider:Private Investment ManagementIf the goal of an individual account plan were toproduce widespread life annuities at retirementthat resemble aspects of Social Security, then thefederal government might decide to providethose annuities directly. The federal government


already has experience paying Social Securitybenefits. As discussed in Model 3, if the federalgovernment were to be the guarantor of individ-ual account annuities, then it might be simplerto have the government bear the risks, ratherthan have it involved in regulating and ultimate-ly guaranteeing the solvency of private insurers.Figure 4-5 illustrates a framework for arrange-ments in which the federal government is theprovider of inflation-indexed annuities.

As is the case with Social Security benefits, thefederal government would bear mortality riskand inflation risk. The government would alsodesign annuities as called for in the individualaccount law, inform account holders about thenew system, educate consumers, and carry outall administrative tasks in dealing directly withannuitants.

If the government were the annuity provider,important issues remain with regard to theinvestment management function. In contrast toSocial Security, genuine life annuities would bewholly prefunded, placing both investment riskand investment management responsibilities onthe federal government. The reserves backing

universal annuities that are funded with 2 per-cent of workers’ earnings could amount to 15percent of GDP when the system is fully mature.This large volume of funds poses a number ofnew questions. What investment policy wouldapply to the funds and who would be responsi-ble for the investments? How would the fundsbe viewed in terms of the unified federal budget?

How Would Annuity Reserves Be Invested?

Should the investment of annuity reserves followthe rules for the Social Security trust funds,which are invested solely in special-issueTreasury securities? Or should annuity premi-ums be invested in a more diversified portfolio?Diversification into corporate bonds and stockswould produce higher expected returns over thelong run, but doing so would create major con-cerns in the business community about federalinvolvement in corporate decisions.

One approach might follow the Thrift SavingsPlan precedent, which invests in private indexedfunds through one or more private fund managersselected through a competitive bidding process.

Figure 4-4. Private Risk Bearing and Federal AdministrationHypothetical Division of Responsibilities


Risk Bearing

Mortality risk XInvestment risk XInflation risk X -- TIPS

Management



Annuity design XConsumer education & marketing XSolvency regulation ?? ??Solvency guaranty ?? ??


The fund managers’ sole responsibility is to earnreturns that match the performance of the indexfunds. Fund managers do not invest in any par-ticular company’s stocks or bonds. With theTSP, the allocation of funds across the indexfunds is determined by the aggregate choices ofplan participants. If policymakers followed thisstrategy for investing annuity reserves, someinvestment board or other entity would need toselect the portfolio allocation across variousindex funds and government securities (Box 4-7).

How Would Annuity Assets Be Counted inthe Unified Budget?

Budget scoring rules can affect how policymak-ers think about various types of federal fundswhen they make taxing and spending decisions.A large volume of annuity reserves – if viewedas government receipts when premiums arereceived – could distort the federal government’staxing and spending decisions. So, mechanismsmight be needed to segregate annuity reservesfrom other federal funds. Such arrangementswould be a top priority if the federal govern-ment were to become an annuity provider. Box

4-8 describes some of the current conventionsand issues in budget treatment of various kindsof federal transactions.

Summary

Existing institutional arrangements for makingannuities available to retirees might suffice for anew system of individual accounts if thoseaccounts were voluntary, supplemental savings,and retirees had wide discretion about fundwithdrawals at retirement. But if policymakerswanted to require, or strongly encourage,retirees to take payments in the form of lifeannuities that resemble Social Security, then newinstitutional arrangements would be needed.Developing a market for inflation-indexed annu-ities would involve the federal government insome way. The government might issue a largevolume of Treasury Inflation-Protected Securitiesto help private insurers hedge inflation risk; itmight reinsure private insurers; or it might issueinflation-indexed annuities directly to retirees.

The volume of reserves required to back wide-spread inflation-indexed annuities would be

Figure 4-5. Federal Annuity Provider and Private Investment Management Hypothetical Division of Responsibilities

Federal Government State Government Private Fund Managers

Risk Bearing

Mortality risk XInvestment risk XInflation risk X

Management



Annuity design XConsumer education & marketing XSolvency regulation n.a.Solvency guaranty X


Box 4-7. Examples of Thrift Savings Plan Index Funds

TSP participants have several investment options. The C Fund aims to match the performance of the S&P500 index; the S Fund aims to track the performance of the Wilshire 4500 index, a broad market indexmade up of stocks of U.S. companies not included in the S&P 500; and the I Fund aims to match the per-formance of the Morgan Stanley Capital International EAFE (Europe, Australasia, Far East) index, a marketindex made up of stocks of companies in 21 developed countries.

Box 4-8. Budget Scoring Conventions

Treasury securities. When Treasury issues securities, no budget transaction occurs. It is simply an exchangeof assets. The government receives cash and issues securities of equal value. There is no budget outlay andno receipt.

Federal loan programs. Under the Credit Reform Act of 1990, federal loan programs are scored in a waysimilar to Treasury securities. Loans are an exchange of assets, but the assets are not of equal value if theloan program subsidizes certain groups. Only the present value of the subsidy is counted as a budget outlayin the year the loan is approved.

Entitlements programs. Under the unified budget, Social Security revenues are counted as receipts andbenefit payments are counted as outlays. Reserves held by the trust funds are viewed as governmentmoney.

Contributions to the Thrift Savings Plan (TSP). When federal agencies put money in the TSP accountsof their employees, the money is counted as a federal outlay when it is contributed. The money in TSPaccounts is outside the federal government and owned by account holders.

Federal insurance programs. Federal insurance activities are generally scored on a cash basis. For exam-ple, Pension Benefit Guaranty Corporation premiums are counted as receipts and pension payments arecounted as outlays. Experts have recommended that concepts reflected in the Federal Credit Reform Act of1990 would provide better accounting of federal insurance.

Federal investment in corporate bonds or stocks. There are conflicting views about how the federalgovernment’s purchase of corporate stocks and bonds should be scored. Office of Management and BudgetCircular A-11, a basic document on budget rules, says that such a purchase is a government outlay, whilethe sale of such assets is a government receipt. Many budget experts believe this rule is wrong. They arguethat it would be more consistent to treat the transaction as an exchange of assets. When the assets are ofequal value, there is no budget transaction. Consistent with this view, Congress in 2002 allowed theRailroad Retirement Board to invest its assets in stocks and corporate bonds and both the CongressionalBudget Office and the Office of Management and Budget scored the transactions as an exchange of assets.

Sources: Douglas Elliott, Center on Federal Financial Institutions, and Richard Kogan, Center on Budget and Policy Priorities, personalcommunication, 2003


substantial. Reserves backing universal annuitiesfunded with 2 percent of workers’ earningscould amount to 15 percent of GDP when thesystem is fully mature. Whether the governmentprovides annuities directly to retirees or providesTIPS to back privately issued annuities, the gov-ernment could be holding very large amounts ofassets backing the annuities. A key question forpolicymakers to address would be who wouldmanage and invest the large volume of assets.New arrangements might be needed to segregatethe funds from other taxing and spending func-tions of the federal government and new institu-tions might be needed to provide for prudentand diversified investment of the funds.

If inflation-indexed life annuities were provided

on a widespread or universal basis by private

insurance companies with backing from TIPS,

policymakers might want the federal govern-

ment to be involved in guaranteeing the solvency

of the insurance companies. Unlike federal

insurance funds – such as the Federal Deposit

Insurance Corporation, which insures bank

deposits – solvency of private insurance compa-

nies is left to the 50 states. Proposals for the fed-

eral government to charter life insurance

companies might gain broader interest if life

annuities were to be encouraged or required by

federal policy.

Chapter Four Endnotes

1 Annuities that are automatically adjusted forinflation would also expose the insurer to infla-tion risk. Such products are rarely available inthe U.S. annuity market. Issues and options forproviding inflation-indexed annuities are dis-cussed later in this chapter.

2 The $100,000 cap in some cases applies to the sumof all life insurance and annuity products a cus-tomer has with the company. In some jurisdictionsthe cap, or combined cap is higher, say $300,000.

3 With rising life annuities, as described in Box 4-1, annuitant would have inflation protection aslong as inflation did not rise faster than theannual adjustment in annuity payments.

4 Some consumers might find it advantageous tobuy annuities in the private market if they haveshort life expectancies that permit more favorablepricing. This adverse selection could cause pricedifferentials to reemerge to some degree, notwith-standing uniform pricing for federally negotiatedoptions. See Box 4-2, Adverse Selection, UniformPricing and Selective Marketing.

5 Cost-of-living adjustments to Social Security ben-efits are based on the increase in the ConsumerPrice Index for Urban Wage Earners and ClericalWorkers (CPI-W), referred to here simply asinflation.

6 The estimate assumes that individual accountfunds earn a net real return of 4.6 percent peryear and that annuity reserves earn a net realreturn of 3.0 percent per year. When the system isfully mature, total assets in accounts plus annuityreserves would be about 45 percent of GDP.Annuity reserves would be about 15 percent ofGDP, while accounts not yet annuitized would beabout 30 percent of GDP. For comparison, TIAA-CREF, the largest defined-contribution retirementsystem in the United States, has combined assetsin retirement accounts and annuity reserves ofabout 3 percent of GDP today.

7 The entire portfolio of annuity reserves might notbe invested in inflation-indexed securities. Thepoint of this estimate is to suggest the size of theassets backing the inflation-indexed annuities.

8 The market valuation of domestic based equitiesis roughly equal to GDP. Adding all governmentand corporate bonds brings the total to roughlytwo times the size of GDP.

9 The Supreme Court found that “To engraft uponthe Social Security system a concept of 'accruedproperty rights' would deprive it of the flexibilityand boldness in adjustment to ever-changing con-ditions which it demands.” — Fleming v. Nestor,363 U.S. 603 (1960).

Chapter

97

5Pre-Retirement Access toIndividual Accounts

Individual savings accounts for retirement posean essential question: would workers haveaccess to the money in the accounts beforeretirement? If so, what limitations would therebe on pre-retirement access to the funds? Thischapter reviews issues that might influence therules for early access to the funds, including thevarious precedents and options for access rules.

Four options are presented for access rulesbefore retirement. Option One: Broad EarlyAccess allows unconstrained access to the fundsand is based on the precedent of IndividualRetirement Accounts (IRAs). Option Two: NoEarly Access is at the other end of the spectrum.Through banning access to the retirement funds,Option Two aims to resemble features of SocialSecurity. Option Three: Limited Early Accessaims to discourage, but not wholly ban, pre-retirement access to the funds. Finally, OptionFour: Access Only Above a Threshold bansearly access in the same way as Option Two, butthe ban is lifted on account accumulations thatexceed some threshold thought to produce anadequate level of retirement income.

Specific policy and implementation issues areraised by the various options. For instance,allowing early access raises policy questionsabout the type and form of access—whetherloans or withdrawals, and for what purposes—and the types of restrictions, if any. Any restric-tions, as well as an outright ban on early access,would require enforcement mechanisms andoversight.

No matter which rules are chosen, questions willarise about how early access rules would affectcreditor claims and the account holder’s eligibili-ty for means-tested assistance.

Framework for AnalyzingAccess Rules

Would workers be allowed to withdraw fundsfrom their accounts before retirement? If so,would there be any limits on the amount orform of the withdrawals? The answers to thesequestions will depend, in part, on at least fourfactors: the intended purpose of the accountfunds; the nature of the property rights associat-


Box 5-1. Lessons from Individual Development Accounts

Individual development accounts (IDAs) are subsidized savings accounts earmarked for low- and moder-ate-income individuals’ wealth building. Savings are subsidized for individuals through matching funds,not tax breaks. Participants earn matching funds for specified uses, usually the purchase of a home, post-secondary education, or to start a business; sometimes automobile purchases and retirement are permit-ted uses. IDAs have been tested in demonstrations funded by public and private sources starting in thelate 1990s, but have not yet been established on a permanent, national basis. At least 20,000 Americansare saving in IDAs.

The American Dream Demonstration, funded by a consortium of private foundations,1 enrolled about2,400 participants in 13 sites around the country, and another 840 in one experimental site. The demon-stration ran from 1997-2003 (Schreiner et al., 2002).

Separately, the 1998 Assets for Independence Act authorized federal funds for IDA demonstrations runby the Office of Community Services in the U.S. Department of Health and Human Services. Community-based, nonprofit organizations (project grantees) received federal funds to sponsor IDA programs for low-income workers. The workers enter into a savings plan agreement that sets goals and schedules. Savingsused for approved purposes are matched (which under federal rules could range from $1 for $1 to asmuch as $8 for each $1 saved, although the typical match rate was $2 for every $1 saved). The matchedsavings can be used only for: (1) the purchase of a first home; (2) to start a business; or (3) post-second-ary education. Participants must attend financial education classes. Withdrawal of subsidized savingsrequires the written approval of both the low-income saver and a project grantee official. Withdrawals ofthe saver’s own money for other purposes are discouraged and result in a loss of matching funds. Certainstatutorily defined emergency withdrawals are permitted, but matching funds are forfeited if withdrawnfunds are not replaced within 12 months (Office of Community Services, 2001).

No formal evaluations are yet available from the Assets for Independence demonstration; however, AbtAssociates and others have evaluated the privately funded American Dream Demonstration, yielding someuseful findings. Most important, Abt reported “the findings from this evaluation provide important newevidence that an IDA program can have significant favorable impacts on asset-building among low-income persons”(Mills, et al., 2004).

The key lessons from IDAs are that low-income people can and do save, and that low-income peoplehave multiple savings needs, not just long-term asset accumulation. As expected, the amounts are mod-est. The researchers found that saving was not strongly related to income, welfare receipt, or most otherindividual characteristics. Accordingly, institutional factors appear to shape savings success. Those factorsinclude: (a) access to a savings plan; (b) incentives in the form of matching funds; (c) financial education;(d) making savings easy through direct deposit and default participation; (e) setting targets and creatingsocial expectations; and (f) restrictions on access (Sherraden and Barr, 2004).

This “hands on” approach is costly relative to the net savings achieved (and higher than pure savingsproducts such as 401(k)s, but costs are comparable to other intensive interventions, such as Head Start.

Chapter Five: Pre-Retirement Access to Individual Accounts 99

ed with the accounts; whether participation isvoluntary or mandatory; and the funding sourcefor the accounts.

Purpose of Account Funds If the accounts are principally designed to pro-vide baseline economic security in old age (simi-lar to that now provided by Social Security),then there is a case for limiting access to accountfunds before retirement except, perhaps, if theaccount holder dies or becomes disabled.

But if the main purpose of the accounts were toprovide additional avenues for the accumulationof retirement wealth, then a ban on early accesswould be less important. Allowing some limitedaccess to the funds might be more appropriate.

A third type of individual account system mightaim to encourage fund usage before retirement.Such accounts could promote investment inhuman capital or wealth building—for example,to buy a home, pay tuition, or start a business—along the lines of individual developmentaccounts. In this case, the relevant questionsconcern whether and how to ensure that thefunds are used only for these earmarked expen-ditures (Box 5-1).

Property Rights to the AccountsThe intended use of the accounts also dependson the extent to which the funds are viewed aspersonal property of the account holder. If theaccounts are viewed less as personal propertyand more as social insurance, then it might beconsistent to limit early access to the funds. Ifthe accounts were personal property—either in alegal sense or in the popular view of accountholders—then it would be consistent to allowready access to the accounts.

Voluntary or Mandatory ParticipationA voluntary individual account system might beable to boost participation by allowing earlyaccess to the retirement savings. If contributionsto the system were voluntary, a ban on earlyaccess could discourage workers from participat-ing or lead them to contribute less. If workers

were required to participate and contribute spec-ified amounts, incentives to boost participationwould not be relevant.

Source of Account FundsThe government could make direct contributionsor extend tax credits to help workers build upaccount funds. If so, the case for banning earlyaccess might seem particularly strong, at leastwith respect to the portion of the funds not con-tributed by workers. Such government contribu-tions would likely not be viewed as an accountholder’s property in the same way as his or herown contributions, so more restrictions might beappropriate. Applying different rules to differentcomponents of the account balance, however,would make the overall system more complex.

Precedents for Access toRetirement Funds

U.S. retirement plans offer varied rules aboutearly access to set-aside funds. When participa-tion is voluntary and funds are viewed as per-sonal property, an account holder generally hasready access to the money, although taxes orpenalties may be levied on early withdrawals.Yet, social insurance systems are not personalproperty and do not allow access to the moneyexcept as prescribed benefits. Details about rulesfor early access to these plans are in AppendixFigure 5-A and are summarized briefly below.

Individual Retirement AccountsAmericans have unlimited access to funds theyset aside in tax-favored IRAs, although with-drawals are subject to federal income taxes if thefunds were tax-deferred when contributed.Withdrawals before age 591/2 are also subject toa 10 percent tax penalty unless certain condi-tions are met.

401(k) PlansEmployees’ access to tax-favored retirementfunds in 401(k) plans is more limited than withIRAs. But employees can usually get access tothe funds, either through loans (if employers


offer this feature), by taking hardship with-drawals, or by cashing out of the plan whenthey leave their jobs. Early withdrawals, includ-ing hardship withdrawals, are subject to incometaxes and generally to an additional 10 percenttax penalty.

Thrift Savings PlanThe Thrift Savings Plan (TSP), enacted in 1984for federal employees, is analogous to a private401(k) plan. The TSP allows employees to takeloans for any reason (with special terms formortgage loans) and allows withdrawals ifemployees can prove financial hardship, subjectto spousal consent or notice.2

Defined-Benefit Pension PlansIn a traditional defined-benefit pension plan,workers earn vested rights to future benefit pay-ments rather than to the plan’s assets. A defined-benefit plan may not pay benefits to workersbefore they leave employment or reach theplan’s normal retirement age. Many plans donot pay benefits to terminated employees untilthey have reached a defined retirement age, butit is becoming more common for plans to allowworkers to choose a lump-sum payment of theirvested benefits.

Social Security BenefitsThe Social Security program does not allowaccess to retirement benefits before retirement.The program pays benefits only as specified bylaw. Risks are shared broadly and benefits arenot the personal property of the recipient untilhe or she establishes entitlement to the benefitsand receives monthly checks. Benefits are paidonly when a worker dies, becomes disabled, orreaches retirement age.

Approaches in Individual AccountProposals Most proposals for individual accounts designedto replace part of Social Security would notallow access to the money before retirement;only if the worker dies would the money bereleased from the account to go to his or herheirs. The President’s Commission to Strengthen

Social Security examined this issue at length andconcluded in its 2001 report that early accessshould not be allowed. Given the prominence ofthe Commission and the relevance of its conclu-sions in this area, we quote its rationale in full:

While prohibiting pre-retirement accessmight seem very restrictive at first glance, itis important to recognize that even amongpeople facing difficult circumstances duringpre-retirement years, most are still expectedto spend some years in retirement. Difficul-ties in pre-retirement years do not justifyfacing even greater difficulties during retire-ment due to a lack of resources. While somepeople might suggest that accounts shouldbe accessible in some “hard cases” (e.g., dis-ability) we believe that those needs are besthandled with other government policy, andnot with funds set aside for retirement.Furthermore, allowing for pre-retirementaccess in the “hard cases” potentially opensPandora’s Box for less discriminatingaccount access in the future. In the sameway that Social Security benefits cannot beaccessed before retirement or used as collat-eral for a loan, neither should assets held inpersonal accounts be available for other pur-poses. However, unlike Social Security, assetsheld in personal retirement accounts can bebequeathed to heirs if the account ownerdies before retirement. In this way, wealthaccumulation in the family need not be cutshort with the death of the primary earner(PCSSS, 2001).

In general, other proposals for individualaccounts designed to partially replace SocialSecurity benefits also prohibit pre-retirementaccess to the funds, as shown in Figure 5-1. Oneexception is the Bipartisan Retirement SecurityAct of 2004 proposed by Representatives JimKolbe (R-AZ) and Charles Stenholm (D-TX)(HR 3821, 108th Congress). This plan wouldallow early access only to the account funds thatexceed the amount needed, when combined withany traditional benefits, to provide monthlyincome equal to 185 percent of the poverty line.


In sharp contrast with plans that ban or restrictearly access, the Social Security Plus plan, pro-posed by former Commissioner of SocialSecurity, Robert M. Ball, in January 2003,would follow IRA rules with regard to earlyaccess. In this proposal, the accounts are a vol-untary supplement to Social Security and wouldbe subject to the same tax treatment and with-drawal rules that apply to IRAs. Other propos-als for voluntary savings (including subsidized

savings) might also follow existing early accessrules for IRAs or 401(k) plans.

Options for Early Access

This section explores four options that put vary-ing restrictions on access to funds in individualaccounts before retirement.

Figure 5-1. Pre-Retirement Withdrawal Rules in Selected Individual Account Proposals

Proposal Pre-Retirement Withdrawal Provisions


ACSS (Gramlich): Individual Account Plan, 1996 • No withdrawals before age 62

Committee on Economic Development, 1997 • No pre-retirement withdrawals


ACSS (Schieber & Weaver): Personal Security Accounts, 1996 • No pre-retirement withdrawals

Reps. Kolbe-Stenholm: Bipartisan Retirement Security • Early withdrawals allowed only for funds in excess of Act of 2004 (H.R. 3821 in 108th Congress) those needed to produce income of 185 percent of

the poverty line (annuity combined with Social Security) taken as a lump sum

National Commission on Retirement Policy, 1999 • No withdrawals prior to retirement, disability, or death


President Clinton’s Retirement Savings Accounts, 2000 • Withdrawals only for purchase of a first home, medical emergencies, or paying for a college education; withdrawals allowed only after five years

Social Security Plus (proposed by R.M. Ball, 11/2003) • Withdrawals and tax treatment would follow IRA rules


President’s Commission (PCSSS) • No withdrawals prior to retirement or deathModels 1,2, 3 (2001)

Reps. DeMint-Armey: Social Security Ownership and • No withdrawals prior to retirement or deathGuarantee Act of 2001 (H.R. 3535 in 107th Congress)

Rep. Smith: Social Security Solvency Act of 2003 • No withdrawals prior to reaching age 591/2(H.R. 3055, 108th Congress)


Rep. Shaw: Social Security Guarantee Plus Act of 2003 • No withdrawals prior to retirement, death, or (H.R. 75 in 108th Congress) disability

Sources: U.S. Social Security Administration, 2004b, Selected solvency memoranda; National Academy of Social Insurance, December 1996,Social Insurance Update; Robert M. Ball, 2003, Social Security Plus, and November 2002 communication; Committee on EconomicDevelopment, 1997, Fixing Social Security


Option One: Retirement Accounts withBroad Early AccessOption One allows the broadest access toaccount funds and builds on experience withIRAs. It is consistent with retirement payoutsOption One in Chapter Three, which wouldallow unconstrained access to funds at retire-ment as well. Following the IRA model, contri-butions to the accounts could be voluntary, taxdeductible, and supplemental to both SocialSecurity and any employer-sponsored pensionsthe participant might have. If treated like tradi-tional IRAs, withdrawals would be subject toincome taxes and, if taken before age 591/2,would also be subject to a 10 percent tax penal-ty (unless certain other conditions are met, asexplained in Appendix Figure 5-A). A broadearly access policy might increase workers’ will-ingness to participate in voluntary accounts andto contribute more than they would in a lessaccessible system.

Option Two: No Pre-Retirement Access Option Two would ban access to retirementfunds before retirement except, perhaps, in thecase of the account holder’s death or disability.

(See Chapter Seven for a discussion on issuesconcerning access at disability.) The ban isdesigned to resemble features of Social Security,which ensures benefit payments at certain events– retirement, death or disability – but there is noaccumulated wealth for participants to with-draw or borrow against before benefits becomepayable.

Prohibiting pre-retirement withdrawals is consis-tent with a retirement payout requirement thatthe account be used to buy life annuities. In fact,a mandate to purchase life annuities at retire-ment could be rendered meaningless by a policythat permitted individuals to withdraw andspend the money before retirement. Similarly, arequirement for spousal protections in the formof joint-life annuities could be seriously erodedif workers could use the money for other pur-poses before reaching the mandatory annuitiza-tion age.

Option Three: Limited Early AccessOption Three is designed to discourage, but notban, early access to retirement savings. It is con-sistent with a goal of encouraging voluntary par-

Box 5-2. Permitting Loans and Banning Withdrawals

Both the federal employees Thrift Savings Plan (TSP) and 401(k) plans permit tax-free loans and hardshipwithdrawals subject to taxes and penalties. When a participant does not (or cannot) repay a loan, theunpaid balance is converted to a taxable withdrawal. In effect, the withdrawal is an escape valve toresolve unpaid loans.

If a system permitted loans and banned withdrawals, how would it deal with delinquent loans? The TSPhad such a policy before 1997; loans were permitted, but no in-service withdrawals were allowed. Theprogram avoided delinquent loans by requiring loan repayments through payroll deductions. Employeescould renegotiate loan repayments with their payroll office, but they could not default on their loan pay-ments as long as they remained on the federal payroll. If the employee left the government, any unpaidloan balance would become a termination withdrawal subject to taxes and penalties. Separation of serv-ice served as the ultimate safety valve to convert a loan to a withdrawal.

A national system of individual accounts would not be linked to a specific employer. Consequently, therewould be no “separation of service” safety valve to convert a delinquent loan to a withdrawal. Withoutthat safety valve, it might not be feasible to permit loans and ban withdrawals.


ticipation in a new retirement savings accountsystem, while minimizing leakage from it.Participants could be granted access to theiraccounts in cases of genuine need, but with fair-ly strict limits. For example, access could be per-mitted only as a last resort in financial hardshipcases. This option might offer only loans ratherthan outright withdrawals, although there arepractical considerations about how such a policywould work. As a practical matter, if loans arepermitted, it might be necessary to also allowwithdrawals, if for no other reason than toreach closure on unpaid loans. The ThriftSavings Plan originally permitted only loans, butthat policy was changed in 1997 (Box 5-2). Apolicy of allowing loans and discouraging with-drawals would need a way to close out unpaidloans.

Option Four: Ban Access Up to aThresholdOption Four would ban early access to accountfunds, as called for in Option Two, but only upto a certain threshold. It is consistent with aretirement payout option that would mandateannuities up to a threshold. Chapter Three dis-cusses various approaches to setting a thresholdfor mandatory annuitization. Possible thresholdswould include the poverty line, wage replace-ment goals, or means-tested program thresholds.Similar approaches could apply to a thresholdabove which workers could withdraw or borrowfunds before retirement.

All of these options would ban access to accountfunds for those most likely to need the access –that is, low-income individuals who have fewother sources of ready cash. At the same time,these approaches would expand choices forupper-income individuals who already haveother fungible assets; the accounts would add totheir repertoire of assets potentially available forborrowing or spending during their work lives.

If the mandatory annuitization threshold alsoserved as a threshold for pre-retirement with-drawals, new issues would arise about how totranslate a retirement-adequacy threshold to a

lump-sum threshold mid-way during the worklife – when it would not yet be known what theworker’s ultimate Social Security benefit or ulti-mate account balance would be by retirementage.


Any choice of pre-retirement access rules wouldlead to a number of key design and implementa-tion decisions. Allowing early access raises poli-cy questions about the type and form ofaccess—whether loans or withdrawals, and forwhat purposes—and the type of restrictions, ifany. Significant restrictions, as well as an out-right ban on early access, would require enforce-ment mechanisms and oversight. Questions ariseabout how early access rules would affect credi-tor claims and the account holder’s eligibility formeans-tested assistance, and finally, decisionswould be needed about how best to administerearly access rules.

Type and Terms of Early AccessDecisions about the type and terms of earlyaccess reveal competing goals. Participantsmight want, or even demand, access to themoney when they need it, particularly if theyview their accounts as personal property. But ifthe accounts were to serve as a basic source ofretirement security, minimizing leakage from theaccounts would be an important goal. So, ifaccess were to be allowed, this goal would sug-gest offering only loans and only for hardship.Finally, minimizing administrative burdens andcosts becomes another important objective.Administrative simplicity would argue for noearly access. If that policy failed, the secondchoice for administrative simplicity would allowoutright withdrawals for any reason – a policythat would maximize leakage.

Loans or Outright Withdrawals

As an early-access feature, loans could reducepermanent losses from the accounts, especially ifthey are paid back with interest. In 401(k)


plans, repaid loans are not subject to penaltiesor taxes, which makes 401(k)s appealing to participants.

The availability of loans does seem to increaseparticipation in 401(k) plans, although not asmuch as when the employer makes matchingcontributions to the plan. Loans appear toincrease the participation rate from about 54percent to 60 percent among workers whoseemployers did not provide matching funds, andfrom about 78 percent to 84 percent amongworkers whose employers did provide matchingfunds (U.S. GAO, 1997). Other studies suggestsimilar results (Hungerford, 1999; FidelityInvestments, 1999). The availability of loansalso appears to increase the amount that work-ers contribute; in one study, the ability to bor-row 401(k) funds increased workers’contribution rate by about 1 percentage point(Munnell et al., 2001).

Loans from 401(k) plans are fairly common—about 17 percent of eligible 401(k) participantshad outstanding loans in 2002. Loan balances,on average, amounted to about 16 percent ofthe remaining account balance (Holden andVanDerhei, 2003). Those who borrow from401(k) plans tend to be participants who havefew other financial options; borrowers tend tohave lower family income, lower net worth, andmore non-housing debt, on average, than donon-borrowers (U.S. GAO, 1997). AfricanAmerican and Hispanic participants are almosttwice as likely as white participants to borrowagainst their 401(k) accounts (U.S. GAO, 1997).Few defaults are recorded on 401(k) plan loans,but outstanding loan balances converted towithdrawals when employees change jobs donot count as defaults.

The primary drawback of permitting loans(compared with outright withdrawals) is greateradministrative burden. Outright withdrawals areone-time transactions in that plans can issuefunds to participants, withhold any requiredtaxes or penalties and, if required, report trans-actions to the Internal Revenue Service. Loans,

in contrast, create a series of ongoing adminis-trative tasks extending over an indefinite timeperiod. The administering entity must arrange arepayment schedule, monitor compliance overmonths or years, deal with late payments ordefaults, and ultimately determine whether fail-ure to repay a loan constitutes a withdrawal,which must be reported to the Internal RevenueService to assess taxes and penalties. Indeed,experts advise private employers to weighadministrative burden before adding loans totheir 401(k) plans, as shown by the questions inBox 5-3.

Reasons for Access: Hardship, Investment, orAny Reason

If either loans or withdrawals were allowed,access could be restricted to only certain purpos-es, such as buying a home or investing in highereducation, or access could be allowed only fordocumented emergencies or hardship. At theother extreme, early access could be allowed forany purpose.

Hardship. Private 401(k) plans allow access toaccount funds for financial hardship. Employers,as plan sponsors, serve as gatekeepers to deter-mine that hardship actually exists, typically fol-lowing the “safe harbor” guidelines establishedby the Internal Revenue Service. Those guide-lines stipulate that the account holder mustestablish that the withdrawal is essential to meetan immediate and heavy financial need. Safeharbor hardship withdrawals include: (a)Certain medical expenses for the account holder,the account holder’s spouse or dependents; (b)Purchase of a primary residence (excludingmortgage payment); (c) Payments of certainpost-secondary education expenses over the nextyear for the account holder, the account holder’sspouse, or dependents; or (d) To prevent evictionfrom or foreclosure on a primary home.

Human Capital Investments. Individual accountscould also permit the use of funds for variousincome-generating or asset-building purposes.Individual development accounts, for example,permit participants to access subsidized matched


savings only for higher education expenses, first-time homebuyer costs, and business capitaliza-tion or expansion costs. Likewise, IRAs allow

penalty-free withdrawals for the purchase of afirst home, as well as for higher education andmedical expenses.

Box 5-3. Check List for Plan Sponsors before Offering Loans from 401(k) Plans

1. How will the loan provision be communicated to plan participants?

2. How will your plan handle the Department of Treasury requirement that participants must borrow themaximum amount from their 401(k) plans before hardship withdrawals are permitted?

3. Will loans be permitted for any reason, or for only specified purposes such as buying a home? Whowill decide whether the loan has been requested for a valid purpose?

4. For loans that are used to buy a home, will the maximum loan term be limited to five years?

5. Will more than one loan be permitted per participant? If so, how many?

6. What participant collateral will secure loans?

7. Who will be responsible for the loan modeling? Who will prepare the amortization schedule?

8. Who will prepare the promissory note and other loan documents?

9. Will loans be repaid only through payroll deductions?

10. How is the loan payment schedule affected if the participant is temporarily laid off or on an unpaidleave of absence?

11. What happens when participants who have outstanding loan balances leave the company? If theseparticipants leave their 401(k) account balances in your plan, will they be kept on a monthly loan pay-ment schedule or will there be automatic acceleration? Will these participants be permitted to takeout new loans?

12. How will the interest rate be established and will it be fixed or variable?

13. How often will the interest rate for new loans be adjusted?

14. Will each loan be considered an asset assigned to the participant’s account, or will loans be combinedinto a general loan fund?

15. Will loans be initiated through manual paperwork, or after participants call a voice response unit?

16. How will loan payments be invested in the plan?

17. Will participants be able to select the investment fund(s) from which the loan withdrawal will bemade, or will plan provisions dictate a withdrawal priority order?

18. From which contribution sources, such as employee pre-tax and employer matching funds, and inwhat order, will the borrowed money be taken?

19. Will participants be charged a fee for setting up the loan in addition to the annual administrative fee?

Source: Benna, 1997. Helping Employees Achieve Retirement Security: Features Answers to the Most Often Asked 401(k) Questions


Unrestricted Access. From the perspective ofboth participants and account administrators,permitting access for any reason is much easierto administer because it does not require evi-dence or determination that the situation match-es the criteria. The access rules governing IRAscould provide one model for a regime of unre-stricted pre-retirement access to individualaccounts.

Segmented Account Funds from DifferentSources

Early access rules could also differ depending onwhether the participant, employer, or govern-ment generated the funds. For example, 401(k)plan rules differ between the employee’s owncontributions and the employer’s matching con-tributions. Similarly, individual developmentaccounts have different access rules for thesaver’s own contributions and for the matchingfunds. If individual accounts contain funds fromdifferent sources, it may be necessary to spellout separate access rules for each source offunds.

Other Limits on Loans or Withdrawals

Policymakers would have other means for limit-ing (or expanding) early access through the rulesof the account system. Spousal consent or noticemight be required (see Chapter Six). Technicaldecisions about the terms and amounts of loansand withdrawals could also be used to help meetany intended balance between access and fundpreservation.

If withdrawals were permitted, policymakerswould have to decide how much could be with-drawn. Withdrawals could be restricted to a cer-tain percentage of the account balance orpermitted only once the balance had reached aset threshold dollar level. For example, someproposals permit withdrawals only when theaccount is sufficient to produce a life annuityabove the poverty threshold. (See Chapter Threefor further discussion of the poverty threshold inthe context of retirement income.)

If loans were allowed, a more complex set ofdecisions about the loan terms and conditionswould have to be made, such as:

Loan ceilings. The system could fix a standardlimit—for example, no more than the lesser of50 percent of the account balance or $50,000—on the amount of total outstanding loansallowed to any participant. Alternatively, thepercentage and dollar caps could vary with thesize or age of the account.

Interest rate policy. The interest rate on individ-ual account loans could be designed to encour-age, or discourage, borrowing from individualaccounts as opposed to credit cards, bank loans,or home equity loans. A low interest rate wouldmake the loan more affordable and wouldencourage borrowing from the account ratherthan from other commercial lenders. A lowinterest rate would also result in lower interestpayments back to the individual’s account.

Limit on duration of loan. The repayment peri-od could be limited, following the example ofTSP loans (must be repaid in four years) and401(k) loans (repayment in five years).

Limit on number of loans. The system couldlimit the number of outstanding loans or thenumber of new loans allowed per participanteach year.

Death or disability. Forgiveness or special treat-ment of an outstanding loan could be consideredin the event of death or disability.

Finally, loan policies would need to specify theconditions under which failure to repay a loanwould constitute a withdrawal. In an individualaccount system modeled after the current TSP ora 401(k) plan, the consequences of converting aloan into a withdrawal would be taxes andpenalties on the unpaid balance.


Enforcing and Sustaining Restrictions Once decisions were made concerning pre-retire-ment access to individual account funds, itwould be necessary to determine what entitywould be responsible for enforcing the system’sregulations.

Finding a Gatekeeper: Who Bears the Risk ofDecision-Making, and With WhatConsequences?

If account holders could access their funds onlyunder certain conditions, a gatekeeper would beneeded to determine whether a particular with-drawal meets the requirements. If the gatekeeperwere expected to deny some requests (for exam-ple, in cases of failure to demonstrate hardship),procedures would likely be needed for reviewand appeal of the gatekeeper’s decision.

What incentive could (or should) be used tomotivate the gatekeeper to make accurate deci-sions? In particular, what incentives might beneeded to discourage wrongful withdrawals?What would be the penalty, and on whomwould it fall? Existing savings vehicles offersome precedents.

Individual Retirement Accounts. IRAs set norestrictions on withdrawals, but some are sub-ject to penalty taxes. For example, the accountholder must pay a 10 percent surtax on with-drawals before age 591/2 and a 50 percent surtaxon insufficient withdrawals after age 701/2. Thefinancial institution is responsible for informingthe IRS of the withdrawal, but the penalty fallson the account holder.

401(k) Plans. The fiduciary in charge of the401(k) plan, usually the plan sponsor (employ-er), is responsible for deciding whether employ-ees’ withdrawals or loans meet IRS and planrules and, if so, for notifying IRS of any with-drawals taken. In theory, the government has ablunt means of punishing and deterring wrong-ful determinations: the loss of tax-favored statusfor the entire retirement plan. But the employercan avoid that fate by reporting the violation tothe IRS and paying a (usually small) penalty.

IDA Experience. Community-based organiza-tions that receive federal or foundation grantsare responsible for approving a particular use ofIDA matching funds. Only after the purchase isapproved are the funds released. If a participantwithdraws funds for an unapproved purpose, heor she forfeits the matching funds. During recentefforts to authorize a federal tax credit to finan-cial institutions that administer and match IDAs,questions of accountability and legal liabilitybecame paramount. Specifically, should thefinancial institution, its contractual partners(typically nonprofit organizations) who helpadminister the program, or the IDA accountholders bear the ultimate legal responsibility forusing the IDAs in accordance with the law? Thisquestion was thoroughly discussed by policy-makers, financial institutions, and others butwas not resolved. This suggests that similarissues may arise in efforts to find an appropriate“gatekeeper” for restricting withdrawals from anew individual account system.

None of these three precedents appear to bepromising models for enforcing restricted accessto a national individual account system. TheIRA model allows unlimited access, as long asthe account holder pays a 10 percent penalty.The 401(k) model relies on employers forenforcement, but it is not clear that employerswould be willing or able to serve as gatekeepersfor a mandatory, universal system. Whileemployers who already have 401(k) plans mightbe able to serve this function, many others lackthe capacity to manage 401(k) plans. Still othershave highly mobile and transitory work forcesthat would make gatekeeping far more difficultthan it would be for a large employer with a sta-ble work force. The IDA model that relies oncommunity-based organizations as gatekeepersis not a cost effective model for a national pro-gram, and IDA proposals to place financial insti-tutions in a gatekeeper role were met with someresistance.

In brief, the question of who would serve asgatekeeper and what would motivate the gate-keeper to do its job well is complex. If the over-


all purpose of the accounts is retirement incomesecurity, then punitive treatment of the accountholder for wrongful withdrawal seems contraryto the purpose of the overall scheme. If theaccounts were held in private financial institu-tions, could those institutions be made gatekeep-ers? What incentives – carrots or sticks – wouldmake financial institutions diligent enforcers ofwithdrawal restrictions? If the government werethe gatekeeper, would this arrangement requirethat the government also be the holder of theaccounts? Such questions merit further attentionif restrictions on access to retirement funds arecontemplated.

Sustaining Restrictions on Access

As a matter of political reality, a ban or signifi-cant restrictions on pre-retirement access toaccount funds might not be popular. The UnitedStates has no precedent for a total ban on accessto individually-owned retirement savings. If poli-cymakers do decide to put such limits in place,significant pressures could be placed onCongress to ease access to retirement fundswhen deserving and sympathetic claims aremade, such as preventing eviction or paying formedical emergencies. As discussed further inChapter Seven, this is a central issue in designingthe appropriate policy when account holdersbecome sufficiently disabled that they qualify forSocial Security disability benefits.

The federal laws governing IRAs and the federalemployees’ TSP have, in recent years, expandedearly access to account funds. When IRAs werecreated in 1974, participants faced a 10 percentpenalty tax on withdrawals before age 591/2.The Health Insurance Portability and Accounta-bility Act of 1996 waived that penalty for with-drawals used to pay for medical care and, a yearlater, the Taxpayer Relief Act of 1997 waivedthe penalty on withdrawals to pay for highereducation and a first home. Similarly, when theTSP was first enacted in 1984, it allowed onlyloans and only for specified purposes; amend-ments in 1996 expanded access to loans for any purpose and permitted withdrawals forhardship.

This history has led the Joint Committee onTaxation (1999) to emphasize, in describing pos-sible systems of individual accounts, that nomatter what rules policymakers originallydecide, political pressures will build for pre-retirement access:

As a practical matter, it may be difficult torestrict access to private accounts. Pressuresunder present law have resulted in numerousexceptions to the 10-percent early withdraw-al tax for IRAs and, in some cases, qualifiedplan distributions. For example, the 10-per-cent early withdrawal tax does not apply toIRA withdrawals for education, medical,and first-time homebuyer expenses. In addi-tion, many employer-sponsored retirementplans permit individuals to borrow fromtheir retirement accounts. The ability tomaintain withdrawal restrictions maydepend in part on how individuals view theprivate accounts. If they are viewed as partof Social Security or as a pension benefit,then they may be more willing to not haveaccess to funds prior to retirement, as that isconsistent with the way in which SocialSecurity works today (as well as privatedefined-benefit pension plans). On the otherhand, if individuals view private accountsmore as personal savings, or as supplementsto other retirement income, they may bemore opposed to restrictions on pre-retire-ment access to funds.

Given the novel legal and political questionsraised by a national individual account system,and the history of the federal TSP and IRAs,restricting access to funds that account holdersconsider their own money is likely to be a diffi-cult challenge.

Impact of Access Rules on Third PartyClaims and Means-Tested BenefitsAccess can be a two-edged sword. Under exist-ing law, the extent to which individuals canreach into their retirement accounts sometimesaffects whether others can make a claim on thefunds as well. Account holders’ access can also


influence how the asset is treated in means-test-ed programs.

Claims of Third Parties

The more the account holder has access to theaccount, the more likely it is that creditors, suchas the IRS for overdue taxes or creditors inbankruptcy filings, can make a claim on theaccount as well. Ex-spouses or parents owedchild or spousal support, or entitled to a shareof marital property, may make claims againstretirement plans, but procedures are simpler foraccounts to which account holders have easieraccess. If policymakers wish to protect accountsfrom creditors, specific provisions could be putinto law.

Bankruptcy. Under federal law, assets held inmost retirement plans are usually not subject tothe claims of creditors, garnishment, or assess-ments and are not included in bankruptcy pro-ceedings. Some courts have ruled that federalbankruptcy protections are not applicable toIRAs because liberal access rules make themmore like a savings account than a traditionalpension plan; other courts have disagreed. TheU.S. Supreme Court heard arguments on thisissue in the case, Rousey v. Jacoway, 03-1407,in November and a ruling is expected by July2005 (Yen, 2004). Assets in IRAs may receiveprotection under state bankruptcy laws.

Child and Spousal Claims. A QualifiedDomestic Relations Order (QDRO) is requiredto obtain child or spousal support, or a share ofmarital property, from defined-benefit and 401(k)-type plans, but is not required to obtain ashare of an IRA.

Overdue Federal Taxes. The IRS may place alevy against most retirement plans and IRAs forunpaid taxes.

Access and Means-Tested Benefits

To qualify for means-tested public benefits suchas food stamps or Medicaid, applicants general-ly must demonstrate limited financial assets andincome. These asset tests exclude some

resources, such as defined-benefit pensions, butoften count assets in defined-contribution plansand IRAs toward allowable limits. So, low-income workers often must draw down, orspend, most or all of the balance in theirdefined-contribution accounts or IRAs (regard-less of early withdrawal penalties or other taxconsequences) before they can qualify formeans-tested programs. The rules for the majorfederal benefit programs are discussed below.

Medicaid. The Medicaid asset test was a federalstandard tied to the former Aid to Families withDependent Children (AFDC) program test.3

Today, eligibility determinations for Medicaidhave been devolved to the states. Some 22 stateshave eliminated asset tests for low-income fami-lies with children, and all but a few states haveeliminated asset tests for children. Most, but notall, states that do have asset tests for familiescount funds in defined-contribution plans andIRAs against the asset limits. In addition, moststates apply asset tests in determining eligibilityfor Medicaid for individuals age 65 and olderand those receiving disability benefits. For theseindividuals, Medicaid asset tests are often linkedto the asset test used in the federal SupplementalSecurity Income program; the SSI asset testrequires that defined-contribution accounts andIRAs be counted. States can make alterations,however, and some states use different Medicaidasset tests for different categories of elderly per-sons and people with disabilities.

Food Stamps. The food stamp rules generallyrequire accessible assets to count toward theprogram’s asset test. That standard is commonlyunderstood to mean that defined-contributionplans and IRAs—which are generally accessibleprior to retirement, albeit at the cost of a signifi-cant tax penalty—are counted. There is anexception for assets held in 401(k) and 403(b)accounts; they are not counted. (IRAs andKeogh accounts are counted.) A provision in theFarm Security and Rural Investment Act of 2002(Section 4107 of P.L. 107-171) allows states toconform the definition of assets used in theFood Stamp Program to the definition in TANF-


funded cash assistance or the Medicaid familycategory. Until final regulations are issued, it isunclear whether this provision will allow a stateto exclude IRAs from the food stamp asset testif the state excludes IRAs from counting asassets under TANF or Medicaid.

Supplemental Security Income. The federal gov-ernment sets the SSI asset test in which defined-benefit pension plans do not count against theprogram’s asset limit, but assets in defined-con-tribution plans do. The general principle is thatif the funds are accessible, they are countableassets.

Tuition Assistance. The rules governing access tostudent loan programs and Pell Grants excludebalances in 401(k) plans and IRAs from counta-ble assets.

Given the complexity of asset tests, it is possiblethat individual account balances could be count-able assets that would make a family ineligiblefor some means-tested programs, particularly ifthe account holder has access to the funds. Ablanket exemption already excludes funds inindividual development accounts from asset lim-its used in any means-tested programs.Policymakers might want to consider an analo-gous blanket exemption for any new nationalsystem of individual accounts so account holderswould not need to spend down their accountassets to qualify for Medicaid, food stamps, SSI,TANF, or tuition assistance programs.

Administrative IssuesThis chapter is based on the assumption that

any individual account proposal would desig-nate a central administrative entity to track con-tributions, balances, and withdrawals. In theabsence of such a centralized entity, pre-retire-ment withdrawals (along with rest of the indi-vidual account system) could prove difficult tomonitor.

Administrative issues pose a particular challengewith regard to loans, as discussed earlier. Theimpact of loans on account balances will depend

on the method of repayment. Loans from 401(k)plans have a very low default rate when repay-ment is made through payroll deductions, butloans generate considerable leakage from thesystem when borrowers leave their jobs.Typically, the outstanding loan is offset againstthe account balance and never restored to theaccount.

One approach to repayment of loans wouldrequire individual account participants to repayvia automatic payroll deductions and directdeposits from their current employers to theaccount system administrator (or its designee).Many employers already have direct depositarrangements that use a portion of an employ-ee’s paycheck to make mortgage or auto loanpayments. If individual account borrowers lefttheir employers, repayment could be made byautomatic debit from an account in a bank orother financial institution unless and until a newemployer assumed the tasks. Participants whoare unemployed or who are unable or unwillingto make such an arrangement with their employ-ers would be required to post collateral for anyloans. Requiring employers to offer this type ofloan repayment could place a significant burdenon the nation’s many small employers whomight not have automated payroll systems.

The feasibility of a large-scale, employment-based payroll deduction direct deposit systemfor repayment of individual account loans other-wise unrelated to employment has not been test-ed. Payroll deduction repayment in thesecircumstances might not work as effectively aswhen the employer sponsors the plan. So, theadministrative savings involved in employeradministration of loans are uncertain in the con-text of individual accounts. Repayment mecha-nisms that do not rely on the borrower’semployment status would have certain adminis-trative advantages, in that termination ofemployment would not trigger an account distri-bution, nor would it interrupt repayment of aloan.


Summary

The pros and cons of allowing early access toindividual accounts would depend, in large part,on the intended use of the accounts, whetherindividuals have any choice about participation,and the extent to which participants viewaccounts as personal property. If the accountsare supposed to provide baseline economic secu-rity in old age (similar to that now provided bySocial Security), there is a case for banningaccess to the funds before retirement. But in avoluntary system, allowing access to fundsmight encourage more people to participate andto save.

Precedents for early access rules for retirementaccounts include IRAs, 401(k)s, and the federalTSP. IRAs allow unlimited access as long asaccount holders pay taxes and, in certain cases,a 10 percent tax penalty on amounts with-drawn. Employer-sponsored 401(k)s permitmore limited access than IRAs, but employeescan often get their money if they need it—through a loan or hardship withdrawal, ifoffered by the employer, or by leaving the joband cashing out the account; such withdrawalsare taxable and usually subject to a 10 percenttax penalty. Most U.S. proposals that envisionindividual accounts as a partial replacement forSocial Security retirement benefits would banearly access to the money. Proposals that viewthe accounts as separate from Social Securitywould allow more liberal access.

Rules addressing the form and purpose of earlyaccess to retirement accounts would create ten-sions among three competing goals. Participantswould want access to their money when theyneed it, but the policy goal of providing basiceconomic security in retirement calls for mini-mizing leakage from the accounts. The goal ofretirement security argues for a total ban onearly access but, if access were allowed at all,concerns about retirement security would arguefor allowing only loans and only for hardship.The competing goal of administrative efficiencyargues for a total ban on access. As a second

choice, administrative efficiency points to mini-mal regulation– withdrawals for any reason –because loans (which involve repayments) andrestrictions on reasons for access (which requiredocumentation and determinations) requiremore administrative resources.

If access to individual accounts were allowedwith some restrictions, a gatekeeper would beneeded to determine whether a particular with-drawal meets the criteria—and procedureswould be needed to give participants an oppor-tunity to have a denial reconsidered. In 401(k)plans, the fiduciary in charge of the plan, usual-ly the plan sponsor (employer), is responsible fordeciding whether employees’ withdrawals orloans meet IRS and plan rules. The employerbears the risk of losing tax-favored status for theentire retirement plan in cases of wrongful deter-mination, although the IRS has procedures forlevying lesser penalties.

In a national system of individual accounts,what entity could play an analogous gatekeeperrole? What incentives would prompt the gate-keeper to prevent wrongful withdrawals? Whatmight the penalty be for noncompliance, and onwhom would it fall? If the overall purpose of theaccounts is retirement income security, a penaltyon the account holder for a wrongful withdraw-al may undermine the ultimate goal.

Early access to retirement funds can be a two-edged sword. Under existing laws, account hold-ers’ access to their own retirement fundssometimes also permits creditors to make aclaim on the funds in cases of bankruptcy orunpaid federal taxes. Moreover, some means-tested government benefit programs treat acces-sible retirement funds as assets for the purposesof determining benefit eligibility; in such cases, ifthe account holder has access to the funds, he orshe must spend down the account balance inorder to qualify for means-tested assistance.

Despite these potential drawbacks to pre-retire-ment individual account access, the politicalreality remains that there might be significant


pressure to allow individuals to withdraw theirfunds before retirement. No U.S. precedentexists for a total ban on access to individually-owned retirement savings and even if policy-makers create such a ban, the history of IRAsand the federal TSP shows that Congress might

experience significant pressure to ease legislativerestrictions on access to retirement savings. So,no matter what the policy chosen at the outset,sustaining restrictions on access to retirementfunds that account holders view as their ownmoney is likely to be an ongoing challenge.

Chapter Five Endnotes

1 Funders of the American Dream Demonstrationinclude the Ford Foundation, the Charles StewartMott Foundation, the Joyce Foundation, the F.B.Heron Foundation, the John D. and Catherine T.MacArthur Foundation, Citigroup Foundation,Fannie Mae Foundation, Levi StraussFoundation, Ewing Marion KauffmanFoundation, the Rockefeller Foundation, and theMoriah Fund.

2 The TSP for participants in the FederalEmployees Retirement System requires spousalconsent for loans and withdrawals. The TSP forparticipants in the Civil Service RetirementSystem requires spousal notice.

3 AFDC was replaced by the Temporary Assistanceto Needy Families (TANF) program under legis-lation enacted in 1996.


Figure 5-A. Federal Rules and Tax Treatment of Withdrawals and Loans from IRAs, 401(k)s, and the Thrift Savings Plan

Type of plan Withdrawals before Retirement Loans

Appendix: Precedents for Early Access

Retirement savings plans have different early access rules. Table 5-A provides further detail on therules for early access to funds in Individual Retirement Accounts, 401(k) plans, and the Thrift SavingsPlan for federal employees.

Rules: Withdrawals are permitted at any age.Taxation: All withdrawals are subject to federal income tax.Those taken before age 591/2 are subject to an additional 10percent penalty, unless: (a) The owner has died and the account passes to a beneficiary; (b) The owner has become disabled; (c) The funds are used to pay non-reimbursed medical expenses

that exceed 7.5 percent of adjusted gross income; (d) The funds are used to pay a federal tax liability; (e) The withdrawals are part of a series of “substantially equal

periodic payments” made over the lifetime or life expectancyof the account holder (or the holder and designated beneficiary);

(f) The funds, up to $10,000, are used for purchasing a firsthome; or

(g) The funds are used to pay health insurance premiums whileunemployed for at least 12 weeks.

Rules: Withdrawals are permitted at any age.Taxation: Contributions withdrawn are not taxable. Earningsheld in a Roth IRA for at least five years are not taxable if takenafter age 591/2, death, disability, or for purchase of first home.Earnings taken for other purposes, or those held less than fiveyears, are subject to federal income tax and a 10 percent penal-ty on the same terms as for traditional IRAs.

Rules: Employers who sponsor 401(k) plans generally cannotallow employees to withdraw their own contributions until theemployee: (a) reaches age 591/2; (b) terminates employment; (c)dies, (d) becomes disabled, or (e) experiences financialhardship.i

Departing workers can: (a) leave the funds in the planii (if lessthan $5,000, the sponsor can roll it over or cash it out); or (b)take a distribution.Taxation: Funds withdrawn by departing workers that are notrolled over into another tax-deferred retirement plan are subjectto federal income taxes on the distribution, including 20 per-cent income tax withholding. An additional 10 percent penalty

No loans allowed

No loans allowed

Plans may offer loansunder certain condi-tions, but are notrequired to do so.

Federal rules permitloans up to the lesserof $50,000 or half ofthe employee’s vested401(k) account. Theloan must be repaid infive years, unless it is

Individual RetirementAccounts (regular)Contributions are taxdeductible

Roth IRAsContributions fromafter-tax income

401(k) plansEmployee contributionsnot taxable


Figure 5-A. Federal Rules and Tax Treatment of Withdrawals and Loans from IRAs, 401(k)s, and the Thrift Savings Plan (continued)


applies unless:(a) The worker is age 591/2;(b) The worker has terminated employment with the plan’s

sponsor after reaching age 55;(c) The worker has died and the account passes to a

beneficiary;(d) The worker has become disabled;(e) The distribution is made to a former spouse or child under a

Qualified Domestic Relations Order incident to a divorce; (f) The funds are used to pay non-reimbursed medical expenses

that exceed 7.5 percent of adjusted gross income;(g) The funds are used to pay a federal tax liability; or(h) The withdrawals are part of a series of “substantially

equal periodic payments” made over the lifetime or lifeexpectancy of the account owner (or the owner and desig-nated beneficiary).

Rules: 401(k) plans generally are permitted to allow employeesto withdraw most types of employer contributions duringemployment, if the employee has at least five years of participa-tion in the plan or if the contribution has been in the plan forat least two years.

Plans can impose more stringent withdrawal restrictions, andcertain types of employer contributions must be subject torestrictions similar to those applicable to employee pre-tax contributions.

used to buy a home.Amortization must belevel over the term ofthe loan, and pay-ments must be madeat least quarterly.Interest is payable atcommercial rates setby the plan.

Non-repaid loansbecome withdrawalsand are subject toincome taxes and a10 percent tax penaltyunless one of thepenalty exceptions ondeparting workerwithdrawals applies.

Otherwise, the 10percent tax penaltyand the 20 percentincome tax withhold-ing do not apply toloans.

Same loan provisionsas for employee contributions.

401(k) planscontinued

401(k) plansEmployer matchingcontributions


Figure 5-A. Federal Rules and Tax Treatment of Withdrawals and Loans from IRAs, 401(k)s, and the Thrift Savings Plan (continued)


Until 1997, in-service withdrawals were not permitted for anyreason.

Rules: Since 1997, in-service withdrawals are permitted in twocases: (a) employees who can prove “financial hardship,” and(b) an employee age 591/2 or older can take a one-time with-drawal for any reason. Spousal consent required for FERS TSPparticipants; spousal notice for CSRS TSP.

Taxation: Either type of withdrawal is subject to federal incometaxes (and 20 percent withholding). Financial hardship with-drawals are also subject to a 10 percent tax penalty. (The avail-ability of general purpose loans eased the demand for earlywithdrawals.)

Until 1997, loanswere permitted onlyfor specified reasons,including financialhardship.

Since 1997, two typesof loans are offered.General purpose loansbetween $1,000 and$50,000 (but limitedto the account bal-ance) must be repaidin 4 years. Residentialloans must be repaidwithin 15 years. Theinterest rate is thatpaid on the G-fund, afund of treasury secu-rities offered by theTSP. Loans are repaidthrough payrolldeductions. Spousalconsent required forFERS TSP participants;spousal notice forCSRS TSP.

Thrift Savings PlanFederal worker, non-taxed contributions

i 401(k) plans can permit employees to take “hardship” withdrawals of their own contributions. The worker must show an “immediate andheavy financial need” such as medical costs, the purchase of a principal residence, college tuition, and expenditures to avoid eviction from aprincipal residence. Participants must also have exhausted all other withdrawal options and non-taxable loans, and they must suspend all newcontributions to retirement plans for at least six months. The hardship withdrawal may not be rolled over into another employer plan or IRA.

ii Unless a departing worker asks for a cash withdrawal or directs the plan to roll over the funds to a specified IRA, pension rules require thatcompanies roll over accounts of between $1,000 and $5,000 to an IRA established by the company for the worker. These requirements willnot become effective until the Department of Labor issues regulations interpreting them.


6Chapter

Spousal Rights

117

Amid all the work on creating individual SocialSecurity accounts for retirees, policymakers havepaid less attention to individuals whose eligibili-ty for Social Security benefits is based on familyrelationship or disability. This is an importantoversight, as about half of all beneficiaries havetheir eligibility based on family status or disabil-ity. This chapter focuses on the rights of spous-es. The next two chapters focus on disabledworkers and children, respectively.

This chapter reviews spousal rights precedentsfound in Social Security, in state family law, andin federal rules that apply to retirement plansother than Social Security. It also explains a keydecision policymakers must make on individualaccounts—whether to establish uniform federalpolicies concerning spousal rights or leave thosepolicies to be resolved by state law. Spousalrights are considered in the context of otherindividual account features, such as whetherparticipation is voluntary or mandatory and thegoals and purposes of the plan. It explores ten-sions and tradeoffs in blending the spousal pro-tections in Social Security with the propertyconcepts inherent in individual accounts. Thechapter also examines new information and pro-

cedures that might be required to implementspousal rights.

Precedents for Spousal Rights

Rules that define spousal protections in individ-ual accounts could be based on precedents inSocial Security law, in state family law, or in theretirement plans that currently supplementSocial Security. New concepts have been putforth in proposals for individual accounts aspart of Social Security reform.

Social Security Benefits for SpousesSocial Security benefits are designed to providebaseline income security for a spouse when aworker’s earnings end due to retirement, disabil-ity, or death. The wife or husband of a retired ordisabled worker is eligible to receive an amountequal to 50 percent of the worker’s benefit; anelderly widow or widower, an amount up to 100percent of the worker’s benefit (providing noearly retirement reductions apply). Disabledwidowed spouses may claim benefits equal to71.5 percent of the deceased worker’s benefitstarting at age 50 (non-disabled widowed spous-


es may claim benefits starting at age 60).Spouses’ benefits continue to be paid for life andare annually indexed for inflation. A spousalbenefit is paid only to the extent that it exceedsthe benefit that would be paid to the spousebased on his or her own work record. Forexample, if Jane’s own benefit as a retired work-er is $700 and her benefit as a widow is $850,she would receive her $700 benefit plus $150 asa widow for a total of $850. Benefits are paid tohusbands and widowers on the same terms asbenefits paid to wives and widows, but becausemen rarely earn less than their wives over theirlifetimes, few men receive benefits as husbandsor widowers.1 While this chapter speaks of“wife” or “widow” benefits, the same benefitsare payable to and the same issues arise with

respect to benefits payable to husbands and widowers.

Of the nearly 46 million individuals receivingSocial Security benefits, only about half receivebenefits solely as retired workers, while theother half receive benefits based on disability oron family relationships to a worker (Figure 6-1).About 12 percent of Social Security recipientsare disabled-worker beneficiaries and 8 percentreceive benefits as children of workers.

In all, about 14 million individuals – 30 percentof all beneficiaries – receive Social Security bene-fits based at least in part on a spouse’s workrecord. These beneficiaries are overwhelminglywomen. About 6.0 million women are entitledto Social Security as workers and to higher ben-

Figure 6-1. Number of Beneficiaries by Basis for Entitlement, December 2001

Number PercentType of Benefit (thousands) of Total

Total 45,874 100.0Entitled based on own work record only 28,031 61.1

Retired worker only 22,766 49.6Disabled worker 5,265 11.5

Dually entitled to a higher benefit as a family member 6,076 13.2Dually entitled retired worker and spouse of deceased worker 3,467 7.5 Dually entitled retired worker and spouse of retired or disabled worker 2,609 5.7

Entitled only as a family member 11,767 25.7(i.) Of deceased worker 6,915 15.0

Non-disabled widow(er)s 4,625 10.1Children 1,890 4.1Widowed mothers and fathers 195 0.4Disabled widow(er)s 202 0.4Parents 3 0.0

(ii.) Of retired worker 3,205 7.0Spouses, age 62 or older 2,738 6.0Children 467 1.0

(iii.) Of disabled worker 1,647 3.5Spouses 157 0.3Children 1,490 3.2

Note: Children include children under age 18, students under age 19 and adults disabled since childhood (before age 22) who receivebenefits on the account of a deceased, disabled or retired parent.

Source: U.S. Social Security Administration, 2002a. Annual Statistical Supplement to the Social Security Bulletin, 2002

Chapter Six: Spousal Rights 119

efits as a widow, wife, or divorced wife. Another7.8 million women receive Social Security solelyas widows, wives or divorced wives.

Benefits for divorced wives are calculated thesame way as benefits for wives and widows.2

To be eligible for Social Security benefits as adivorced wife or widow, the marriage must havelasted at least 10 years and the individual mustbe unmarried when applying for benefits. As ofDecember 2001, about 475,000 individualsreceived benefits solely as divorced spouses orsurviving divorced spouses and an additional490,000 women were dually entitled to retire-ment benefits from their own earnings as well asa secondary benefit as divorced wives or widows(U.S. SSA, 2003b). Taken together, almost onemillion women age 62 or older receive SocialSecurity benefits based in full or in part on thework record of an ex-spouse. They account forjust less than 5 percent of the 21.4 millionwomen age 62 and older who receive SocialSecurity benefits.

As more women work in paid employment,more women will be eligible for benefits basedon their own covered work records and fewerwill receive benefits solely as wives or widows.At the same time, more women will be duallyentitled. That is, they will be eligible both for aretired-worker benefit based on their own cov-ered earnings and for a higher benefit as a wife

or widow. By 2040, the proportion of womenwho will qualify for benefits only as wives orwidows is projected to decline from 34 percentto 7 percent, while those who receive benefitssolely as retired workers will increase from 38percent to 55 percent (Figure 6-2). The share ofwomen dually entitled as workers and widows isestimated to rise from 16 percent to 28 percent(Favreault and Sammartino, 2002). The share ofwomen dually entitled as wives will decline asmore women’s own work records provide a ben-efit equal to at least half of their husband’s ben-efit; the percentage of dually entitled widowswill increase because the benefit on their workrecord will still be lower than their husbands.For most dually entitled widows, the top-upfrom the secondary benefit will be significant:20 percent to 69 percent of their total benefit, or$300 to $1,000 per month in 1998 dollars.

Because the dual entitlement rule limits paymentof spousal benefits in two-earner couples, exist-ing spousal protections pay more relative to pastearnings and contributions to one-earner couplesthan to two-earner couples with the same com-bined earnings. In addition, because spousalbenefits are based on the retired-worker benefitsof the primary beneficiary, the higher replace-ment of lower average earnings than of higheraverage earnings is duplicated in spousal bene-fits. This redistribution occurs in the Social

Figure 6-2. Type of Social Security Benefits Received, 2000 and 2040Women Age 62 and Older (Percent distribution)

Type of Benefit 2000 2040

Total – All Benefit Types 100 100Insured on own record as retired worker 66 93

Entitled as worker only 38 55Dually entitled 28 38

Wife 12 10Widow 16 28

Entitled only as spouse 34 7

Wife 13 3Widow 22 4

Source: Favreault and Sammartino, 2002. “The Impact of Social Security Reform on Low-Income and Older Women”


Security system because benefits are based on aformula that varies with average covered earn-ings. It would be more difficult in an individualaccount system, where funds reside in personalaccounts rather than a pool of federal tax revenue.

Family Law Precedents for SpousalRights If individual accounts were considered property,then state laws on spousal property rights mightdetermine account disposition during marriage,at divorce, and/or when the account holder dies.

All but nine states follow English common lawprinciples. The other nine states follow commu-nity property rules derived largely from Spanishcivil law, in which wives and husbands have ahalf interest in all property accumulated during marriage.

Over time, common law states began to incor-porate community property principles by enact-ing laws that called for equitable distribution ofproperty at divorce. Before these laws werepassed, wives were entitled only to the propertyto which they held title. Today, all common lawstates have adopted equitable principles for divi-sion of property at divorce.

States still vary significantly in how they treatmarried couples. As family structures havegrown more complex (children from multiplemarriages, for instance), states have adoptedvarying solutions to resolve issues presented bycontemporary family life. Most states—bothcommon law and community property—allowstate rules on property rights to be overriddenby a contract that is mutually and fairly agreedto by the husband and wife before or during amarriage or at divorce.

Community Property States

About 29 percent of all Americans live in thenine community property states—Arizona,California, Idaho, Louisiana, Nevada, NewMexico, Texas, Washington, and Wisconsin. Inthese states, all property acquired during a mar-riage is held jointly by the spouses, regardless of

whose name is on the deed or title. Propertyacquired before the marriage and inheritancesduring the marriage are considered separateproperty, although the increase in the value ofthis property during marriage may be consideredmarital property (unless the asset is clearlydeclared and separately held as individual prop-erty). The value of property, including retire-ment accounts, acquired prior to residence in acommunity property state (what is sometimeslabeled the “determination date”) may be con-sidered individual property, but subsequent con-tributions and increases in value to thoseaccounts will be marital property. So, moves bymarried couples between community propertyand common law states will result in mixedproperty that is only partially community property.

Control of community property during mar-riage. Most community property states alloweither the husband or the wife to singly managethe community property. If title is in the name ofonly one spouse, for example, only that spousemay be able to manage the property. If a spouseis operating a business that is community prop-erty, exclusive control of the business may begiven to that spouse. However, a spouse whomanages community property generally has afiduciary responsibility to protect the interests ofthe other spouse. Most community propertystates require that both spouses agree to trans-fers or mortgages of community real property(Dukeminier and Krier, 2002). As a practicalmatter, enforcement of state laws on manage-ment or control of community property general-ly becomes an issue only in cases of maritaldispute.

Community property at divorce. In general, atdivorce each spouse is entitled to half of themarital or community assets, and each spouseretains his or her separate property that is out-side the community assets.

Community property at death. When a spousedies in a community property state, the widowor widower automatically retains his or her sep-


arate personal property and his or her half ofthe marital community property. At issue iswhether the widowed spouse has a claim to thedecedent’s half of the marital property or thedecedent’s separate property. This disposition isdetermined by the decedent’s will. In the past, inthe absence of a will, a widowed spouse wouldtypically get only his or her half of the commu-nity property and the children would get thedecedent’s half; most states now give the wid-owed spouse all of the marital property if thepartner dies intestate, although children of thedeceased from a former marriage may also havean automatic claim to a share of the maritalproperty. Divorce and remarriage increase theprobability that marital property will be dividedbetween a current widowed spouse and childrenfrom a former marriage.

Common Law States

In common law states, property belongs exclu-sively to the spouse who holds title. Spousesmay, of course, hold joint title, and such anarrangement is particularly important in thesestates because courts may look at the title whendistributing assets at divorce.

Common law property at divorce. Traditionally,marital property remained the property of thespouse holding the title at divorce. A wife wasentitled to alimony as long as she was not atfault for the divorce. Today, common lawdivorces look more and more like those in com-munity property states. All common law statesenacted equitable distribution of property laws,with these laws calling for allocation of maritalassets according to equitable principles. Theseprinciples vary considerably from state to stateand some states require a determination of faultto help guide the equitable division concept.

Common law property at widowhood. Widowswere not, historically, entitled to property, onlyto upkeep. Today, if a spouse dies without awill, common law states generally stipulate thata share of the decedent’s property goes to thewidowed spouse. Even if there is a will, thedecedent cannot unilaterally disinherit a spouse.

Most common law states stipulate a minimumshare for the widowed spouse. If the will pro-vides less than this share, the widowed spousecan sue in state court to obtain that allotment.

Spousal Rights in Pensions andRetirement Savings AccountsSocial Security and state family law are not theonly precedents for spousal rights. Tax-favoredretirement plans established under federal lawalso offer a model for an individual account sys-tem. The appendix to this chapter summarizesspousal rights in these plans. Pensions andretirement savings programs include defined-benefit pensions, defined-contribution pensions,such as 401(k) plans and 403(b) plans, and indi-vidual retirement accounts (IRAs). Each definesthe rights of account holders and spouses differently.

Pension Plans

Defined-benefit pension plans and money pur-chase defined-contribution plans provide broadspousal rights that extend beyond the worker’sdeath and, in some cases, beyond divorce. Thedefault pension paid to a married worker by aprivate, employer-provided pension must be ajoint and (at least) 50 percent survivor annuity.A married participant who seeks to receive pen-sion benefits with a less generous or no paymentto the survivor must obtain the spouse’s nota-rized consent. The spouse must also consent toany loan that would use the benefit as collateral.A widowed spouse has a waivable entitlement toat least 50 percent of the annuity that wouldhave been paid to the deceased worker, whetherthe death occurs before or after retirement.Divorced spouses, by contrast, do not have anyautomatic rights, though at divorce they mayseek rights to a share of the account or benefit.This settlement must be expressed in a QualifiedDomestic Relations Order (QDRO). The extentof the divorced spouse’s rights varies from stateto state, depending upon whether a state usescommon law or community property rules.3


401(k) Plans

Defined-contribution plans such as a 401(k)provide fewer rights for spouses. At retirement,no spousal consent is needed if funds are takenas a lump sum (the typical payout) or in period-ic payments. Only if the plan offers annuitiesand the worker chooses one is spousal consentrequired for a payment other than a joint-and-survivor annuity. Account holders can take loansor withdrawals against their accounts withoutspousal consent. A divorced spouse may obtainsome portion of the account in a divorce pro-ceeding through a QDRO but he or she doesnot automatically receive a share. A widowedspouse does, however, have significant inheri-tance rights to account assets held in the plan.Unless the spouse had previously consented tothe participant naming another beneficiary, heor she will automatically receive the entireaccount balance as a death benefit.

Thrift Savings Plan

The Thrift Savings Plan (TSP), a 401(k)-typeplan for federal employees, has different rulesfor spousal protection.4 In general, spousal con-sent is needed before a worker can withdraw orborrow against the TSP account before retire-ment. As with 401(k) plans, a divorced spousemay seek some portion of the account in adivorce settlement, but he or she has no auto-matic share. Unlike 401(k) plans, the workercan name anyone as a death beneficiary; if nobeneficiary is named, the spouse is the defaultbeneficiary in the TSP line of succession. Likedefined-benefit pension and money purchasedefined-contribution plans—and unlike401(k)s—the TSP requires notarized spousalconsent before a worker can take a post-employ-ment withdrawal from the account other than inthe form of a joint-life annuity with a 50 percentsurvivor benefit.5

Individual Retirement Accounts

IRAs give the least protection to spouses and themost freedom to account holders. No spousalconsent is needed for pre-retirement with-drawals. In the event of divorce, spouses mayseek IRA funds as part of the settlement under

state law, but awards are not automatic. Theaccount holder can choose anyone as the deathbeneficiary.

Summary

The spousal benefit provisions in other pensionand retirement savings plans provide usefulcomparisons for the discussion of spousal rightsin individual accounts. It is important to note,however, that these employer-based pensionsand retirement savings account systems werecreated assuming the existing Social Security sys-tem. These benefits were intended to build onthe Social Security base—not to replace SocialSecurity in whole or in part. Should Congressdevelop individual accounts that would replacesome share of retired-worker or disability bene-fits, and alter the spousal safety net that SocialSecurity now provides, the rules regulating such accounts would acquire much greater significance.

Federal or State JurisdictionState laws already address spousal rights, butstates define these rights in a variety of ways. IfCongress establishes a system of individualaccounts, a key question is whether to have uni-form federal policies on spousal rights or leavethose policies to state jurisdiction.

Federal Legislation Can Trump or Defer toState Law

The Social Security program provides uniformbenefit rights throughout the country for spous-es, widowed spouses, and ex-spouses. At thesame time, the federal program relies on statelaw to determine whether a claimant is married,widowed, or divorced, with one exception—theDefense of Marriage Act of 1996 prohibits pay-ing spousal benefits from Social Security or vet-erans’ benefits to same sex couples, even if theyare married under state law.

Under the federal system of government, lawsthat govern property rights during marriage, atdivorce, and at the death of a spouse have tradi-tionally been the province of state courts andlegislatures. This historical delegation has pro-


duced marked differences in law and policyamong states. If policymakers wish to imple-ment uniform spousal rights under an individualaccount system, they will need to define therules explicitly in federal legislation. Absent afederal pronouncement, state courts and legisla-tures will make determinations about spousalprotection, the treatment of accounts at divorce,and account inheritance.

It is well established that when Congress createsa federal benefit program, it has the power tolegislate on matters that would otherwise fallwithin the province of the states. Despite thegeneral deference accorded to state jurisdictionin the area of family law, the Supreme Court hasheld that federal legislation preempts state lawwherever state jurisdiction would do “‘majordamage’ to ‘clear and substantial’ federal inter-ests.” For example, in Hisquierdo v. Hisquierdo,439 U.S. 572 (1979), the Supreme Court decid-ed that a California decision to award the wifean interest in the husband’s expected benefitsunder the Railroad Retirement Act of 1974 wasan impermissible conflict with the federal act,which provides benefits similar to those underthe Social Security Act. Similarly, in McCarty v.McCarty, 453 U.S. 210 (1981), the SupremeCourt decided that federal law precluded thestate court of California from dividing militarypensions upon divorce. In the following year,Congress enacted the Uniformed ServicesFormer Spouses’ Protection Act, Title 10 U.S.C.sec. 1408, which permits state courts to treatmilitary retired pay as property and divide itupon divorce.

The courts are likely to allow the federal gov-ernment substantial leeway in structuringspousal rights for individual accounts. However,these precedents suggest that no spousal protec-tions would be guaranteed unless specified infederal legislation on individual accounts.

Congress may decide to define spousal rights asa matter of federal law, or to defer to state law,or to do some of both.6 If Congress does notdeal with these issues, uncertainty about

whether these powers remain with the state orwere implicitly preempted by federal law wouldgenerate considerable litigation and inconsistentresults.

Pros and Cons of Federal Treatment ofSpousal Rights

The advantage of having national rules is clear:they ensure uniform treatment for all accountholders, no matter where they work or reside.Moreover, uniform rules can reduce costs byreducing the need for lawyers to represent therights of account holders and spouses and bysimplifying plan administration. But creatingfederal policy on spousal rights in an individualaccount system would require making difficultchoices because individual accounts represent afinite pool of assets—when one person receives ashare, another person loses. In contrast, spousalbenefit awards under Social Security do notreduce the worker’s benefit, as the cost of pro-viding family benefits is spread among all SocialSecurity participants.

State determination of spousal rights in individ-ual cases would not provide the uniformity oradministrative simplicity of federal rules. Butstate jurisdiction, arguably, might ensure moreequitable treatment for individuals. State courts,for example, routinely decide how to divide themarital property of divorcing spouses who havebeen unable to reach settlement. Other retire-ment accounts (such as IRAs and 401(k)s), arealready subject to division by state courts atdivorce. The advantage of this approach lies inits flexibility: one divorcing spouse might wantto exchange his or her right to a retirementaccount for the family home, while anotherdivorcing spouse who expects to live a long lifemight prefer the interest in a retirement account.State courts might arbitrate these disputes andsupervise settlements that better address themarital breakup circumstances.

At the same time, relying on state courts todetermine spousal rights presents a number ofchallenges, particularly in the context of an indi-vidual account plan designed to reach low- and


moderate-income people who may be unable toafford lawyers. It is common for people to getdivorced without the benefit of an attorney.Studies in individual states consistently find thatin most family law cases, one or both parties isunrepresented. In California, for example, two-thirds of petitioners in family law cases did nothave attorneys; by the time a divorce was grant-ed, 80 percent were unrepresented (JudicialCouncil of California, 2004). Likewise, aFlorida study found that about two-thirds ofdomestic relations cases began with at least oneparty unrepresented and the share of unrepre-sented parties grew by the time the cases ended(Greacen, 2002a). Of litigants filing for maritaldissolution in the state of Washington, 62 per-cent of those without children and 47 percent ofthose with children were unrepresented(Greacen, Administrative Office of the Courts,2002b). So, in determining the rules and proce-dures for dividing assets in an individualaccount at the time of divorce, it is important torecognize that at least one party often does nothave a lawyer. The unrepresented spouse couldbe in a very precarious position.

Framework for AnalyzingSpousal Rights

Rules for delineating spousal rights or spousalprotections for individual accounts would beinfluenced by the overall purpose of theaccounts, the level of Social Security benefitsthat accompany the accounts, and whether par-ticipation is mandatory or voluntary.

Purpose of AccountsThe suitability of spousal rights rules woulddepend on what role individual accounts areintended to play in relation to the Social Securitysystem. More extensive federal spousal rightswould be called for if individual accounts weredesigned to replace, in whole or in part, tradi-tional federal Social Security benefits—whichinclude spousal benefits. To the extent that indi-vidual accounts would supplement SocialSecurity, more flexibility could be considered foraccount holders and/or for the role of state law.

Level of Social Security BenefitsSimilarly, the level of traditional Social Securitybenefits that accompany an individual accountsystem could influence policymakers’ viewsabout spousal rights. The lower the level ofremaining Social Security benefits, the strongerthe case would be for requiring that accountsprovide some mandatory protections found inSocial Security. If traditional Social Security ben-efits are thought to provide an adequate baselineof retirement income, then more flexibility couldbe considered.

Voluntary or Mandatory Participation inAccounts If participation in an individual account systemwere voluntary, restrictive policies could deterparticipants. Individuals might be less interestedin participating if contributions must be sharedwith the spouse, if accounts are automaticallydivided at divorce, if bequests cannot be madeto beneficiaries other than the spouse, or if joint-and-survivor annuities must be purchased.

A voluntary individual account system also rais-es some specific challenges to designing andimplementing spousal rights policies. If partici-pation were voluntary, there would be no guar-antee that spouses would make the samedecision to participate. Indeed, many individualsenter the labor force and start contributing toSocial Security before they marry. Consequently,spouses might have made different decisionsabout participation. Even if the plan allowedindividuals to change their initial election aboutparticipation when they marry, efforts to requirespouses to make the same choice would face theproblem of multiple marriages. So, decisionswould have to be made about how spousalrights rules would apply if only one spouseelects to participate. Mandatory spousal rightsare not likely to achieve their intended purposein a voluntary participation system.

Worker-Specific Offsets and SpousalRightsA new set of spousal rights issues arise in pro-posals that would allow workers voluntarily to


shift part of the scheduled Social Security taxesto personal accounts. These proposals call forworker-specific offsets against the worker’sfuture benefits or account proceeds to compen-sate the Social Security Trust Fund, so as not tofurther jeopardize system solvency and to treatparticipants and non-participants in accountsequitably. How these offsets would interact withfuture benefit rights of the worker’s spouse, ex-spouse, or widowed spouse raise complex ques-tions, which are discussed in Chapter Nine.

Policy Choices for SpousalRights

Policymakers will need to decide how spousalrights would be addressed at different possibledistribution points: during marriage, at divorce,at death, or at retirement. Many differentanswers are suggested by various Social Securityproposals that call for individual accounts(Figure 6-3). We consider each life event in turn.

Spousal Rights during MarriageFederal policymakers would need to considerwhat spousal rights to recognize during mar-riage. For example, whether contributions toaccounts would be split equally between hus-bands and wives and whether spousal consentwould be required for pre-retirement loans orwithdrawals, if such options exist.

Contribution Splitting or Not

A contribution splitting approach would divideaccount deposits equally between the husbandand wife. Only contributions made during mar-riage would be divided in such a way—eachspouse would retain any contributions andinvestment earnings accumulated while single orfrom previous marriages. A contribution-split-ting approach would build community propertyprinciples into an individual account system.7

Contribution splitting is consistent with thepartnership view of marriage and it gives eachspouse a share of property to manage independ-ently. Such a partnership view has gainednational currency as women and men enter the

workforce in more nearly equal numbers. Unliketraditional Social Security, contribution splittingwould provide portable spousal rights forwomen who divorce after fewer than 10 years ofmarriage.

If contributions were divided between spouses,policymakers could give each spouse the abilityto singly control their own account.Policymakers could decide that accounts wouldnot be subject to division at divorce (since eachspouse had already received and invested theirshare of the marital earnings); spousal consentwould not be required for pre-retirement with-drawals, and each spouse could decide what todo with his or her own account at retirement.Another option is that policymakers couldestablish some additional spousal rights, such as the right to inherit the spouse’s account atwidowhood.

A key issue with contribution splitting would behow to get accurate and on-going marital statusreports so that contributions could be creditedto each account in a timely and accurate man-ner. If contribution splitting were mandatory,this would be particularly important in cases ofworkers not wanting to report their marital sta-tus. Administrative issues are discussed later inthis chapter.

Spousal Consent for Loans or Withdrawals

Contribution splitting would make each spouse’sdecisions about loans or withdrawals independ-ent. If contributions were not split, decisionswould have to be made about spousal-consentrules for account holders to withdraw or borrowagainst funds in their own individual accounts.The requirement of spousal consent would logi-cally depend on whether a spouse had a claimon the partner’s account at retirement, divorce,or widowhood. If the plan provided for dividingaccounts at divorce or for spousal survivorshiprights, then spouses would have a clear interestin preserving their partners’ accounts. If thespouse had no such claims, the case for spousalconsent would be reduced.


Figure 6-3. Spousal Protection Rules in Selected Individual Account Proposals

Treatment of Individual Account at Certain Events

During At At death At Use after divorce Proposal marriage divorce of worker retirement or widowhood


ACSS (Gramlich): Individual No split; Not specified Widowed spouse Married buy Surviving spouseAccount Plan, 1996 no access must inherit. joint-life must keep

before age 62 Unmarried worker inflation- acquired fundsnames any heir. indexed annuity for own

unless spouse retirement. Otherdeclines it. heirs spend it.


Reps. Kolbe-Stenholm: Not specified Not specified Account holder Not specified Not specifiedBipartisan Retirement names any heirSecurity Act of 2004


Ball, Social Security Plus, No split; Court can Worker can Open access Former spouse2003 IRA rules include in name any at retirement. can use

for access settlement death acquired funds.like IRA beneficiary


President’s Commission No split; 50-50 split of Widow(er) must Solvency Former spouse &(PCSSS) Models 1, 2, and 3 no early contributions inherit. estimates surviving spouse(2001) access made during Unmarried assumed must keep IA

marriage worker names married must for own retirement.any heir. buy joint and Other heirs can

2/3-survivor spend it.annuity (either CPI-indexed or variable)

Rep. Smith, Social Security 50-50 split; Divorce Funds go to Purchase of No provisionsSolvency Act of 2003 no early terminates worker’s estate CPI-indexed life (H.R. 3055, 108th access 50/50 sharing annuity such thatCongress) Social Security

plus the annuity meets poverty line

Pozen, 2002 No split; Not specified Funds go to Default joint and Surviving spouseno early access widowed spouse 2/3 survivor an- must keep for own

or estate nuity (solvency retirementestimates assumed CPI-indexed)



Even if federal law did not require spousal con-sent for all pre-retirement withdrawals or loans,it could require account administrators to com-ply with a preliminary order in a divorce caseconstraining account holders from depletingtheir accounts, and require spousal notice topromote consultation about a decision withpotentially long-term financial consequences forthe couple’s retirement income.

Spousal Rights at Divorce beforeRetirement Any individual accounts proposal must alsoaddress what happens at divorce. If contributionsplitting does not occur during marriage, thendivorce might trigger a division of accountsbetween husbands and wives. The following sixapproaches show various ways to combine fed-eral mandates for contribution-splitting duringmarriage, dividing accounts at divorce, and vari-ous roles for state courts. It will be importantfor policymakers to provide explicit guidance ontwo issues; whether or not accounts would beautomatically divided at divorce and whether or not state courts would have authority to reallocate the funds as part of an overall divorcesettlement.

(1) Contribution splitting only. If policymakersadopted contribution splitting during marriage,then at divorce each spouse could retain his orher own account—nothing more and nothingless. This approach might be the simplest toadminister at the time of divorce if contributionsplitting had already created two separateaccounts. This policy would also represent acongressional decision to shield accounts fromstate divorce proceedings.

(2) Contribution splitting and allow divorcereallocation. This approach would establish con-tribution splitting with no further division atdivorce as the default, but would allow a statecourt to reallocate accounts at divorce if neces-sary to avoid an inequitable result. This wouldbe analogous to the approach taken by somestates on the reallocation of separate property

during a divorce, which can be done in specialcircumstances.

If contribution splitting had not alreadyoccurred during marriage, then possible rules atdivorce might be:

(3) Mandate a 50/50 split of account contribu-tions and accumulations during marriage andpermit no exceptions;

(4) Set the default as a 50/50 split of accountcontributions and accumulations during mar-riage, but permit other allocations if agreed byboth parties or ordered in a divorce settlement;

(5) Defer to state divorce proceedings explicitly.Federal law would permit reallocation ofaccounts if negotiated or ordered as part of thedivorce settlement, as with 401(k)s or IRAs;

(6) Shield the accounts from state divorce pro-ceedings altogether. Federal law could explicitlypreempt state family laws that otherwise wouldgive spouses a claim to a share of the otherspouse’s account at the time of divorce.

Would Funds Acquired at Divorce beAccessible?

If individual accounts were transferred betweenspouses at divorce, what restrictions, if any,would limit the recipient’s use of the funds?When IRAs or 401(k) funds are transferred atdivorce, the recipient can freely use the funds aslong as taxes and tax penalties are paid on thewithdrawals. Federal laws also permit the tax-favored account to be rolled over into anotherqualified tax-favored account for the ex-spousewithout paying taxes or penalties.

As indicated in Chapter Five, most proposals forindividual accounts as part of Social Securitycall for banning any access to individual accountfunds before retirement. Would that ban alsoapply to funds acquired at divorce? Whether ornot funds acquired at divorce are accessiblecould influence property settlement negotiations.For example, a party in need of immediate cash


might trade off inaccessible retirement funds toget more liquid assets for immediate needs.

When A Spouse Dies before RetirementIf a married worker dies, important questionsare raised about the nature of the widowedspouse’s rights to the deceased worker’s account.State laws on widow’s or widower’s rights differdepending on whether states follow communityproperty rules or common law rules and, inmany cases, on whether children from the cur-rent marriage or a former marriage also survivethe worker (Shammas, et al., 1987).

Pension and retirement savings plans offer avariety of precedents that preempt state propertyand family laws. Many individual account pro-posals that are part of Social Security reforminclude rules designed to protect widowedspouses. Possible options include:

(1) Mandatory spousal inheritance. This optionwould require that the deceased worker’saccount go to the widowed spouse withoutexception. If the account holder was unmarried,the account could go to a named beneficiary.

(2) Spousal inheritance, unless the spouse con-sents otherwise. The default in this optionwould transfer the account to the widowedspouse. If the spouse had waived the inheritanceright in writing, the account could go to anothernamed beneficiary. Federal rules call for this pol-icy in most defined-contribution plans. Indefined-benefit plans, however, widowed spous-es are entitled to a survivor annuity of at least50 percent, unless the spouse has waived thisright.

(3) Free choice for account holder with defaultto a federal order of precedence. In this option,the account holder could name anyone he or shechose as death beneficiary without spousal con-sent. If, however, no death beneficiary werenamed, an order of precedence established byfederal law would apply and the spouse wouldbe first in line. The TSP follows this option.

(4) Free choice for account holder with defaultto the estate and state law distribution. Thisoption would let the account holder name any-one he or she chose as death beneficiary withoutspousal consent. If, however, no death benefici-ary were named, the proceeds would be part ofthe decedent’s estate, unless the account rulesspecified an order of precedence for inheritance.Without such rules, the account would be dis-tributed according to the terms of the decedent’swill or, if there were no will, through state intes-tacy laws. This is the approach used with IRAs.

Would Inherited Funds Be Accessible?

A key question is whether widowed spouses orother heirs would be restricted in what theycould do with funds inherited from individualaccounts. In IRAs, the TSP, and retirementplans, the heir is free to use the funds in anyway he or she chooses. Many proposals that aimto replace part of traditional Social Security ben-efits with individual accounts are more restric-tive with regard to widowed spouses; theseproposals require that the widow(er) always bethe death beneficiary and, in addition, that thefunds be preserved for retirement. If there is nowidowed spouse, the death benefit can go to anyother named beneficiary and no restrictions areplaced on use of the funds (Figure 6-3). In brief,these plans treat the widow’s (or widower’s)inheritance as a transfer that remains subject tothe restrictive rules of the retirement system,while inheritance by others is an unrestrictedbequest outside the retirement system.

While restricting a widow’s use of funds mightbe motivated by a desire to protect the widow’sretirement security, the policy could be viewedas unfairly restrictive. At a spouse’s death, awidowed spouse might have a number of imme-diate and pressing financial needs, such as funer-al expenses. Restricting use of death benefitsonly for widowed spouses and not for otherbeneficiaries could encourage account holders toname someone else as a death beneficiary – suchas a child or trusted relative –to provide the sur-viving family with immediate access to themoney.


Social Security currently pays benefits to twogroups of non-elderly widows: widows withchildren under age 16 and disabled widows age50 and older. Even if the rules for individualaccounts generally restrict widows’ pre-retire-ment access to the funds, the rules might allowaccess for these widows. However, allowing forearly payment would reduce the future retire-ment benefits that could be paid from individualaccounts. (See Chapter Eight for a discussion ofpayment options for young survivor families,which could include benefits for widows caringfor minor children as well as children’s benefits.)

Requiring the widow to preserve inherited fundsfor retirement might also be important if a planutilizes worker-specific offsets as part of itsdesign and financing (see Chapter Nine). If onlybequests to a spouse were subject to an offset,this could also provide an incentive to nameanother death beneficiary, if permitted by theplan.

Spousal Inheritance and Second Marriages

A number of Social Security reform plans wouldrequire that accounts always go to the widowedspouse and that the inherited account must bepreserved for the widowed spouse’s retirement.This policy could generate unintended results inthe event of second marriages. For example,Mary inherits John’s retirement account at hisdeath and later marries Peter. If Mary diesbefore retirement, this policy would mean thatPeter would get all of Mary and John’s accumu-lated accounts. From a property perspective,other potential heirs of John and Mary – such as their children - might believe they have amore legitimate claim to their deceased parents’accounts than does their mother’s second husband.

If individual account policy followed communityproperty concepts, then inherited accounts pre-served for retirement would not be co-mingledwith marital property in a subsequent marriage.Separate accounting could facilitate separatetreatment of inherited accounts in future distri-butions. It also could raise new administrative

hurdles in situations where individuals holdmultiple accounts from different sources.

Implementation Issues in Free Choice ofDeath Beneficiaries

The IRA and TSP models allow free choice ofdeath beneficiary. In a large national system,administrative complexities will arise if a deathbeneficiary has not been named, cannot befound, or has predeceased the account holder.The TSP retirement plan and IRA systems usesomewhat different procedures to determineinheritance rights in cases of ambiguity. The TSPhas a federal statutory standard of inheritancerights in cases where the distribution at death isunclear. In contrast, retirement plans and IRAsoften include rules for determining beneficiarieswhere none has been named. Otherwise, theassets become part of the participant’s estate or,if there is no will, the question of inheritance issettled by state law.

The TSP model provides federal statutory stan-dards to guide inheritance where there is no sur-viving designated beneficiary; it turns to statelaw only after pursuing an order of precedencethat is consistent throughout the country. Thefederal default procedures follow an order ofprecedence similar to that found in many stateintestacy laws, with the widowed spouse first inline, followed by the decedent’s children. Box 6-1 summarizes the TSP order of precedence.

By comparison, if a plan participant or an IRAholder fails to name a death beneficiary, theaccount becomes part of the decedent’s estatelike other property left behind. The funds arethen distributed according to the account hold-er’s will (subject to any claims the widowedspouse might have against the will) or, in theabsence of a will, according to state intestacylaw. If the beneficiary predeceases the IRA par-ticipant, the beneficiary’s heirs lose any claim tothe account and the account is divided amongother named beneficiaries. If no beneficiariessurvive, the balance goes to the account holder’sestate. IRA beneficiary designations also providefor cases in which the named beneficiary cannot


be found (the account goes to the estate), orwhen ambiguities or conflicts arise (the custodi-an can consult counsel and institute legal pro-ceedings to be paid for with account funds).Language from a sample IRA beneficiary desig-nation form is shown in Box 6-2.

Retirement Payouts for Married AccountHoldersRetirees face questions about how to ensure thattheir savings will last the rest of their lives (seeChapter Three for a detailed treatment of thisissue). Outside of any retirement account sys-tem, individuals could purchase private annu-ities, withdraw from their retirement accounts ata fixed rate, or just spend the money as theylike. For its part, Social Security offers no choiceabout the form of payments in old age, consis-tent with its goal of ensuring a basic retirementbenefit throughout the lifespan. When qualifiedretirees or spouses or survivors apply, benefitsare automatically paid in regular monthly

amounts. Payments last for life and are adjustedeach year to keep pace with inflation. Benefitsfor wives, widows and children are supplemen-tal; they do not reduce the worker’s benefit.

Spousal rights in individual accounts present dif-ferent policy issues because any money designat-ed for a spouse would reduce the moneyavailable to the worker. In contrast to SocialSecurity, where contributions are pooled andfamily benefits are funded from Social Securitytax revenue, couples with individual accountscould pull only from their own account funds.Policymakers, then, would have to balance theinterests of workers with the interests of spous-es, and design rules that best coincide with pro-gram goals. More flexible structures might helplure participants if individual account participa-tion is voluntary, but greater flexibility mightnot offer spousal protections that policymakersregard as adequate, especially if individual

Box 6-1. Thrift Savings Plan Default Order of Precedence When No Named Death BeneficiaryExists

If no death beneficiary has been named (or no named beneficiary outlives the account holder), theaccount is distributed according to the following order of precedence:

(1) To the widowed spouse;

(2) If none, to the account holder’s child or children* equally, and descendants of children by represen-tation;**

(3) If none, to the account holder’s parents equally or to the surviving parent;***

(4) If none, to the appointed executor or administrator of the estate;

(5) If none, to the next of kin who is entitled to the estate under the laws of the state in which theaccount holder resided at the time of death.

*Children include biological and adopted children, but not stepchildren who have not been adopted.

**”By representation” means that if a child of the account holder had died, the deceased child’s share will be divided equally amonghis or her children.

***A parent does not include a stepparent, unless the stepparent had adopted the account holder.

Source: Federal Retirement Thrift Investment Board, 2001. Summary of the Thrift Savings Plan for Federal Employees


accounts are intended to replace traditionalSocial Security benefits.

Would Payouts Be Structured Like SocialSecurity?

Some individual account proposals aim to pro-vide benefits somewhat analogous to SocialSecurity’s current protections for spouses wid-owed after retirement by requiring marriedretirees to buy inflation-indexed monthly annu-ities with joint and two-thirds survivor benefits.8

Inflation indexing is particularly important toindividuals expected to live for many years afterretirement because purchasing power can be sig-

nificantly eroded by even modest rates of infla-tion over long periods of time.

Policymakers could make the purchase of joint-and-survivor annuities mandatory, or establishthat option as the default, allowing a spouse towaive the right to receive an annuity, as is thecase with defined-benefit pensions. If both thehusband and wife purchased joint and two-thirds life annuities, when either one died thewidowed spouse would begin receiving two-thirds of his or her own annuity and two-thirdsof the deceased partner’s annuity.

Box 6-2. Sample Death Beneficiary Designation for an IRA

I elect that when I die, my interest in my IRA will become the property of:

• The primary beneficiary, if he or she survives;

• Or if no primary beneficiary survives, the contingent beneficiary.

If no designated beneficiary survives, or if the custodian cannot locate the beneficiary, then the custodianwill distribute the benefits to my estate.

I understand that if I fail to indicate percentage of benefits, the plan administrator will divide benefitsequally among the beneficiaries I designate.

If a primary beneficiary dies after my date of death, but before receiving his or her percentage share of theIRA, his/her share should be transferred/distributed to his/her estate.

I understand that if any primary or contingent beneficiary dies and I wish to name a new beneficiary, ormodify existing beneficiary information, I must complete a new IRA beneficiary form.

I reserve the right to revoke or change this beneficiary designation. I understand that any change or revo-cation will not be effective until it is given in writing to the plan administrator. This designation revokes allprior designations (if any) of primary or contingent beneficiaries.

I understand that if the plan administrator determines that my beneficiary designation is not clear withrespect to the amount of the distribution, the date on which the distribution shall be made, or the identityof the party or parties who will receive the distribution, the plan administrator will have the right, in itssole discretion, to consult counsel and to institute legal proceedings to determine the proper distribution ofthe account, all at the expense of the account, before distributing or transferring the account.

Source: Charles Schwab & Co., 2004. Retirement Accounts for Individuals


Policymakers could provide more spousal pro-tection by requiring (or permitting) retirees tobuy a joint and 100 percent survivor annuity. Inthis case, the widowed spouse would suffer noloss in income.

Impact of Survivor and Inflation Protectionon Annuity Amounts

The extensive protections of joint-and-survivorannuities are not costless. Each increment ofextra insurance—inflation indexing, two-thirdsjoint-and-survivor benefits, 100 percent survivorbenefits—will be reflected in lower monthly pay-ments granted to annuitants; as the protectionsmount, monthly payments decrease. For exam-ple, a 65-year-old with a slightly younger spousewith an account of $10,000 could buy a simpleannuity with no inflation or survivor protectionthat would pay about $80 a month (Figure 3-5in Chapter Three). Inflation protection (assum-ing 3 percent inflation) would lower the individ-ual annuity to about $62 a month. Addingtwo-thirds survivor protection for a spouse aged62 would lower the initial annuity to about $55a month for the account holder, with the sur-vivor benefit starting at about $37 a month.Requiring 100 percent survivor annuities wouldproduce an even smaller annuity while bothwere alive—$48 a month in this example—butthe payment would continue undiminished tothe widowed spouse.

Disparities in Age

Joint-life annuities will vary depending on theages of both the husband and wife. If a retireehad a much younger wife, the annuity would besmaller. For example, if a 65-year-old retiree hada 53-year-old wife, he would take a bigger cut inhis annuity to provide survivor protection forhis wife. If he had $10,000 to spend, giving hiswife a two-thirds survivorship interest in theannuity would reduce his immediate paymentfrom $62 (under a single-life annuity) to about$48—a drop of about 22 percent—well belowthe approximately $55 payment if his wife were62 (Figure 3-9 in Chapter Three). Twelve yearslater, when the wife reached age 65 and her hus-band was 77, she would buy a joint and two-

thirds survivor annuity but, in her case, theimmediate annuity payment would be about 10percent higher than the single-life annuity shecould buy at that age.9

Change in Marital Status After Annuitization

In general, life annuities cannot be rewrittenafter they are purchased to, for example, changefrom a single-life annuity to a joint-and-survivorannuity, or vice versa, or to name a differentbeneficiary for the survivor annuity. So, if anindividual marries after annuitization, it couldbe difficult for the new couple to ensure sur-vivor protection in widowhood. Joint-and-sur-vivor annuities help people who are married atthe time they annuitize, but someone who wassingle, divorced, or widowed at retirementwould have purchased a single-life annuity. Ifthis retiree then got married, he or she would beunable to convert the single-life annuity into onethat protected his or her new spouse as well.10

But if the spouse were younger, she or he wouldbe required, under a pure annuity mandate forall married retirees, to buy a joint-and-survivorannuity when she or he annuitized at retirementage. The result would be an asymmetry of bene-fits between the two spouses.

The treatment of individuals who divorce afterat least one has purchased an annuity posesother issues. If both spouses had purchased jointand two-thirds survivor annuities, then eachmight keep his or her own annuity and the sur-vivorship interest in the other’s account evenafter a divorce. But if either remarried therewould be no survivor annuity to leave to a newspouse.

If only one spouse had annuitized beforedivorce, a variety of approaches are possible.One option would divide the un-annuitizedaccount according to the plan’s usual rules fordivision at divorce. This would leave one spousewith the right to receive a survivor benefit fromthe spouse who had annuitized, and the spousewho had annuitized with a share of the otherspouse’s account. (The plan would also have todecide what options are available to someone


who receives account proceeds through divisionat divorce or by inheritance after annuitization.)Another approach would allow state courts todecide how to address the issues, for example,by dividing the account in an equitable way,ordering the spouse who had not yet annuitizedto name the former spouse as the beneficiary ofa joint-and-survivor annuity at the time of annu-itization (which would eliminate any survivorbenefits for a subsequent spouse), or awardingother property or income to achieve an equitableresult.

In summary, a system providing maximalspousal rights would mandate contribution split-ting during marriage or automatic division ofaccounts at divorce, spousal inheritance of anaccount if death occurs before retirement, andpurchase of joint-and-survivor annuities by allmarried individuals at retirement. At the otherend of the spectrum, a regime with no spousalrights would have federal law preempt state lawand create no federal spousal rights. As such,spouses would lose any state law claims to con-tributions to an account made by the otherspouse during marriage, shield individualaccounts from divorce settlements, allow theaccount holder to name anyone he or she want-ed as a death beneficiary, and give the accountholder unfettered discretion over the use offunds at retirement. Policies for spousal benefitrules could fall anywhere between these twoextremes.

Implementation Issues

Administering spousal rights in an individualaccount system could impose new reporting andverification requirements, beyond those faced bythe Social Security system. For its part, SocialSecurity simplifies administration of family ben-efits in a number of ways.

Determining Family Status at BenefitEntitlement or Over the LifetimeSocial Security benefit entitlement is generallybased on the family relationships in existencewhen individuals establish entitlement to bene-

fits—when workers retire, die, or become dis-abled. The system does not need to track mar-riage and divorce over the working life. Ifindividual accounts required ongoing updates onthe account holder’s family status before becom-ing entitled to benefits, Congress would need toauthorize new administrative arrangements forreporting and resolving disputes or discrepanciesin marital status. Additionally, the ongoingupdates would need to account for less formalfamily relationships such as common law mar-riage (recognized by some states, but not by all),informal separation or abandonment, or parent-child relationships. These less formal family rela-tionships are more time-consuming to documentbecause they are based on evidence of livingarrangements and financial support rather thanon official records of marriage, divorce, birth, oradoption. Documenting these changes only atthe time of benefit entitlement requires fewerresources than year-by-year reporting and verification.

Evidence of divorce and marriage duration isrequired only infrequently to determine eligibili-ty for Social Security benefits. About 1 millionof the 46 million Social Security beneficiaries in2001 received benefits based wholly or in parton an ex-spouse’s work record. Of the 4.2 mil-lion people newly awarded Social Security bene-fits in 2001, roughly 100,000 of the casesinvolved evidence of divorce—a scant one-tenthof the approximately one million divorces thatoccur every year in the United States (U.S.Census Bureau, 2001). As these numbers indi-cate, an individual account system requiring thedivision of accounts at divorce would likelyrequire far more frequent and extensive collec-tion of evidence of marriage, spousal identityand contributions, duration of marriage, anddivorce. This collection would, in turn, mean amajor step-up in administrative capacities andcosts over the current system.

Incentives to Report and DocumentFamily Status The current Social Security benefit structure pro-vides a strong incentive for individuals to report


and document family relationships. Spouses anddivorced spouses receive benefits in addition tothose paid to workers, with no consequentreduction in workers’ own benefits. When aworker divorces and remarries, for example,both the divorced spouse and the subsequentspouse could receive benefits with no change inthe worker’s own benefit income. In otherwords, it is a win-win proposition to report anddocument marital and family information accu-rately: the worker suffers no loss, and it is to theadvantage of the spouse. These rules help ensurethe necessary flow of information fromclaimants to the Social Security Administration.

By contrast, if individual accounts were dividedbetween husbands and wives, either by contribu-tion splitting year-by-year or by dividingaccounts at divorce, account holders might failto report a marriage because they do not want aspouse to receive funds at their own expense. Asa result, administering an automatic division ofaccounts at divorce—a rule that on its face hasthe appeal of simplicity and administrativeease—could entail new reporting and verifica-tion. If the rule for division at divorce were lim-ited to just the contributions and earningsaccumulated during marriage, then historicalrecords of year-by-year individual contributionsand investment earnings would be needed alongwith evidence and dates of marriage anddivorce.

Mechanisms for Reporting Family StatusEmployers, the federal income tax system, statemarriage and divorce records, and individualsthemselves are four possible sources of informa-tion about marital status and spousal identities.

Employers’ W-2 and W-3 reports are the mainsource of information that the Social SecurityAdministration relies on to correctly credit eachworker’s annual earnings to his or her SocialSecurity record. This system works remarkablywell, but it is not foolproof. Wages that cannotbe matched to a Social Security number areplaced in a suspense file. This suspense file cur-rently holds 236 million wage items totaling

about $375 billion, $49 billion of which wasadded in tax year 2000 (U.S. SSA, 2003c).When a wage report does not match identifyinginformation in its records, the Social SecurityAdministration uses a series of automated rou-tines to check for common mismatches – varia-tions in the spelling of common names ornicknames, or transposed digits in the SocialSecurity number. When these routines do notyield a match, the Social Security Administrationcontacts the employer to help resolve the dis-crepancy. In sectors that experience high laborturnover – such as agriculture, food service, andconstruction – employers are often unable tohelp resolve discrepancies because the employeehas moved on and the employer has no furtherinformation.

While employers might be willing to pass onmarital status information that their employeesvoluntarily report, it is doubtful that employerswould want to be responsible for ensuring theaccuracy of spousal information sufficient for anindividual account system administrator tomatch records. Nor would employers likelywant to become intermediaries for resolving dis-crepancies between the Social SecurityAdministration’s records and the spousal identi-fying information supplied by a current or for-mer employee. A system beyond employerreports would be needed to document and verifyspousal identities.

Federal income tax returns might be used to linkinformation about husbands and wives, at leastin cases in which couples chose to file joint taxreturns. Using income-tax reporting as a basisfor enforcing the division of individual accountsbetween husbands and wives could influencewhether or not taxpayers file joint returns.

Because marriage is a state-defined legal statusthat imposes responsibilities on spouses for chil-dren and sharing of property and expenses,states maintain marriage records in order todocument a legal status that cannot be duplicat-ed and that can only be terminated through alegal procedure. Because an individual might


marry in one state and divorce or become wid-owed in another state, validation from severalstates might be needed to document a maritalhistory. One approach to obtaining marriageand divorce data is to establish a nationalreporting system, perhaps through the NationalCenter for Health Statistics (NCHS), the federalagency that now collects data on deaths fromstate and county records. Use of a standardreporting form and paying states for the deathdata has led to fuller state compliance and uni-formity. The National Death Index maintains anational record on each death from computerfiles submitted by state vital statistics offices.This file is currently used by the Social SecurityAdministration to validate deaths. States main-tain marriage and divorce records and somestates, such as Massachusetts, have historicalrecords that are publicly available. Other statesoffer the ability to request marriage and divorcecertificates online.

The NCHS once tried to construct a nationalsystem of state marriage and divorce records,but abandoned this effort in 1995, due to budg-etary pressures and concerns about the qualityand completeness of the divorce data. In theory,the agency could renew effort to collect mar-riage and divorce data for administering individ-ual accounts. Responsibility for creating such asystem could also be assigned to another federalagency. For example, the Federal Case Registryis a national record system maintained by thefederal Office of Child Support Enforcement.The Registry is designed to include informationabout all child support cases handled by statechild support agencies and all support ordersestablished or modified after October 1, 1998,even if they are not being enforced by stateagencies.

A central depository of state marriage anddivorce records would reduce problems ofincomplete reporting associated with a voluntaryreporting system or with an income-tax basedsystem that does not require everyone to file. Itwould require uniform reporting across statesand may raise confidentiality issues that now

vary across the states. While a national index ofmarriage and divorce records would documentmarital histories, it would not resolve changes inproperty rights that might occur as individualsmove from state to state.

New resources would be needed to set up anational marriage and divorce record systemand ensure that states collect the information ina uniform, complete and timely way. Decisionswould be needed about how much of the infor-mation is publicly accessible and how to protect the privacy and security of non-publicinformation.

Finally, individuals could be asked to providecorrect and current information about theirmarital status through the Social Security state-ment that is currently mailed every two years toworkers age 25 and older.

Depending on how accounts were managed,information about marital status might need tobe transmitted from the government to a privateentity managing the account. Policymakerswould need to determine how to assign respon-sibility for verifying the information.

Resolving Disputes

The division of an individual account wouldproduce winners and losers. Since money, oncepaid out of an account, would be difficult torecover, disputes about family status are morelikely to arise with an individual account planthan with Social Security. Policymakers wouldneed to develop procedures for hearing andresolving disputes.

Summary

Social Security currently provides spousal bene-fits to wives, widows, husbands, widowers, aswell as ex-wives and ex-husbands, protectingthem against the income consequences of aworker’s death, retirement, or disability. Thesebenefits are provided without reduction in thebenefit paid to the worker. Spousal benefits arepaid only to the extent that the benefit exceeds


what the spouse would receive based on his orher own work record; benefits paid for life andare annually indexed for inflation. The introduc-tion of individual accounts would require reso-lution of what rights for spouses and survivorswould be required in those accounts, with thatresolution being more important if the accountswere designed to replace a part of SocialSecurity benefits.

If Congress establishes a system of individualaccounts, a key question is whether to have uni-form federal rules concerning spousal rights orleave the issues to state jurisdiction. Rules thatdefine spousal rights and protections in individ-ual accounts could be based on federal lawprecedents in Social Security, federal provisionsin current pension and retirement plans, orcould be delegated to state family law.

The federal Employee Retirement IncomeSecurity Act and the Internal Revenue Codedetermine spousal rights in private pensions andretirement savings. These spousal rights varywidely across types of retirement accounts, butgenerally protect only current spouses. Privateemployer-provided defined-benefit plans requirethat a spouse receive at least a 50 percent sur-vivor pension from the plan, unless the spousehas waived that right. Providing the survivorpension lowers the pension amount for theretiree. Tax-deferred individual retirementaccounts provide no special rights to spouses,although the accounts can be divided at divorceunder state family law.

In the United States, the historical delegation tostates of jurisdiction over family property rightshas produced marked differences in law and pol-icy among states. Common law states considerthe title-holder to be the owner of the property,although all common law states now call forequitable distribution of property at divorce ordeath. Community property states view propertyacquired during the marriage as belongingequally to husbands and wives.

If policymakers wish to implement uniformspousal rights under an individual account sys-tem, they will need to define the rules explicitlyin federal legislation. Absent a federal pro-nouncement, state courts and legislatures willmake determinations about spousal protection,the treatment of accounts at divorce, andaccount inheritance. These determinations willlead to differences in the way accounts are treat-ed for account holders residing in differentstates. State rules may also lead to changes inthe property treatment of accounts as accountholders move between common law and com-munity property states.

If policymakers wish to have uniform federalrules, important tradeoffs remain in choosingbetween spousal protections that resemble fea-tures of traditional Social Security and spousalrights that relate to property concepts withregard to inheritance rights and equitable divi-sions of marital property at divorce.

A voluntary individual account system also rais-es some specific challenges to designing andimplementing spousal rights. If participationwere voluntary, there would be no guaranteethat both spouses would make the same decisionto participate. Rules that seem equitable whenboth parties are participants could have unin-tended outcomes if only one party participatesin the plan. Proposals for voluntary accountsalso often call for worker-specific offsets againstfuture Social Security benefits or account pro-ceeds; how these offsets would interact withspousal payments from traditional SocialSecurity or the individual account raise complexquestions that are discussed in Chapter Nine.

Administering spousal rights in an individualaccount system could impose new reporting andverification requirements and dispute-resolutionprocedures beyond those needed to implementSocial Security. Social Security family benefitsare generally based on family relationships ineffect when people claim benefits – when aworker retires, dies or becomes disabled.Divorced spouses claims are also determined at


that time. In contrast, updated marital statusdata would be needed if marital property con-cepts were used to allocate spousal rights duringmarriage. While a national marriage and divorceregistry would accomplish this, such a systemdoes not now exist. Without such a system,some administrative mechanisms would be need-ed to resolve factual disputes about marital sta-tus, timing, and duration between parties whostand to gain or lose by the allocation of spousalrights. Such disputes are minimized in tradition-al Social Security because benefits for a worker

and current spouse are not affected by the bene-fit claims of an ex-spouse.

As the foregoing discussion illustrates, any indi-vidual account proposal must look beyond theindividual account holder and address the issuesof spousal rights to the account during marriage,at divorce, at retirement, and at death. A cleararticulation of congressional intent as to therights of current and ex-spouses would be neces-sary to clarify the process of payouts from indi-vidual accounts.


Chapter Six Endnotes

1 In this chapter “spousal benefit” describes bene-fits based on a former or current marriage to aprimary beneficiary. The term includes benefits towives, husbands, widows, and widowers. “Wife”benefits refer to the benefits paid to women andmen married to a retired or disabled worker.“Widow” benefits refer to benefits paid to wid-ows and widowers of deceased workers.

2 Divorced spouses are included as spouses or wid-owed spouses in Figure 6-1 but are not identifiedseparately.

3 In community property states an employer-pro-vided defined-benefit pension plan may be con-sidered community property while both husbandand wife are alive, but the interest of the second-ary spouse may terminate upon that spouse’sdeath. That is, the non-pensioner may not have aright to bequeath a share of the other spouse’sdefined-benefit pension.

4 There are two versions of the TSP. For employeescovered under Social Security and the newerFederal Employees’ Retirement System (FERS)that began in 1986, the TSP involves employer-matching contributions and has somewhatstricter rules for spousal consent and protections.Employees who remain in the old Civil ServiceRetirement System (who are not covered bySocial Security) can contribute to the TSP buthave no employer matching funds and their TSPaccounts have somewhat weaker spousal protec-tions, in that the rules call for spousal notice, butnot spousal consent.

5 The FERS TSP requires spousal consent asdescribed; the CSRS TSP requires spousal notice.

6 For example, under federal law, a widowedspouse has a right to a 50 percent survivor’sannuity in a defined benefit plan regardless ofstate inheritance laws, but the division of benefitsat the time of divorce is explicitly left to statecourts.

7 Contribution splitting has some similarity to pro-posals for earnings sharing. Earnings sharingproposals would divide Social Security wagecredits on which Social Security defined benefitsare based. See, Technical Committee on EarningsSharing, Center for Women’s Policy Studies,Earnings Sharing in Social Security: A Model forReform (Edith Fierst and Nancy Duff Campbell,eds., 1988). Contribution-splitting as discussedhere would divide contributions to individualaccounts each year as contributions are made.

8 As discussed in Chapter Three, joint-life annu-ities produce somewhat different results thanSocial Security benefits for both one-earner cou-ples and two-earner couples in retirement.

9 The joint-life annuity is larger than the single lifeannuity because the joint-life annuity pays lesswhen the primary annuitant is widowed, whichis highly probable when the spouse is 12 yearsolder.

10 The current Social Security system does providethis benefit, protecting the new spouse againstloss in income upon the death of the remarriedcovered worker.


App

endi

x: P

rece

dent

s fo

r Sp

ousa

l Rig

hts

IRA

1

401(

k)

Non

e. T

he a

ccou

nt h

olde

rm

ay s

pend

IRA

fun

ds w

ith-

out

spou

sal c

onse

nt(a

lthou

gh t

ax p

enal

ties

appl

y fo

r w

ithdr

awal

sbe

fore

age

591 /2

).

Non

e. T

he p

artic

ipan

t m

ayta

ke lo

ans2

or h

ards

hip

with

draw

als

from

his

or

her

acco

unt

whi

leem

ploy

ed b

y th

e pl

ansp

onso

r w

ithou

t sp

ousa

lco

nsen

t.

If th

e pa

rtic

ipan

t le

aves

the

spon

sorin

g em

ploy

er,

he o

rsh

e m

ay s

pend

401

(k)

fund

s (a

lthou

gh t

ax p

enal

-tie

s ap

ply)

or

roll

the

fund

sin

to a

n IR

A w

ithou

tsp

ousa

l con

sent

.

Non

e. T

he a

ccou

nt h

olde

rm

ay s

pend

any

or

all I

RAfu

nds

with

out

spou

sal

cons

ent.

Non

e if

acco

unt

fund

s ar

edi

strib

uted

as

a lu

mp

sum

or in

per

iodi

c pa

ymen

ts(t

he m

ost

com

mon

and

ofte

n th

e on

ly o

ptio

ns).

The

part

icip

ant

may

spe

ndan

y or

all

ther

eby

dist

rib-

uted

401

(k)

fund

s w

ithou

tsp

ousa

l con

sent

. If

annu

-iti

es a

re a

n op

tion,

spo

usal

cons

ent

is r

equi

red

for

any

annu

ity p

urch

ase

othe

rth

an a

join

t-an

d-su

rviv

or.

IRA

fun

ds g

ener

ally

are

subj

ect

to d

ivis

ion

atdi

vorc

e. T

he s

pous

e ca

nse

ek s

ome

port

ion

of IR

Afu

nds

in d

ivor

ce p

roce

ed-

ings

, bu

t do

es n

ot a

uto-

mat

ical

ly r

ecei

ve a

sha

re.

Any

ann

uity

aw

ard

redu

ces

the

fund

s th

at t

he a

ccou

ntho

lder

will

rec

eive

.

401(

k) f

unds

are

sub

ject

to

divi

sion

at

divo

rce.

A p

ar-

ticip

ant’s

spo

use

can

seek

som

e po

rtio

n of

401

(k)

fund

s in

div

orce

pro

ceed

-in

gs,

but

does

not

aut

o-m

atic

ally

rec

eive

a s

hare

.A

ny a

war

d w

ill r

educ

e th

efu

nds

that

the

par

ticip

ant

will

rec

eive

.

IRA

fun

ds w

ill b

e di

strib

-ut

ed t

o th

e de

sign

ated

bene

ficia

ry.

The

acco

unt

hold

er m

ay d

esig

nate

any

-on

e he

or

she

choo

ses

as a

bene

ficia

ry.

If a

part

icip

ant

dies

bef

ore

retir

emen

t, h

is o

r he

rsp

ouse

will

rec

eive

the

401(

k) a

ccou

nt b

alan

ce,

unle

ss t

he s

pous

e ha

s pr

e-vi

ousl

y co

nsen

ted

in w

rit-

ing

to t

he d

esig

natio

n of

an a

ltern

ate

bene

ficia

ry.

Non

e. If

IRA

fun

ds r

emai

n,th

ey w

ill b

e di

strib

uted

to

the

desi

gnat

ed b

enef

icia

ry.

Non

e, t

o th

e ex

tent

the

bala

nce

was

alre

ady

dis-

trib

uted

bef

ore

or a

t re

tire-

men

t. If

any

fun

ds r

emai

nin

a p

artic

ipan

t’s 4

01(k

)ac

coun

t, t

he s

pous

e w

illre

ceiv

e th

e ac

coun

t ba

l-an

ce u

nles

s th

e sp

ouse

has

prev

ious

ly c

onse

nted

inw

ritin

g to

the

des

igna

tion

of a

n al

tern

ate

bene

ficia

ry.

Fig

ure

6-A

. Sp

ou

sal R

igh

ts a

nd

Ben

efit

s in

Var

iou

s R

etir

emen

t Sa

vin

gs

and

Pen

sio

n P

lan

s

Pre-

Ret

irem

ent

Dis

trib

uti

on

at

Pre-

Ret

irem

ent

Post

-Ret

irem

ent

Dis

trib

uti

on

Ret

irem

ent

Div

orc

eD

eath

Dea

th

1

Gen

eral

ly, s

tate

law

gov

erns

IRA

s. S

pous

al r

ight

s va

ry b

oth

betw

een

com

mon

law

and

com

mun

ity p

rope

rty

stat

es, a

nd a

mon

g st

ates

in e

ach

cate

gory

.

2 4

01(k

) pla

ns t

ypic

ally

allo

w p

artic

ipan

ts t

o bo

rrow

50

perc

ent

of t

heir

acco

unt

bala

nces

, up

to $

50,0

00.


Def

ined

-Ben

efit

/M

on

ey P

urc

has

ePe

nsi

on

Pla

n

CSR

S an

d F

ERS7

A p

artic

ipan

t’s s

pous

e m

ust

cons

ent

if th

e pa

rtic

ipan

tse

eks

to u

se t

he a

ccru

edbe

nefit

as

colla

tera

l for

alo

an.

Oth

er t

han

for

loan

s,w

hich

ver

y fe

w p

lans

allo

w,

part

icip

ants

are

gen

-er

ally

una

ble

to a

cces

sth

eir

accr

ued

bene

fit p

rior

to r

etire

men

t.

Und

er F

ERS,

a p

artic

ipan

t’ssp

ouse

mus

t co

nsen

t to

in

-ser

vice

loan

s an

d w

ithdr

awal

s.

Und

er C

SRS,

a p

artic

ipan

tm

ay m

ake

volu

ntar

y co

ntri-

butio

ns t

o fu

nd a

hig

her

annu

ity.

The

part

icip

ant

can

with

draw

the

se f

unds

at a

ny t

ime,

with

out

spou

sal c

onse

nt.8

A p

artic

ipan

t’s s

pous

e m

ust

cons

ent

if th

e pa

rtic

ipan

tse

eks

to r

ecei

ve p

ensi

onbe

nefit

s ot

her

than

a jo

int

and

at le

ast

50 p

erce

ntsu

rviv

or a

nnui

ty.

A s

ur-

vivo

rshi

p in

tere

st f

or t

hesp

ouse

will

red

uce

the

par-

ticip

ant’s

ann

uity

.

A p

artic

ipan

t’s s

pous

e m

ust

cons

ent

if th

e pa

rtic

ipan

tse

eks

to r

ecei

ve p

ensi

onbe

nefit

s ot

her

than

a jo

int

and

no m

ore

than

55

per-

cent

sur

vivo

r an

nuity

. A

join

t-an

d-su

rviv

or a

nnui

tyw

ill r

educ

e th

e an

nuity

paym

ents

tha

t th

e pa

rtic

i-pa

nt w

ill r

ecei

ve.

Pens

ion

bene

fits

are

sub-

ject

to

divi

sion

at

divo

rce

thro

ugh

a pr

oper

ly e

xecu

t-ed

QD

RO.3

A s

pous

e ca

nse

ek a

por

tion

of t

he p

ar-

ticip

ant’s

ann

uity

and

/or

asu

rviv

or a

nnui

ty,

but

does

not

auto

mat

ical

ly r

ecei

veth

ese

bene

fits.

Any

ann

uity

awar

d re

duce

s th

e be

nefit

sth

at t

he p

artic

ipan

t w

illre

ceiv

e.4

Pens

ion

bene

fits

are

sub-

ject

to

divi

sion

at

divo

rce.

A p

artic

ipan

t’s s

pous

e ca

nse

ek s

ome

port

ion

of t

hepa

rtic

ipan

t’s a

nnui

ty a

nd/o

ra

surv

ivor

ann

uity

indi

vorc

e pr

ocee

ding

s, b

utdo

es n

ot a

utom

atic

ally

rece

ive

thes

e be

nefit

s. A

nyan

nuity

aw

ard

redu

ces

the

bene

fits

that

the

par

tici-

pant

will

rec

eive

.

A p

artic

ipan

t’s s

pous

e w

illre

ceiv

e a

surv

ivor

ann

uity

equa

l to

not

less

tha

n 50

perc

ent

of t

he a

nnui

ty t

hat

wou

ld h

ave

been

pai

d to

the

part

icip

ant,

unl

ess

the

spou

se h

as p

revi

ousl

yw

aive

d th

is r

ight

inw

ritin

g.5

A p

artic

ipan

t’s s

pous

e w

illre

ceiv

e a

surv

ivor

ann

uity

.9

Each

chi

ld o

f a

part

icip

ant

will

rec

eive

a s

urvi

vor

annu

ity,

if th

e ch

ild is

depe

nden

t; u

nmar

ried;

and

eith

er (

1) u

nder

18,

(2

) in

capa

ble

of s

elf-

sup-

port

bec

ause

of

a di

sabi

lity,

or (

3) b

etw

een

the

ages

of

18 a

nd 2

2 an

d a

full-

time

stud

ent.

The

am

ount

of

the

annu

ity d

epen

ds o

nw

heth

er t

he c

hild

’s ot

her

pare

nt is

stil

l liv

ing.

A p

artic

ipan

t’s s

pous

e w

illre

ceiv

e a

surv

ivor

ann

uity

equa

l to

not

less

tha

n 50

perc

ent

of t

he a

nnui

ty t

hat

was

pai

d to

the

par

ticip

ant

whi

le t

he p

artic

ipan

t w

asal

ive,

unl

ess

the

spou

seha

s pr

evio

usly

wai

ved

this

right

in w

ritin

g.6

A p

artic

ipan

t’s s

pous

e w

illre

ceiv

e a

surv

ivor

ann

uity

equa

l to

no m

ore

than

55

perc

ent

of t

he a

nnui

ty t

hat

was

pai

d to

the

par

ticip

ant

whi

le t

he p

artic

ipan

t w

asal

ive,

unl

ess

the

spou

seha

s pr

evio

usly

wai

ved

this

right

.10

Fig

ure

6-A

. Sp

ou

sal R

igh

ts a

nd

Ben

efit

s in

Var

iou

s R

etir

emen

t Sa

vin

gs

and

Pen

sio

n P

lan

s (c

on

tin

ued

)

Pre-

Ret

irem

ent

Dis

trib

uti

on

at

Pre-

Ret

irem

ent

Post

-Ret

irem

ent

Dis

trib

uti

on

Ret

irem

ent

Div

orc

eD

eath

Dea

th

3 P

rivat

e pe

nsio

n pl

ans

are

not

requ

ired

to r

ecog

nize

ord

ers

that

do

not

satis

fy t

he r

equi

rem

ents

for

Qua

lifie

d D

omes

tic R

elat

ions

Ord

ers

set

fort

h in

ERI

SA. E

ven

if a

divo

rced

spo

use

is a

ble

to o

btai

na

revi

sed

orde

r in

sta

te c

ourt

, pla

ns s

omet

imes

ref

use

to r

ecog

nize

rev

ised

QD

ROs

or Q

DRO

s is

sued

aft

er a

div

orce

is f

inal

.4

U

nder

fed

eral

law

, a p

artic

ipan

t’s f

orm

er s

pous

e is

elig

ible

to

rece

ive

a su

rviv

or a

nnui

ty if

he

or s

he h

ad b

een

mar

ried

to t

he p

artic

ipan

t fo

r at

leas

t on

e ye

ar.

5

A p

lan

may

spe

cify

tha

t be

nefit

s w

ill n

ot b

e pa

yabl

e to

a p

artic

ipan

t’s s

urvi

ving

spo

use

unle

ss t

he p

artic

ipan

t an

d sp

ouse

hav

e be

en m

arrie

d fo

r on

e ye

ar p

rece

ding

the

dea

th o

f th

e pa

rtic

ipan

t.

6

A p

lan

may

spe

cify

tha

t be

nefit

s w

ill n

ot b

e pa

yabl

e to

a p

artic

ipan

t’s s

urvi

ving

spo

use

unle

ss t

he p

artic

ipan

t an

d sp

ouse

hav

e be

en m

arrie

d fo

r on

e ye

ar p

rece

ding

the

com

men

cem

ent

of a

nnui

typa

ymen

ts o

r th

e de

ath

of t

he p

artic

ipan

t.7

In

198

7, f

eder

al e

mpl

oyee

s w

ere

give

n th

e op

tion

to e

lect

cov

erag

e un

der

eith

er t

he F

eder

al E

mpl

oyee

s’ R

etire

men

t Se

rvic

e (F

ERS)

or

the

Civ

il Se

rvic

e Re

tirem

ent

Syst

em (C

SRS)

. U

nles

s ot

herw

ise

indi

cate

d, t

he e

lem

ents

of

the

CSR

S an

d FE

RS p

ensi

on p

lans

set

for

th in

thi

s se

ctio

n of

the

cha

rt a

re id

entic

al.

8

Ther

e is

no

cons

ent

requ

irem

ent

for

the

form

of

paym

ent

of a

n an

nuity

fun

ded

by v

olun

tary

con

trib

utio

ns; a

par

ticip

ant

may

, but

nee

d no

t, r

educ

e hi

s or

her

hig

her

annu

ity in

ord

er t

o pr

ovid

e a

surv

ivor

ann

uity

for

his

or

her

spou

se.

9

Und

er C

SRS,

the

wid

ow o

r w

idow

er’s

surv

ivor

ann

uity

will

be

equa

l to

55 p

erce

nt o

f th

e an

nuity

tha

t th

e pa

rtic

ipan

t w

ould

hav

e re

ceiv

ed. U

nder

FER

S, t

he s

urvi

vor

annu

ity w

ill b

e eq

ual t

o ei

ther

50 p

erce

nt o

f th

e pa

rtic

ipan

t’s a

nnua

l rat

e of

pay

plu

s $1

5,00

0 or

50

perc

ent

of t

he a

nnui

ty t

hat

wou

ld h

ave

been

pai

d to

the

par

ticip

ant,

dep

endi

ng h

ow lo

ng t

he p

artic

ipan

t w

orke

d fo

r th

e fe

dera

lgo

vern

men

t.10

In

add

ition

, at

the

time

of r

etire

men

t, a

par

ticip

ant

may

ele

ct t

o re

duce

his

or

her

annu

ity in

ord

er t

o pr

ovid

e a

surv

ivor

ann

uity

for

a b

enef

icia

ry w

ith a

n in

sura

ble

inte

rest

in t

he p

artic

ipan

t’sin

com

e. S

pous

al c

onse

nt is

not

req

uire

d, u

nles

s th

e pa

rtic

ipan

t se

eks

to p

rovi

de a

sur

vivo

r an

nuity

for

his

or

her

spou

se u

nder

the

insu

rabl

e in

tere

st a

nnui

ty p

rovi

sion

s. C

urre

nt s

pous

es, b

lood

or

adop

ted

rela

tives

clo

ser

than

firs

t co

usin

s, f

orm

er s

pous

es, s

omeo

ne t

o w

hom

the

par

ticip

ant

is e

ngag

ed, a

nd c

omm

on la

w s

pous

es a

re p

resu

med

to

have

an

insu

rabl

e in

tere

st; f

or a

ll ot

hers

, the

par

-tic

ipan

t m

ust

subm

it ev

iden

ce s

how

ing

the

exte

nt t

o w

hich

the

ben

efic

iary

is d

epen

dent

on

the

part

icip

ant

and

“the

rea

sons

why

the

nam

ed b

enef

icia

ry m

ight

rea

sona

bly

expe

ct t

o de

rive

finan

cial

bene

fit f

rom

the

con

tinue

d lif

e of

the

em

ploy

ee o

r m

embe

r.”

The

exac

t am

ount

of

the

insu

rabl

e in

tere

st a

nnui

ty is

not

set

for

th b

y st

atut

e, b

ut t

he p

artic

ipan

t’s a

nnui

ty m

ay n

ot b

e re

duce

d by

mor

eth

an 4

0 pe

rcen

t to

pay

for

suc

h an

ann

uity

.


CSR

S Th

rift

Sa

vin

gs

Plan

11W

hen

a pa

rtic

ipan

t ap

plie

sfo

r in

-ser

vice

loan

s or

with

-dr

awal

s, t

he p

artic

ipan

t’ssp

ouse

mus

t be

not

ified

.

Onc

e a

part

icip

ant

has

left

the

civi

l ser

vice

, he

or

she

may

with

draw

fun

ds f

rom

the

TSP

acco

unt

in a

sin

gle

or m

onth

ly p

aym

ent,

or

asan

ann

uity

.12If

a pa

rtic

i-pa

nt w

ho h

as le

ft c

ivil

serv

-ic

e ap

plie

s fo

r a

with

draw

al a

nd h

is o

r he

rTS

P ac

coun

t ba

lanc

e is

mor

e th

an $

3,50

0, t

hepa

rtic

ipan

t’s s

pous

e m

ust

be n

otifi

ed w

ithin

a y

ear

ofth

e ap

plic

atio

n.

If th

e TS

P ba

lanc

e is

less

than

$3,

500,

a p

artic

ipan

tw

ho h

as le

ft c

ivil

serv

ice

may

app

ly f

or a

with

draw

alw

ithou

t no

tifyi

ng h

is o

r he

rsp

ouse

.

If a

part

icip

ant

who

has

left

civ

il se

rvic

e ap

plie

s fo

ra

with

draw

al a

nd h

is o

rhe

r TS

P ac

coun

t ba

lanc

e is

mor

e th

an $

3,50

0, t

hepa

rtic

ipan

t’s s

pous

e m

ust

be n

otifi

ed.

If th

e TS

P ba

lanc

e is

less

than

$3,

500,

no

notic

e is

requ

ired.

A p

artic

ipan

tw

ho h

as le

ft c

ivil

serv

ice

may

with

draw

TSP

fun

dsw

ithou

t no

tifyi

ng h

is o

r he

rsp

ouse

.

TSP

fund

s ar

e su

bjec

t to

divi

sion

at

divo

rce.

Con

sequ

ently

, a

part

ici-

pant

’s sp

ouse

can

see

kso

me

port

ion

of T

SP f

unds

in d

ivor

ce p

roce

edin

gs,

but

does

not

aut

omat

ical

lyre

ceiv

e a

shar

e. A

ny a

war

dre

duce

s th

e fu

nds

that

the

part

icip

ant

will

rec

eive

.

If a

part

icip

ant

dies

bef

ore

elec

ting

a pa

yout

, th

efu

nds

will

be

dist

ribut

ed t

oth

e de

sign

ated

ben

efic

iary

.Th

e pa

rtic

ipan

t m

ay d

esig

-na

te a

nyon

e he

or

she

choo

ses

as t

he b

enef

icia

ry.

Thus

, a

part

icip

ant’s

spo

use

will

rec

eive

TSP

fun

ds a

tth

e pa

rtic

ipan

t’s d

eath

on

ly if

des

igna

ted

as t

hebe

nefic

iary

. 13

If a

part

icip

ant

dies

aft

erel

ectin

g to

rec

eive

his

or

her

TSP

acco

unt

bala

nce

asa

join

t-an

d-su

rviv

or a

nnui

tyfo

r th

e pa

rtic

ipan

t an

d hi

sor

her

spo

use,

the

spo

use

will

rec

eive

the

TSP

fund

s.14

If a

part

icip

ant

dies

aft

erel

ectin

g to

rec

eive

his

or

her

TSP

acco

unt

bala

nce

asa

sing

le o

r m

onth

ly p

ay-

men

t, a

sin

gle-

life

annu

ity,

or a

join

t-an

d-su

rviv

oran

nuity

for

the

par

ticip

ant

and

a be

nefic

iary

with

an

insu

rabl

e in

tere

st,

none

.15

If a

part

icip

ant

elec

ted

ajo

int-

and-

surv

ivor

ann

uity

,th

e sp

ouse

will

rec

eive

asu

rviv

or a

nnui

ty e

qual

to

50 p

erce

nt o

f th

e an

nuity

that

was

pai

d to

the

par

tic-

ipan

t w

hile

the

par

ticip

ant

was

aliv

e.

If a

part

icip

ant

elec

ted

tore

ceiv

e hi

s or

her

TSP

acco

unt

bala

nce

as a

sin

gle

or m

onth

ly p

aym

ent,

a s

in-

gle-

life

annu

ity,

or a

join

t-an

d-su

rviv

or a

nnui

ty f

orth

e pa

rtic

ipan

t an

d a

bene

-fic

iary

with

an

insu

rabl

ein

tere

st,

none

.16

Fig

ure

6-A

. Sp

ou

sal R

igh

ts a

nd

Ben

efit

s in

Var

iou

s R

etir

emen

t Sa

vin

gs

and

Pen

sio

n P

lan

s (c

on

tin

ued

)

Pre-

Ret

irem

ent

Dis

trib

uti

on

at

Pre-

Ret

irem

ent

Post

-Ret

irem

ent

Dis

trib

uti

on

Ret

irem

ent

Div

orc

eD

eath

Dea

th

11

The

Thr

ift S

avin

gs P

lan

(TSP

) for

fed

eral

em

ploy

ees

is s

imila

r to

a 4

01(k

) pla

n fo

r pr

ivat

e em

ploy

ees.

FER

S, b

ut n

ot C

SRS,

TSP

par

ticip

ants

rec

eive

aut

omat

ic a

genc

y co

ntrib

utio

ns t

o th

eir

TSP

acco

unts

as

wel

l as

agen

cy m

atch

ing

fund

s.12

T

he p

artic

ipan

t m

ay e

lect

a s

ingl

e-lif

e an

nuity

, a jo

int-

and-

surv

ivor

ann

uity

for

the

par

ticip

ant

and

his

or h

er s

pous

e, o

r a

join

t-an

d-su

rviv

or a

nnui

ty f

or t

he p

artic

ipan

t an

d a

form

er s

pous

e or

othe

r in

divi

dual

with

an

insu

rabl

e in

tere

st in

the

par

ticip

ant’s

inco

me.

The

par

ticip

ant

may

als

o el

ect

for

the

paym

ent

or p

aym

ents

to

occu

r at

a f

utur

e da

te. T

he p

artic

ipan

t m

ay c

hang

e or

can

cel a

with

draw

al e

lect

ion

befo

re t

he w

ithdr

awal

is d

isbu

rsed

, but

a C

SRS

part

icip

ant’s

spo

use

mus

t be

not

ified

.13

A

ltern

ativ

ely,

if t

here

is n

o de

sign

ated

ben

efic

iary

, his

or

her

spou

se w

ill r

ecei

ve t

he f

unds

; if

ther

e is

no

spou

se, t

he p

artic

ipan

t’s c

hild

or

child

ren

(or

desc

enda

nts

of d

ecea

sed

child

ren)

will

rece

ive

the

fund

s.14

T

he s

pous

e w

ill r

ecei

ve t

he T

SP f

unds

as

eith

er a

n an

nuity

or

a lu

mp

sum

, dep

endi

ng o

n w

heth

er a

n an

nuity

had

bee

n pu

rcha

sed

at t

he t

ime

of d

eath

. 15

If

the

par

ticip

ant

elec

ted

to r

ecei

ve a

sin

gle

or m

onth

ly p

aym

ent,

the

spo

use

or o

ther

fam

ily m

embe

r w

ill r

ecei

ve t

he T

SP a

ccou

nt b

alan

ce if

des

igna

ted

as t

he b

enef

icia

ry b

y th

e pa

rtic

ipan

t—or

,if

ther

e is

no

bene

ficia

ry, a

ccor

ding

to

the

stat

utor

y or

der

of p

rece

denc

e. If

the

par

ticip

ant

elec

ted

but

had

not

purc

hase

d a

sing

le-li

fe a

nnui

ty, t

hen

the

spou

se o

r ot

her

fam

ily m

embe

r w

ill r

ecei

ve t

heTS

P ac

coun

t ba

lanc

e if

desi

gnat

ed a

s th

e be

nefic

iary

by

the

part

icip

ant—

or, i

f th

ere

is n

o be

nefic

iary

, acc

ordi

ng t

o th

e st

atut

ory

orde

r of

pre

cede

nce.

If t

he p

artic

ipan

t se

lect

ed a

join

t-an

d-su

rviv

oran

nuity

with

a b

enef

icia

ry w

ith a

n in

sura

ble

inte

rest

, the

spo

use

or o

ther

fam

ily m

embe

r w

ill r

ecei

ve t

he T

SP a

ccou

nt b

alan

ce if

des

igna

ted

as t

he b

enef

icia

ry b

y th

e pa

rtic

ipan

t.

16

If t

he p

artic

ipan

t el

ecte

d to

rec

eive

a s

ingl

e or

mon

thly

pay

men

t an

d TS

P fu

nds

rem

ain,

the

spo

use

or o

ther

fam

ily m

embe

r w

ill r

ecei

ve t

he T

SP a

ccou

nt b

alan

ce if

des

igna

ted

as t

he b

enef

icia

ryby

the

par

ticip

ant—

or, i

f th

ere

is n

o be

nefic

iary

, acc

ordi

ng t

o th

e st

atut

ory

orde

r of

pre

cede

nce.

If t

he p

artic

ipan

t el

ecte

d a

sing

le-li

fe a

nnui

ty, b

enef

it pa

ymen

ts w

ill b

e m

ade

acco

rdin

gly.

If t

he p

artic

i-pa

nt s

elec

ted

a jo

int-

and-

surv

ivor

ann

uity

with

a b

enef

icia

ry w

ith a

n in

sura

ble

inte

rest

, the

spo

use

or o

ther

fam

ily m

embe

r w

ill r

ecei

ve t

he T

SP a

ccou

nt b

alan

ce if

des

igna

ted

as t

hat

bene

ficia

ry b

y th

epa

rtic

ipan

t.


FER

S Th

rift

Sa

vin

gs

Plan

Spou

se m

ust

cons

ent

befo

re p

artic

ipan

t re

ceiv

esan

in-s

ervi

ce d

istr

ibut

ion

orlo

an.

Onc

e a

part

icip

ant

has

left

civi

l ser

vice

, he

or

she

may

with

draw

fun

ds f

rom

the

TSP

acco

unt

in a

sin

gle

paym

ent,

in m

onth

ly p

ay-

men

ts,

or a

s an

ann

uity

.17

If th

e TS

P ba

lanc

e ex

ceed

s$3

,500

and

a p

artic

ipan

tw

ishe

s to

with

draw

TSP

fund

s af

ter

leav

ing

civi

lse

rvic

e, t

he s

pous

e m

ust

cons

ent.

Oth

erw

ise,

the

TSP

fund

s w

ill b

e di

strib

-ut

ed in

the

for

m o

f a

join

tan

d 50

per

cent

sur

vivo

ran

nuity

, un

less

the

spo

use

wai

ves

this

rig

ht.

If th

e TS

P ba

lanc

e is

less

than

$3,

500,

no

cons

ent

isre

quire

d.

If th

e TS

P ac

coun

t ba

lanc

eex

ceed

s $3

,500

, a

part

ici-

pant

’s sp

ouse

mus

t co

n-se

nt if

the

par

ticip

ant

seek

sto

rec

eive

TSP

acc

ount

fund

s in

any

for

m o

ther

than

a jo

int

and

50 p

er-

cent

sur

vivo

r an

nuity

.

If th

e TS

P ba

lanc

e is

less

than

$3,

500,

non

e. A

retir

ed p

artic

ipan

t m

ayw

ithdr

aw T

SP f

unds

with

-ou

t sp

ousa

l con

sent

.

TSP

fund

s ar

e su

bjec

t to

divi

sion

at

divo

rce.

A p

ar-

ticip

ant’s

spo

use

can

seek

som

e po

rtio

n of

TSP

fun

dsin

div

orce

pro

ceed

ings

, bu

tdo

es n

ot a

utom

atic

ally

rece

ive

a sh

are.

Any

aw

ard

redu

ces

the

fund

s th

at t

hepa

rtic

ipan

t w

ill r

ecei

ve.

If a

part

icip

ant

dies

bef

ore

elec

ting

a pa

yout

of

his

orhe

r TS

P fu

nds,

non

e. T

hefu

nds

will

be

dist

ribut

ed t

oth

e de

sign

ated

ben

efic

iary

.A

par

ticip

ant

may

des

ig-

nate

any

one

he o

r sh

ech

oose

s as

the

ben

efic

iary

.Th

us,

a pa

rtic

ipan

t’s s

pous

ew

ill r

ecei

ve T

SP f

unds

at

the

part

icip

ant’s

dea

th o

nly

if de

sign

ated

as

the

bene

ficia

ry.18

If a

part

icip

ant

has

elec

ted

a jo

int-

and-

surv

ivor

ann

u-ity

, th

e sp

ouse

will

rec

eive

the

TSP

fund

s.19

If a

part

icip

ant’s

spo

use

wai

ved

the

right

to

rece

ive

a jo

int-

and-

surv

ivor

an

nuity

, no

ne.20

If th

e TS

P ac

coun

t ba

lanc

eex

ceed

s $3

,500

, a

part

ici-

pant

’s sp

ouse

will

rec

eive

asu

rviv

or a

nnui

ty e

qual

to

50 p

erce

nt o

f th

e an

nuity

that

was

pai

d to

the

par

tic-

ipan

t w

hile

the

par

ticip

ant

was

aliv

e, u

nles

s th

esp

ouse

has

pre

viou

sly

wai

ved

this

rig

ht.

If a

part

icip

ant’s

spo

use

wai

ved

the

right

to

rece

ive

a jo

int-

and-

surv

ivor

an

nuity

, no

ne.21

Fig

ure

6-A

. Sp

ou

sal R

igh

ts a

nd

Ben

efit

s in

Var

iou

s R

etir

emen

t Sa

vin

gs

and

Pen

sio

n P

lan

s (c

on

tin

ued

)

Pre-

Ret

irem

ent

Dis

trib

uti

on

at

Pre-

Ret

irem

ent

Post

-Ret

irem

ent

Dis

trib

uti

on

Ret

irem

ent

Div

orc

eD

eath

Dea

th

17

The

part

icip

ant

may

ele

ct a

sin

gle-

life

annu

ity, a

join

t-an

d-su

rviv

or a

nnui

ty f

or t

he p

artic

ipan

t an

d hi

s or

her

spo

use,

or

a jo

int-

and-

surv

ivor

ann

uity

for

the

par

ticip

ant

and

a fo

rmer

spo

use

orot

her

indi

vidu

al w

ith a

n in

sura

ble

inte

rest

in t

he p

artic

ipan

t’s in

com

e. T

he p

artic

ipan

t m

ay a

lso

elec

t fo

r th

e pa

ymen

t or

pay

men

ts t

o oc

cur

at a

fut

ure

date

. The

par

ticip

ant

may

cha

nge

or c

ance

l aw

ithdr

awal

ele

ctio

n be

fore

the

with

draw

al is

dis

burs

ed, b

ut a

FER

S pa

rtic

ipan

t’s s

pous

e m

ust

cons

ent

(unl

ess

the

spou

se p

revi

ousl

y w

aive

d th

e jo

int-

and-

surv

ivor

ann

uity

opt

ion)

.18

A

ltern

ativ

ely,

if t

here

is n

o de

sign

ated

ben

efic

iary

, his

or

her

spou

se w

ill r

ecei

ve t

he f

unds

; if

ther

e is

no

spou

se, t

he p

artic

ipan

t’s c

hild

or

child

ren

(or

desc

enda

nts

of d

ecea

sed

child

ren)

will

rec

eive

the

fund

s.19

Th

e sp

ouse

will

rec

eive

the

TSP

fun

ds e

ither

in t

he f

orm

of

an a

nnui

ty o

r as

a lu

mp

sum

, dep

endi

ng o

n w

heth

er t

he a

nnui

ty h

ad b

een

purc

hase

d at

the

tim

e of

dea

th.

20

If th

e pa

rtic

ipan

t el

ecte

d to

rec

eive

a s

ingl

e or

mon

thly

pay

men

t, t

he s

pous

e or

oth

er f

amily

mem

ber

will

rec

eive

the

TSP

acc

ount

bal

ance

if d

esig

nate

d as

the

ben

efic

iary

by

the

part

icip

ant—

or if

ther

e is

no

bene

ficia

ry, a

ccor

ding

to

the

stat

utor

y or

der

of p

rece

denc

e. If

the

par

ticip

ant

elec

ted

a si

ngle

life

ann

uity

but

an

annu

ity h

as n

ot b

een

purc

hase

d, t

hen

the

spou

se o

r ot

her

fam

ily m

embe

rw

ill r

ecei

ve t

he T

SP a

ccou

nt b

alan

ce if

des

igna

ted

as t

he b

enef

icia

ry b

y th

e pa

rtic

ipan

t—or

if t

here

is n

o be

nefic

iary

, acc

ordi

ng t

o th

e st

atut

ory

orde

r of

pre

cede

nce.

If t

he p

artic

ipan

t se

lect

ed a

join

t-an

d-su

rviv

or a

nnui

ty w

ith a

ben

efic

iary

with

an

insu

rabl

e in

tere

st, t

he s

pous

e or

oth

er f

amily

mem

ber

will

rec

eive

the

TSP

acc

ount

bal

ance

if d

esig

nate

d as

tha

t be

nefic

iary

by

the

part

icip

ant.

21

If

the

part

icip

ant

elec

ted

to r

ecei

ve a

sin

gle

or m

onth

ly p

aym

ent

and

TSP

fund

s re

mai

n, t

he s

pous

e or

oth

er f

amily

mem

ber

will

rec

eive

the

TSP

acc

ount

bal

ance

if d

esig

nate

d as

the

ben

efic

iary

by

the

part

icip

ant—

or if

the

re is

no

bene

ficia

ry, a

ccor

ding

to

the

stat

utor

y or

der

of p

rece

denc

e. If

the

par

ticip

ant

elec

ted

a si

ngle

-life

ann

uity

, ben

efit

paym

ents

will

be

mad

e ac

cord

ingl

y. If

the

par

tici-

pant

sel

ecte

d a

join

t-an

d su

rviv

or-a

nnui

ty w

ith a

ben

efic

iary

with

an

insu

rabl

e in

tere

st, t

he s

pous

e or

oth

er f

amily

mem

ber

will

rec

eive

the

TSP

acc

ount

bal

ance

if d

esig

nate

d as

tha

t be

nefic

iary

by

the

part

icip

ant.

Prep

ared

by

the

Nat

iona

l Wom

en’s

Law

Cen

ter,

Oct

ober

200

2. T

he C

ente

r ap

prec

iate

s th

e co

mm

ents

of

Pam

ela

Peru

n.

7Chapter

Disabled Workers and their Families

143

Introduction

When considering individual accounts as part ofSocial Security, it is important to take account ofdisabled-worker beneficiaries1 and their families.In 2003, about 16 percent of all Social Securitybeneficiaries were disabled-worker beneficiariesand their dependent children or spouses.Changes in basic disabled-worker benefits wouldaffect other beneficiaries, including former dis-abled workers who are reclassified as retireeswhen they reach normal retirement age, childrenand widowed spouses of deceased disabledworkers, widows in old age who receive benefitsbased on the work record of a spouse disabledbefore he died, and disabled adult children whoreceive benefits based on the work record of aparent who is deceased, disabled or retired.More information about disabled adult childbeneficiaries is in Chapter Eight.

Any changes in Social Security defined benefitsthat would affect disabled-worker beneficiariesand their families throughout the rest of theirlives should be thought through carefully. It is

important to distinguish between two reasonswhy Social Security benefits might be reduced.First, because Social Security is not in long-termfinancial balance, various proposals call forreducing traditional defined benefits as part of asolvency proposal. These benefit reductions forsolvency are not the topic of this report. We arenot assessing ways to bring Social Security intobalance, nor are we weighing tradeoffs betweenbenefit reductions and revenue increases torestore solvency. Rather, our purpose is to helppolicymakers think through payout issues thatarise in various types of proposals that createindividual accounts as part of Social Security.

Second, plans that shift Social Security taxes toindividual accounts generally call for additionalreductions in scheduled retirement benefits tophase in as the accounts build up. These so-called “benefit offsets” are typically designedwith retirement benefits in mind, and, dependingon how they are designed, could have unintend-ed effects on the benefits of disabled workers orother beneficiaries who may not share in theproceeds of the individual account. These bene-


fit offsets usually differ depending on whetherparticipation in the accounts is mandatory orvoluntary.

Consider a case in which account participationis mandatory (that is, all Social Security contrib-utors would automatically have part of theirSocial Security taxes put into individualaccounts). Of the many ways of adjusting thedefined benefit formula to accommodate thischange, one simple approach would be to grad-ually phase in reductions in the primary insur-ance amount (PIA) formula. This type of changecould have unintended outcomes for beneficiar-ies who might not benefit from creation of theindividual account, such as young disabledworkers or young survivor beneficiaries. Theseissues are discussed in this chapter.

A different set of issues arise when SocialSecurity contributors can choose whether or notto shift part of their Social Security taxes to per-sonal accounts. In this case, worker-specific off-sets would be designed to apply only to thosewho shift Social Security taxes to personalaccounts. These worker-specific offsets are dis-cussed in Chapter Nine.

Disability Issues Depend on the Purposeof Individual AccountsPayout policies for disabled workers and theirfamilies vary greatly depending on the purposeof the individual accounts. If the purpose is tosupplement Social Security defined benefits, thentreatment of accounts at the onset of disabilitycould be based on existing rules for supplemen-tal retirement savings plans – such as tax-favored individual retirement accounts (IRAs),employer-sponsored 401(k) plans, or the ThriftSavings Plan (TSP) for federal employees.

If the accounts’ purpose is to fill part of the rolethat Social Security has traditionally filled forretirees, then treatment of the account at disabil-ity onset might be more complex. A reduction inthe PIA formula as a way to accommodatemandatory creation of individual accounts withSocial Security taxes would automatically reduce

scheduled benefits for disabled workers andtheir families, unless special rules preventedthose reductions.

If special rules exempt disabled workers frombenefit cuts for retirees, then the discontinuitybetween retirement and disability benefits wouldraise new issues. When and how disabled work-ers would have access to their individualaccount funds are also key questions. Would dis-abled workers have access at disability or wouldthey be required to preserve the accounts forretirement? Would disabled workers be requiredto buy annuities under the same terms thatapply to healthy workers? Would joint-lifeannuities be required of disabled workers andtheir spouses on the same terms that apply toother married individuals? This chapter exam-ines these issues.

Many Disabled Workers Are FinanciallyVulnerableAbout 5.9 million individuals aged 18-64received Social Security disabled-worker benefitsin January 2004; their average benefit was $862a month, or about $10,000 a year. In addition,1.6 million children of disabled workers receivedbenefits, averaging $254 a month.

The test of disability in the Social Security pro-gram is strict – the worker must be unable towork because of a medically determinable physi-cal or mental impairment expected to last for atleast one year or to result in death within a year.The person must also have recent work inemployment covered by Social Security. Forthose who qualify, Social Security disability ben-efits begin five full months after the onset of thedisabling condition. In this chapter, we assumethat any individual account proposal would con-tinue this disability definition for Social Securitybenefits.

Social Security retirement benefits are normallybased on a worker’s highest 35 years of earningsover his or her lifetime. In the case of disability,benefits are based on the period of potentialwork years before the disability occurred. The

Chapter Seven: Disabled Workers and their Families 145

disabled worker’s PIA is the same as that of aretiree at normal retirement age with the sameaverage earnings.

When a disabled worker reaches the full benefitretirement age, his or her benefits remain thesame but are classified as retirement benefits.Benefits for qualifying family members of thedisabled worker are subject to a family maxi-mum cap that is lower than the maximum thatapplies in retirement and survivor cases.

When compared to other people aged 18through 64, disabled-worker beneficiaries aredisproportionately male, due in part to menbeing more likely than women to have therecent work needed to be eligible for benefits.About 60 percent of disabled workers are men,as are about 49 percent of other people aged 18-64. As women are working more continuouslythan in the past, more women will be insuredfor disability in the future. On other measures,disabled-worker beneficiaries are a relatively disadvantaged group (U.S. SSA, 2001a). Whencompared to other working aged adults, disabled-worker beneficiaries are more likely to be:

• Black or Hispanic (17 percent compared to10 percent);

• 50 years of age or older (60 percent compared to 21 percent);

• Unmarried (51 percent compared to 42 percent);

• Divorced (24 percent compared to 12 percent);

• Without a high school diploma (37 percentcompared to 13 percent);

• Without education beyond high school (75 percent compared to 48 percent);

• Living alone (23 percent compared to 11 percent).

The median adjusted family income of disabled-worker beneficiaries is about half that of otherpeople aged 18-64. Disabled workers are at highrisk of being poor or near poor, with familyincomes below 125 percent of the povertythreshold. About 34 percent of disabled workersare poor or near poor, compared to 13 percentof others aged 18-64.2

Social Security is half or more of total familyincome for about one in two (48 percent) disabled-worker beneficiaries. For nearly one infive (18 percent), Social Security is 90 percent ormore of their income (Figure 7-1).

Married disabled workers often rely, in part, oncontinued income from a working spouse, whilethose who are unmarried typically have littleincome other than Social Security. Of marrieddisabled workers, about one in three relies onSocial Security for half or more of total income.Of those who live alone, nearly four out of five(78 percent) rely on Social Security for half ormore of their total income and one in three (33percent) relies on Social Security for 90 percentor more of total income (Figure 7-2).

The Risk of Becoming Disabled IsSignificant The risk of becoming so disabled that one quali-fies for disability benefits is greater than manypeople think. About three in ten men and one infour women will become disabled before reach-ing normal retirement age, according to projec-tions by the Office of the Chief Actuary of theSocial Security Administration (Figure 7-3).

Further, disability is not the last risk to incomesecurity that workers and their families mightface. Some workers will die after starting toreceive Social Security disability benefits, leavingchildren, including disabled adult children, whowill rely on survivor benefits from SocialSecurity. And, a widowed spouse might laterturn to survivor benefits in old age based on thedisabled-worker beneficiary’s record. Over a life-time, about 8 percent of men and 5 percent ofwomen in an age cohort will receive disabled-


worker benefits and then die before reachingretirement age. Others – about 20 percent ofmen and 18 percent of women in an age cohort– will receive disability benefits and then shift toretirement benefits at the normal retirement age.These workers’ Social Security income will need

to span disability as well as retirement yearsand, perhaps, the remaining life of a spouse wid-owed in old age or a disabled adult child.Finally, a smaller number of individuals – about1 or 2 percent of an age cohort – will receivedisability benefits, then stop receiving benefitsbecause they recover or return to work, andlater come back to claim Social Security asretirees.3

Payment Options for DisabledWorkers

Individual account proposals treat disabled-worker beneficiaries in a variety of ways. Figure7-4 illustrates some of the ways disability bene-fits are addressed in selected Social Securityreform proposals.

The design of payout rules for disabled workerswill depend on the nature and purpose of theindividual account proposal. Option One –Access at Disability Onset: The IRA Approachis based on the precedent of individual retire-ment accounts (IRAs) and other savings thatsupplement Social Security.

Figure 7-2. Disabled Workers’ Reliance on Social Security by Living Arrangements

Source: U.S. Social Security Administration, 2001a. Income of Disabled-Worker Beneficiaries

Figure 7-1. Disabled Workers’ Reliance onSocial Security as a Share of FamilyIncome

6%

18%

48%

0%

10%

20%

30%

40%

50%

60%

50% or more 90% or more 100%

Perc

ent

of D

isab

led

Wor

kers

Social Security as a Share of Family Income

4%

13%

36%

0%

10%

20%

30%

40%

50%

60%

50% or more 90% or more 100%

Perc

ent

of D

isab

led

Wor

kers

Not living with relatives Living with relatives

Social Security as a Share of Family Income

70%

80%

13%

33%

78%

Source: U.S. Social Security Administration, 2001a. Income ofDisabled-Worker Beneficiaries

24367 NASI TEXT 1/17/05 11:36 AM Page 146


Five other options described here explore dis-ability payout rules in plans where accounts aremeant to partially fill the role of Social Securitydefined benefits at retirement. For this discus-sion, we assume a generic Social Security indi-vidual account plan with three basic features: (a)accounts would generally be preserved for retire-ment; (b) annuities would be the expected pay-out from the accounts at retirement; and (c)income from the accounts would replace part ofexisting Social Security retirement benefits.

If participation in the accounts were mandatory,then the conforming changes in retirement bene-fits might be accomplished by scaling back PIAsacross the board. These types of proposals arediscussed in Options Two through Six. If partici-pation were voluntary, then worker-specific off-sets might be designed in a number of ways, asdiscussed in Chapter Nine.

Option Two – Treat Disability Like Retirement,would simply apply the retirement rules of thegeneric plan to the case of disability. The lowerretirement PIA would be paid at disability onset.Option Three – Mandate Private DisabilityInsurance explores the notion of adding manda-tory private disability insurance to Option Two,and preserving the account for retirement.Option Four – Pay a Higher Disability PIA andTake Back the Account explores introducing ahigher defined benefit for disability than forretirement. In return, the disabled worker’s indi-vidual account would be turned over to the dis-

ability insurance trust fund. Option Five – Pay aHigher Disability PIA and Preserve the Accountfor Retirement seeks to avoid the potentiallyunpopular feature of taking back the individualaccount at disability. Finally, Option Six – Pay aHigher Disability PIA that Shifts to a BlendedPIA and Annuity at Retirement would, likeOptions Four and Five, pay a higher PIA for dis-ability than for retirement. At retirement, thedisabled worker would shift to a somewhatlower PIA and start receiving an annuity fromthe account. Each option is described more fullybelow.

Option One – Access at Disability Onset:The IRA ApproachThis option for disability payouts is based onthe precedent of tax-favored retirement savingsplans – such as IRAs, 401(k) plans, and theThrift Savings Plan for federal employees. In allof these systems, contributions to the accountsare new funds unrelated to Social Security. Theseaccounts serve as a second or third tier of retire-ment income and make the money available,without penalty, to account holders who suffersevere long-term disabilities. Receipt of SocialSecurity disability insurance benefits would beevidence of such a disability.

The accounts in this option would supplementSocial Security and give the account holderbroad discretion about whether and how to usethe money. The worker could withdraw themoney in a lump sum, or continue to preserve it

Figure 7-3. Probability of Becoming Disabled or Dying Before Age 67Persons Age 20 in 2004

Probability of: Men Women

Living not disabled until age 67 61 69Dying before 67, no disability 9 6Becoming disabled before age 67 — Total 30 25

Then dying before 67 8 5Remaining disabled until 67 20 18Stop receiving benefits and living to 67 2 2

Source: U.S. Social Security Administration, Office of the Chief Actuary, 2004d. Personal correspondence


Figure 7-4. Disability Benefit Rules in Selected Individual Account Proposals

Proposal Provisions Affecting Disabled-Worker Beneficiaries


ACSS (Gramlich): Individual • Reductions in the primary insurance amount (PIA) for retirees applyAccount Plan, 1996 to disabled worker benefits as well.

• Individual accounts must be preserved for retirement.

Committee on Economic • Disability benefits not specifically addressedDevelopment, 1997


Reps. Kolbe-Stenholm: Bipartisan • PIA reductions due to creation of individual accounts do not apply at disability.Retirement Security Act of 2004 • Accounts must be preserved for retirement. (H.R. 3821 in 108th Congress) • At retirement, disabled individual’s benefit is a blend of disability PIA and

retirement PIA, and individual account annuity. • Requires annuitization, such that the annuity plus the Social Security benefit will

pay at least 185 percent of the poverty line.

National Commission on • PIA reductions due to creating the accounts apply to disability benefits.Retirement Policy, 1999 • Accounts must be preserved for retirement.


Clinton Retirement Savings • No special provision. Accounts, 2000

Social Security Plus • Account holder could use funds, penalty free at disability.(proposed by R.M. Ball, 11/2003)


President’s Commission (PCSSS) • Disability benefits would be subject to offset based on the annuity value of theModels 1,2, 3 (2001) account.

• No early access to account funds at disability onset.

Reps. DeMint-Armey: Social • Benefit payable to disabled workers and their dependents and survivors under Security Ownership and age 60 are not subject to an offset based on the annuity value of the account. Guarantee Act of 2001(H.R. 3535 in 107th Congress)


Rep. Shaw: Social Security • Individuals would receive a distribution of 5 percent of their account as a Guarantee Plus Act of 2003 lump-sum distribution upon entitlement to disability benefits. (H.R. 75 in 108th Congress)



for other emergencies later. Since unexpectedexpenses often accompany the onset of disabili-ty, the funds in the individual account couldhelp meet everyday living expenses during thefive-month wait for disability benefits, or beused to help cover health insurance continuationpremiums for the additional 24-month waitbefore Medicare begins,4 or be used for otherpurposes.

In the following options, the individual accountis a mandatory and integral part of a SocialSecurity reform plan. As is common in manyproposals that call for individual accounts inSocial Security, we assume that the basic featuresof the Social Security plan for retirement wouldinclude: (a) generally preserving the individualaccounts for retirement; (b) annuities would bethe expected payout at retirement; and (c)income from the account would be expected toreplace part of existing Social Security retire-ment benefits. Each proposal raises new issuesabout adequacy, complexity, and incentives; thefinal option seeks to balance all three.

Option Two – Treat Disability LikeRetirementUnder this approach, a disabled worker wouldhave access to the individual account at disabili-ty onset and would receive the reduced retire-ment PIA for life. The symmetry of treatingdisability like retirement is simple and appealingon the surface, but this approach might be prob-lematic if the overall goal of the reformed SocialSecurity system is to achieve basic income ade-quacy. The disabled worker’s individual accountbalance would likely be smaller than that of asimilar earner who worked until retirement age.Workers who become disabled in their 30s, 40s,or even 50s might not have enough in theiraccounts to produce the intended level ofincome in retirement, much less for years of dis-ability prior to retirement.

The form of payment from the account wouldpose new questions. Permitting a lump sum pay-ment would leave the disabled individual onlythe reduced Social Security benefit as a source of

monthly income. Phased withdrawals are anoth-er option, but because the length of the worker’sremaining lifetime is particularly uncertain, itwould be difficult to allocate the withdrawalsprudently. The purchase of life annuities at dis-ability onset would also pose many issues. Withthe great variability in life expectancy for dis-abled individuals, pricing annuities might beproblematic. Death rates in the early years afterdisability onset are very high, with about one infour disabled workers dying within five years offirst receiving disability benefits (Mashaw andReno, 1996b). Other disabled workers live along time into retirement.

Option Three – Mandate PrivateDisability InsuranceThis option would require that the individualaccount be preserved for retirement and wouldpay the reduced Social Security retirement PIAat disability. Here, a portion of all workers’ indi-vidual account contributions must be used topurchase long-term disability insurance (LTDI).In theory, the supplemental disability insurancewould take effect when a worker was found eli-gible for Social Security disability benefits.

Private group long-term disability insurance isprovided by some employers either through self-insuring the benefits or by purchasing coveragefrom an insurance company. In 2003, about 28percent of private sector workers had such cov-erage that was financed, at least in part, by theiremployers (U.S. Department of Labor, BLS,2004). Some employment-based LTDI plans arefinanced solely by employees so that benefits,when paid, are tax-free. The typical private poli-cy calls for replacing roughly 60 percent of aworker’s prior earnings and the payments areusually reduced dollar for dollar by SocialSecurity disability benefits. If Social Security dis-ability benefits were lowered, private employersor the insurers that provided LTDI would pickup more of the cost unless they modified theirplans to adapt to Social Security benefit changes.

White-collar employees are more likely thanservice or blue-collar workers to have private


disability coverage. Forty percent of professionaland technical workers had long-term disabilitycoverage in 2003, but only about 20 percent ofblue-collar and 10 percent of service workerswere covered (U.S. Department of Labor, BLS,2004). As Social Security’s weighted benefit for-mula pays lower levels of wage replacement tohigher earners, supplementary private LTDI ben-efits become a larger share of total benefits forhigher earners. Private insurers use their owndefinitions of disability to determine eligibilityfor LTDI benefits. In some cases, the test is lessstrict than the Social Security definition.

In brief, existing private LTDI differs fromSocial Security disability benefits in manyrespects. Private insurance does not use aweighted benefit formula and generally does notpay additional benefits for dependent children ofdisabled workers. Pricing of private disabilityinsurance does not reflect a broad risk pool thatincludes everyone, and private group disabilityinsurers have little or no experience insuring theroughly three-quarters of private sector workerswhose employers do not offer this coverage. Inaddition, many people with disabilities whowork would be uninsurable under private dis-ability plans. If the goal of the Social Securityindividual account plan were to mitigate oravoid benefit offsets for disabled workers andtheir families, then it might be simpler to do sothrough adjustments in the plan directly.

Option Four – Pay a Higher Disability PIAand Take Back the Account

At disability onset, this option would avoid anyacross-the-board reduction or offset in the PIAthat would normally apply at retirement due tothe creation of the individual account. In return,the individual account balance would be turnedover to the DI trust fund to help finance theunreduced PIA. All future Social Security bene-fits for the disabled worker and his or her familymembers or survivors would be based on theunreduced “disability” PIA. The individualaccount would be gone.

Policymakers would need to decide what to doin the case of a disabled worker who stopsreceiving disability benefits because of recoveryor return to work. When a disabled individualgoes back to work, would he or she begin con-tributing to a new individual account? Wouldthe higher disability PIA still prevail when he orshe later retires? Or would the trust fund giveback the account with interest when the individ-ual returns to work, or at retirement? About 1or 2 percent of individuals in an age cohortstopped receiving disability benefits because theyno longer meet the medical criteria for disabilityor because they recovered and/or returned towork and later claimed retirement benefits(Figure 7-3).

This option conflicts with the notion of individ-ual accounts as personal property and might notbe popular among people who place great valueon owning the account. The more the account isviewed as personal property, the more difficult itmight be to require that it be turned over to thetrust funds at the onset of disability.

Option Five – Pay a Higher Disability PIAand Preserve the Account for RetirementThis option aims to avoid the potentially unpop-ular feature of taking back the account; it wouldpay an unreduced disability PIA for life, and itwould require that the individual account bepreserved until normal retirement age, when theaccount would be annuitized.

For workers who become disabled in their 50sor early 60s, this option would produce consid-erably higher old-age benefits than if they hadnot received disability benefits because: (a) thebenefit would be the higher disability PIA that,in this option, is payable for life; and (b) forthose disabled late in their careers, the annuitiesbased on their individual accounts would bealmost as large as that for a comparable retiree.There is some concern that this option couldraise new incentives to claim disability benefitsinstead of early retirement benefits, which arereduced under current law (Larin andGreenstein, 1998).


Early Social Security retirement benefits claimedbetween age 62 and the normal retirement ageare actuarially reduced. The reduction at age 62has historically been 20 percent but is scheduledto increase to 30 percent when the normalretirement age rises to age 67 for people born in1960 and later. The availability of unreduceddisability benefits at age 62 and older could bean incentive to claim disability benefits insteadof early retirement benefits.5 This option wouldfurther widen the disparity between disabilityand early retirement benefits beyond what isalready forecast for the future.

Option Six – Pay a Higher Disability PIAand Shift to a Blended PIA Plus anAnnuity at RetirementThis option, like Options Four and Five, wouldbase benefits for the disabled worker and his orher family on a disability PIA that is not reducedbecause of the creation of the individualaccount. And, the individual account would notbe available until the disabled worker reachesnormal retirement age, at which time it wouldbe annuitized under the rules that apply to otherretirees. If the disabled worker dies before nor-mal retirement age, the account would go to hisor her heirs according to the rules that apply tonon-disabled workers. Additionally, defined ben-

efits for surviving children would be based onthe higher disability PIA.

A unique feature of this plan is the treatment ofthe PIA at retirement. When the disabled workerreaches normal retirement age, he or she wouldshift to a blended PIA that would be a weightedaverage of the lower retirement PIA (after anoffset due to the individual account) and thehigher disability PIA. The relative share of eachwould depend on the portion of the work lifethat the individual was disabled. For example, ifJohn had been disabled one-fourth of his poten-tial working-age years, his blended PIA at nor-mal retirement age would be 25 percent of hisunreduced disability PIA, plus 75 percent of hisreduced retirement PIA, as illustrated in Figure7-5. In addition, he would receive an annuityfrom the individual account. Presumably, theblended PIA at normal retirement age wouldbecome the basis for any family member’s bene-fits payable after that time. This type of feature isincluded in the Bipartisan Retirement Security Act(HR 3821, 108th Congress) introduced byRepresentatives Kolbe and Stenholm (Figure 7-4).

Figure 7-6 illustrates the blended PIA approachif John were able to work only one-quarter ofhis potential working life before becoming disabled.

Figure 7-5. Blended PIA for Worker Disabled One-Quarter of Working Life

Disability PIA Retirement PIA Blended PIA IA Annuity

At time of retirement $900 $700 $400

Adjustment—disability 1/4

of potential working life:

25% of disability PIA $225

75% of reduced retirement PIA $525

Total blended PIA* $750

Full IA annuity $400

Total Retirement Income** $1,150

*Disability PIA and reduced retirement PIA

**Blended PIA and full IA annuity


Under this option, the disabled worker’s incomewould change when he or she reaches normalretirement age; it could go up or down, depend-ing on the difference between the disability PIAand the blended PIA, and the size of the annuityfrom the individual account.

Option Six could have major implications forfamily members. The option of using a blendedPIA might prove beneficial for survivors andother family members. In both of the aboveexamples, survivor benefits based on the blend-ed PIA would be higher than benefits based onthe reduced retirement PIA. Since children mustbe under age 18 or 19 to qualify for benefits ona disabled worker’s record, it is likely that mostchildren’s benefits would be based on the unre-duced disability PIA (the blended PIA is notapplicable until the worker reaches full retire-ment age). Yet, people who receive benefits asdisabled adult children based on their parents’work histories would have their benefits basedon the reduced blended PIA once the parentretired. See Chapter Eight for more informationabout disabled adult child beneficiaries.


Issues discussed in other chapters are pertinentwhen considering disabled workers. For exam-ple, sustaining a ban on early access to themoney (as discussed in Chapter Five) might bereconsidered if the account holder is disabled.Similarly, mandating joint-life annuities for mar-ried individuals (as discussed in Chapter Three)might take on a different cast if one (or both) ofthe spouses is converting from disability toretirement benefits at normal retirement age.

Sustaining a Ban on Disabled AccountHolders’ Access to the FundsChapter Five points out the difficulty of sustain-ing a ban on account holders’ access to themoney when they need it. The case for allowingaccess might become more compelling if theworker were disabled. As noted earlier, disabledworkers are financially vulnerable and mustwait at least five months after they were nolonger able to work to receive benefits thatreplace a fraction of their prior earnings. Also,disabled workers must wait another two yearsbefore they gain Medicare coverage. A hardshipcase for access to the individual account might

Figure 7-6. Blended PIA for Worker Disabled Three-Quarters of Working Life









Total Retirement Income** $625




be compelling, particularly if these workersexpect to die before retirement.

Exceptions for the Terminally Ill

One rationale for banning early access to thefunds is that workers will need the money forretirement. But if an account holder is terminallyill, this rationale would not be convincing.Should policymakers allow an exception to theban on access if an account holder is terminallyill?

There have been proposals to modify the tradi-tional Social Security disability program for terminally ill applicants – for example, by short-ening or eliminating the waiting period for cashbenefits or Medicare. While the Social SecurityAdministration has expedited claims proceduresfor terminally ill claimants, no new or differentbenefits are accorded to designated individuals.Recurring questions on offering new or differentbenefits include: Who would designate terminalillness and how? What is the recourse or out-come when the individual does not die asexpected?

Exceptions for Unmarried Individuals

Another rationale for banning access to theaccount is that the money would be needed toprovide some protection for an account holder’sspouse. Again, this rationale is not compellingfor account holders who are single and terminal-ly ill. Should policymakers consider early with-drawals when account holders are terminally illand have no spouse or dependents?

Would Loans or Viaticals Be Allowed?

Presumably, if disabled workers were notallowed to withdraw funds, loans would be dis-allowed as well. Otherwise, a loan could easilyturn into a withdrawal, simply by failure torepay the loan.

Account holders might instead turn to viaticalsettlements as a way to access the funds. Somestate insurance rules allow viatical settlements,which help policyholders access funds availableonly following their deaths. Under a viatical set-

tlement, the owner of a life insurance policy(viator) sells the policy for a percentage of thedeath benefit. The buyer of the policy, or viaticalprovider, becomes the new owner and deathbeneficiary. The entity that represents the seller,the viatical broker, can shop for viatical offersand is paid a commission by the viaticalprovider (Iowa Insurance Division, 2004). Therehave been cases in which individuals with life-threatening illnesses have sold their insurancepolicies in return for immediate cash. A healthyperson might also sell a life insurance policy toget cash. In brief, if policymakers intend to banaccess to individual accounts, they might alsoconsider rules about new financial products thatcould evolve to circumvent the ban on access.

Disability and Mandatory Joint-LifeRetirement AnnuitiesChapter Three discussed at some length issuesabout requiring that accounts be used to buyretirement annuities and requiring that marriedretirees buy joint-life annuities. Some of thoseissues take on new dimensions with regard toindividuals who enter retirement as disabledworkers or spouses of disabled workers.

Would Disabled Workers Be in the AnnuityPricing Pool?

In 2002, 11 percent of the individuals claimingSocial Security retirement benefits did so afterreceiving disability benefits prior to retirement(U.S. SSA, 2004a). Would these retirees be in theannuity pricing pool on the same terms as otherretirees? Ideally, the appeal to consumers of riskpooling in annuities means that everyone in thepool has an equal chance of at least being aver-age, and a chance of at most living longer thanthe next person. For disabled retirees, the oddsmight work against them. If all account holderswere required to purchase annuities at retire-ment, and if disabled retirees were included in auniversal pool of annuitants, the terms of theannuity purchase might be unfavorable for dis-abled retirees. An individual account plan might,however, create a special disabled retirees annu-ity pool that provides more favorable pricing forthis group.


Annuity products are available in the UnitedKingdom (UK) that take into consideration indi-vidual medical and lifestyle situations. Forinstance, standard annuities are priced withoutregard to medical or lifestyle information, while“smoker annuities” are based purely uponsmoking history and “enhanced annuities” areunderwritten on medical and lifestyle informa-tion. Figure 7-7 illustrates the annual incomegenerated from approximately $55,000 for botha male and female smoker and non-smoker.6

Most annuity providers in the UK offer onlystandard annuities, but the move toward greaterpricing refinement allows those with lower lifeexpectancy to access improved annuity rates. Itis possible, however, that provision of lifestyleannuities might lead to higher-cost standardannuities for those without evidence of substan-dard life expectancy.

Compulsory Joint-Life Annuities When OneSpouse Is Disabled

Compulsory joint-life annuitization of individualaccount balances might prove disadvantageousto disabled workers and their spouses comparedto non-disabled workers. Aside from concernsabout pricing unfairness discussed earlier, twoadditional outcomes need to be considered thatapply specifically to disabled workers and theirspouses. First, a disabled worker’s individualaccount might be significantly smaller than itwould have been at normal retirement age, pro-ducing a much smaller annuity, which would bereduced even further due to a joint-survivormandate. And second, if the disabled workerdies earlier than most non-disabled workers (asis sometimes the case), the disabled worker’s

spouse would need to live on a survivor annuityfor more years than would a non-disabled work-er’s spouse.

More Choices In The Case of Disability

If disabled beneficiaries were allowed to opt inor out of retirement annuity requirements, or ifthey were provided with more types of guaran-tees or more types of joint-life annuities, itwould be important that participants understandtheir choices and how the different optionswould affect their long-term financial security.Additionally, if one group of account holders,such as disabled workers, receives specialoptions not afforded all account holders, law-makers might be pressured to expand the specialoptions to the entire universe of individualaccount holders. Further, any special optionsdesigned specifically to enhance the economicsecurity of disabled workers might inadvertentlyencourage more workers to file for disabilitybenefits.

Summary

At the beginning of 2004, almost 6 million indi-viduals aged 18-64 received disabled-workerbenefits from Social Security. Social Securityaccounts for half or more of total family incomefor about one in two (48 percent) disabled-worker beneficiaries. In designing a SocialSecurity plan that involves individual accounts,it is important to think through how theaccounts, with any accompanying changes inSocial Security defined benefits, would affect dis-abled workers and their families.

Figure 7-7. Annual Income From $55,000 UK Lifestyle Annuity Purchase

Male Female

Smoker Non-Smoker Smoker Non-Smoker

Age at Purchase60 $3,733 $3,418 $3,476 $3,39565 $4,383 $3,768 $3,954 $3,74670 $5,281 $4,250 $4,589 $3,987


Policy issues on payouts for disabled-workerbeneficiaries will vary depending on the purposeof the individual accounts. If the accounts areintended to be discretionary supplemental sav-ings on top of Social Security, then payout rulescould be based on existing supplemental savingsplans. All such existing plans make the money inthe account available, without penalty, at dis-ability onset.

Yet, if individual accounts are intended toreplace part of Social Security, and retirementbenefits are offset on a mandatory basis toaccommodate the accounts, then new issuesarise about whether and how the retirement off-sets apply to disability benefits and how andwhen the account funds become available.

Many of the issues discussed in earlier chapterstake on new dimensions if beneficiaries haveexperienced career-ending disabilities. The chal-lenge of sustaining a ban on access to the fundsbefore retirement age might be revisited when adisabled individual has a pressing need for themoney – particularly if there is no spouse orthere are no children. Mandating joint-life annuities for married retirees could present newissues if one or both members of the couplewere disabled workers. Many key questionsarise for integrating individual accounts withprovisions for disabled workers and their families.


Chapter Seven Endnotes

1 Although it is preferable to refer to people withdisabilities rather than disabled people, the term“disabled-worker” is a term of art in the SocialSecurity system. It refers to workers who areinsured for disability benefits who became enti-tled to benefits when they were found to beunable to work for at least a year due to theonset of a disabling physical or mental condition.

2 Family income is adjusted for family size andcomposition using the equivalence scale built intothe official poverty index, which takes intoaccount economies of scale and different needs ofchildren and adults. Family income is divided bythe ratio of the family’s poverty threshold to theone-person poverty threshold.

3 The return-to-work rate over five to six years fol-lowing receipt of disability benefits is higher foryounger disabled workers; of those who beganreceiving disability benefits in their 30s, nine per-cent recovered or returned to work within sixyears. Testimony of Virginia Reno before theSocial Security Advisory Board’s DiscussionForum on the Definition of Disability, April 14,2004; based on data provided by the SocialSecurity Administration in the mid-1990s inMashaw and Reno, (1996a).

4 Federal law requires employers with 20 or moreemployees who provide health insurance to offercontinuation coverage during the Medicare wait-ing period to former employees who become dis-abled. The employer can charge 150 percent of

the full group premium. In 1999, the averagepremium for individual insurance coverage wasestimated to be roughly 40 percent of the averagedisabled-worker benefit. The family coveragepremium was roughly equal to the average dis-ability benefit (Fronstin and Reno, 2001).

5 The availability of disability benefits at earlyretirement ages under current law does notappear to entice many older workers to claimdisability benefits in lieu of reduced early retire-ment benefits. The number of people who arenewly awarded disability benefits declines at age62, and continues to decline at ages 63 and 64(U.S. SSA, 2002a). Even though early retirementbenefits are reduced, they might be more attrac-tive than disability benefits for the following reasons: (a) no medical evidence or exams arerequired; (b) there is no five-month waiting peri-od after earnings have stopped; (c) partial retire-ment is possible under the retirement earningstest, while earnings above a low threshold causedisability benefits to end; (d) the benefit for fami-lies of three or more is higher for retirees thanfor disabled workers; and (e) disability benefitsare offset for workers’ compensation, whileretirement benefits are not (Mashaw and Reno,1996b).

6 The annuities in this table for non-smokers arebased on rates from Norwich Union, and theannuities for smokers are from BritannicRetirement Solutions, both as of September2003. Information provided by Alec Findlater,2003.

8Chapter

Children, Life Insurance, and Bequests

157

Introduction

Individual accounts as part of Social Securityreform are generally designed with retired work-ers in mind. Less attention has been paid to ben-eficiaries whose eligibility is based on a familyrelationship or disability. The previous twochapters focused on rights of spouses and onbenefits for disabled workers within an individ-ual account system. This chapter focuses on chil-dren, both minor children and those disabledadult children who receive benefits on the basisof their parents’ work history.1

Social Security is most commonly known forpaying retirement benefits to workers who havecontributed to the system throughout theirworking careers. But Social Security also pro-vides life and disability insurance in that it paysbenefits to children under the age of 18 (19 ifthe child is still in school) and to disabled adultchildren when a parent dies, becomes disabled,or retires. It is important to examine if, andhow, new accounts might interact with SocialSecurity benefits for children since assets in indi-

vidual accounts are not expected to spread riskthe way insurance does.

This chapter examines current Social Securitybenefits for children and how those benefitsmight change with the creation of an individualaccount system. It also considers four possiblepayment options for children in young survivorfamilies, what rules might apply to defined bene-fits for children of disabled and retired workers,what, if any, rights minor or disabled adult chil-dren would have to their parents’ individualaccounts, and issues about bequests other thanto spouses and children. The chapter presents aprofile of disabled adult children with the aim ofaiding policymakers in deciding what rulesshould apply to this group.

Current Social Security Benefitsfor Children

About three million children under age 18received Social Security in December 2002 (U.S.SSA, 2004a). These child beneficiaries representabout 7 percent of all Social Security beneficiar-


ies and 4 percent of all children in the UnitedStates (U.S. SSA, 2002a). About half (49 per-cent) of child beneficiaries are survivors ofdeceased workers, 38 percent are dependents ofdisabled workers, and about 12 percent are chil-dren of retired workers (U.S. SSA, 2002a).Another 2 million children live in families inwhich another member receives benefits fromSocial Security, for a total of 5 million childrenwho receive part of their family income fromSocial Security (Hill and Reno, 2003).

Adults who have been severely disabled sincechildhood (before age 22) are eligible for bene-fits on the same terms as minor children. Theybecome eligible for benefits when a parent dies,becomes disabled, or retires. About 749,000 dis-abled adults received these benefits in December2002. Like minor children, disabled adult childbeneficiaries could be considered for special pro-tections in the design of an individual accountplan that is part of Social Security.

Current Benefits for Young SurvivorFamiliesSocial Security is the main source of life insur-ance for most families with children. Almost allU.S. jobs (96 percent) are covered by SocialSecurity (U.S. SSA, 2002a). According to actuar-ies at the Social Security Administration, SocialSecurity protection had a net present valueequivalent to a life insurance policy with a facevalue of $403,000, and a disability policy with apresent value of about $353,000 for a youngaverage earner with a spouse and two youngchildren in 2001 (U.S. SSA, 2001b).2

Workers qualify for Social Security life insurancefairly quickly.3 About 98 percent of Americanchildren would receive Social Security benefits ifa working parent died.

The average monthly Social Security benefit fora widowed mother with two or more childrenwas $1,909, or about $22,900 a year, in January2004 (U.S. SSA, 2004a). The benefits keep pacewith inflation and continue until the child reach-es age 18, or 19 if he or she is still in high

school. Benefits for young survivor families arebased on the same primary insurance amount(PIA) formula used for retirement benefits.Surviving children are eligible for a benefit equalto 75 percent of the deceased workers’ PIA. Awidowed spouse caring for an eligible child isalso eligible for 75 percent of the PIA unless heor she has remarried.4 A family maximum limitsthe total monthly benefits payable to a family ofthree or more.

Social Security benefits are based on a progres-sive formula that replaces a higher proportion ofthe earnings of a lower earner than a higherearner, as discussed in Chapter Two.Consequently, the system pays higher benefits tofamilies of high earners but benefits replace agreater share of the income lost by families oflow earners (Figure 8-1).

For example, a low-earner’s child would have abenefit that replaced about 46 percent of lostearnings, while a worker who had alwaysearned the maximum amount taxed and countedfor Social Security benefits would produce achild survivor benefit that replaced about 22percent of lost earnings. If three or more chil-dren (or two children and a widowed spouse)were eligible for benefits, the replacement rateswould range from 92 percent of lost earningsfor a low-earning worker to 51 percent of lostearnings for a worker who had earned the maxi-mum amount.

The Social Security Act defines the parent-childrelationship broadly. All biological and adoptedchildren, as well as stepchildren in many cases,are eligible for Social Security in the event of aparent’s death. Children of unmarried parentsare also eligible; if paternity is disputed, SocialSecurity follows state law. If a child’s parents aredeceased or disabled, a child can receive benefitsbased on the work record of a custodial grand-parent. Grandchildren who are adopted by theirgrandparents can receive benefits based on agrandparent’s work record as long as a parent isnot living in the same household as the grand-parent and adopted grandchild.

Chapter Eight: Children, Life Insurance, and Bequests 159

Possible Changes in Children’s DefinedBenefits with Individual AccountsIn considering the impact on Social Security ben-efits paid to children, it is important to distin-guish between two reasons why scheduled SocialSecurity benefits might be reduced. First,because Social Security is not in long-term finan-cial balance, various proposals call for reducingtraditional defined benefits as part of a solvencyproposal. These benefit reductions for solvencyare not the topic of this report. Second, plansthat shift Social Security taxes to individualaccounts generally call for offsets in scheduledSocial Security retirement benefits that wouldphase in as the accounts build up. These benefitoffsets differ in design depending on whetheraccount participation is mandatory or voluntary.

When participation is mandatory, some plansphase in a reduction, or offset, in the primaryinsurance amount (PIA) formula as the accountsbuild up. This type of change could have unin-tended results for beneficiaries who might notreceive payments from the individual account,such as children. This chapter considers how

proposals for mandatory accounts financed withSocial Security taxes might be modified if policy-makers wanted to avoid applying offsets to ben-efits of surviving children and disabled adultchildren.

A different set of issues would arise when partic-ipation in the accounts is voluntary. In this case,worker-specific offsets are designed to applyonly to workers who chose to shift their SocialSecurity taxes to personal accounts. These worker-specific offsets are discussed in ChapterNine.

Defined-Benefit PaymentOptions for Young SurvivorFamilies

The design of payment rules for young survivorfamilies in individual account proposals willdepend on the nature and purpose of theaccounts. The first option is based on an indi-vidual retirement account (IRA) model that issupplemental to Social Security. The otheroptions explore potential outcomes for chil-

Figure 8-1. Social Security Benefits for Children of Deceased Workers Compared to Past Earnings, 2004*

"low" "medium" "high"

Benefit for 1 surviving child

Earnings Benefit for 3 or more

Earnings Amount

$0

$40,000

$60,000

$80,000

$100,000

$20,000

"maximum"

92%46%

$14,400

$13,200$6,600

83%

34%

$32,000

$26,000

$10,800

66%

28%

$51,200

$34,000

$14,500

51%

22%

$83,400

$42,200

$18,000

*For family of deceased worker age 40 in 2004.5

Source: U.S. Social Security Administration, Office of the Chief Actuary, 2004d. Personal correspondence

24367 NASI TEXT 1/17/05 11:36 AM Page 159


dren’s benefits in plans that shift Social Securitytaxes to individual accounts on a mandatorybasis and phase in lower defined benefits forretirees. These approaches would also apply tosurvivors who are disabled adult children.

Option One – The IRA ApproachOption One – The IRA Approach – is based onthe precedent of tax-favored retirement savingsthat supplement Social Security, such as IRAs,401(k) plans, and the Thrift Savings Plan forfederal employees. All these systems have volun-tary contributions that represent new fundsunrelated in any way to Social Security. Theaccounts are on top of Social Security and oftenon top of supplemental pensions. Money inthese accounts becomes available to a designatedbeneficiary when a worker dies before retire-ment, a treatment that could apply to individualaccounts that follow an IRA model wholly inde-pendent of Social Security.

The remaining three options assume mandatoryaccounts are financed with Social Security taxesand scheduled Social Security retirement benefitsare scaled back as the accounts phase in.

Option Two – Shield Young SurvivorFamilies from PIA Reductions Option Two would shield children of deceasedworkers from any PIA reductions due to the cre-ation of individual accounts. This would meanhaving a PIA calculation for benefits paid toyoung survivor families that differs from the oneused for retirees. Similar approaches were dis-cussed in Chapter Seven with regard to disabled-worker beneficiaries. The purpose of this optionwould be to avoid benefit reductions for youngfamilies due to creation of the accounts becausethe families would not be expected to receive lifeinsurance protection from the accounts.

Option Three – Mandate Purchase ofPrivate Life Insurance This option would use a scaled-back PIA to payyoung survivor families and require that a por-tion of each worker’s individual account contri-

butions be earmarked to buy life insurance forhis or her family.

About 69 percent of families have some kind oflife insurance (Figure 8-2). About half of allfamilies have term life insurance, which providescoverage for a specified period, such as one year.When the term ends, coverage ends unless thepolicy is renewed. Whole life policies are lesscommon. About 28 percent of families ownwhole life insurance. Whole life insurance oftenincludes a cash value component that policy-holders can borrow against. The typical value ofthese accounts is modest. In 2001, the mediancash value of whole life insurance was $10,000(Copeland, 2003a).

In some cases, workers receive life insurance aspart of their benefit package on the job. Abouthalf (54 percent) of private-sector employeeshad life insurance through their employers in2000 (U.S. Department of Labor, BLS, 2003).These policies typically pay a lump sum atdeath, either a pre-determined flat amount or apayment based on a multiple of the deceasedworker’s annual salary. About 2 percent of pri-vate-sector workers had life insurance coveragethat would pay on-going monthly benefits.

Social Security and private life insurance differin important ways. Social Security pays monthlybenefits to each child, resulting in higher bene-fits to larger families up to the family maximum.Social Security also replaces a higher proportionof income for low earners, as shown in Figure 8-1. Private life insurance pays a lump sum direct-ly related to the premium. Private life insurancepools mortality risk among policyholders, butdoes not pay extra benefits for larger families,nor is it designed to pay more relative to contri-butions for lower earners. In a voluntary privatelife insurance market, insurers typically groupindividuals with similar risks and charge higherrates to those with greater risk. Social Securitypools children’s risk of having a working parentdie among all contributors, including those withyoung children, the childless, and those whosechildren have grown. The social insurance sys-


tem reflects a view that basic income security forchildren when a working parent dies is a sharedresponsibility for society.

To require purchase of private life insurance as areplacement for part of Social Security survivorbenefits brings into clearer relief the distinctionsbetween social insurance concepts and privateinsurance principles. For this option, a key ques-tion for policymakers is which set of principleswould or should prevail if individual accountswere to be part of Social Security.

Option Four – Mandatory Accounts withNo Special Provisions for YoungSurvivorsOption Four –Mandatory Accounts with NoSpecial Provisions for Young Survivors – wouldapply a plan’s general rules for retirement pay-outs to young survivor families. The scaled-backretirement PIA would be used to calculate bene-fits for children in young survivor families.Unlike retirees, these children would not haveaccount annuities to supplement the scaled-backmonthly benefits. Instead, they would experiencethe full change in monthly income reflected inthe PIA adjustment. Many individual accountproposals stipulate that the account could bebequeathed to a widowed spouse to be pre-served for her or his retirement. Even if theaccount were immediately available to theyoung family, assets in the account would gener-

ally not substitute for Social Security life insur-ance protection for the children. Further, insome cases the eligible surviving children live inseparate households with different custodial par-ents. In this case, the assets in the account mightnot be distributed to families in the same waythat Social Security benefits to children wouldbe allocated.

Payment Options for Childrenof Retired and DisabledWorkers

Minor children and disabled adult children ofretired and disabled workers also receive SocialSecurity. Like benefits for child survivors, theirbenefits are based on the working parents’ pri-mary insurance amount. The benefit amount forchildren of retired or disabled workers is 50 per-cent of the parent’s PIA. A key issue is whetherpolicymakers wish to apply any offset in theparent’s PIA (due to creation of the individualaccount) to the children’s benefits.

Chapter Seven discussed various ways that dis-abled workers could be exempted from benefitcuts due to the creation of individual accounts.If the disabled worker were exempt from thePIA cuts, then presumably his or her eligiblechildren would also be exempt.

Figure 8-2. Percent of Families with Life Insurance, 2001

Percent Owning

Number Any Life Term Life Whole Life(in thousands) Insurance* Insurance Insurance

All Families 106,496 69 52 28Family IncomeLess than $19,999 25,321 41 29 15$20,000-$49,000 37,380 68 51 26$50,000-$99,999 28,327 85 67 35$100,000 or more 15,466 88 71 42*Includes families with both types of life insurance.



When retirees have their defined benefits offsetbecause part of their Social Security taxes wereshifted to individual accounts, policymakerscould make a separate decision about whetherthe changes in those retirees’ benefits shouldapply to the benefits of his or her children. Theretiree would have had a full work life to accu-mulate assets in the account and would presum-ably have an annuity from the account, but thechildren might not share in that annuity.

Children’s Rights to theirParents’ Individual Accounts

Another set of questions with regard to childrenand proposals for retirees’ individual accounts iswhether, and to what extent, children wouldhave any special rights to, or claims on, theirparents’ accumulated individual accounts.

As discussed in Chapter Six on Spousal Rights,wives and husbands often have certain minimuminheritance rights under state law. Further, theEmployee Retirement Income Security Act of1974 (ERISA), as amended, spells out additionalrights of spouses to pensions accumulated undertax-favored employer-sponsored retirementschemes. Would, or should, children, includingdisabled adult children, have any special rightsto individual accounts in the event that a parentdied before retirement?

Would minor children or disabled adult chil-dren have any special rights to inheritance ofindividual account balances?

Most individual account plans specify inheri-tance rights only for widowed spouses and onlyfor a current spouse. If there were no spouse,most plans would let the account holder pickanyone he or she chose as the death beneficiary(or beneficiaries). In this case, the account hold-er could name someone other than dependentchildren, or name only some children and leaveout others. State laws have varied rules withregard to children’s rights, generally defining therights of children if a parent dies intestate. If adeceased individual leaves his or her estateentirely to a spouse, children do not have clear

rights to that estate. If Congress establishes asystem of individual accounts, a key question iswhether to have uniform federal policies con-cerning children’s rights or to leave the issues tostate jurisdiction.

Would policymakers want the individualaccount plans to specify any special inheritancerights for minor children or disabled adult chil-dren? If accounts were considered part of SocialSecurity life insurance protection, then a casewould exist for trying to direct the account todependent children, and this would require fed-eral law. Yet, if the accounts are not consideredpart of Social Security life insurance, then shield-ing children from reductions in Social Securitybenefits due to the accounts’ creation becomesmore important. Here, the children’s inheritancerights outside of Social Security might be left tostate law as they are today.

Would a widowed spouse with children havethe option to use inherited funds for imme-diate needs?

Many plans stipulate that accounts bebequeathed to a widowed spouse and, further,the widowed spouse must preserve the individ-ual account for her or his retirement. The fundswould not be available to help with immediatespending needs following the death of the work-er or to care for children. In these plans, if theaccount went to anyone other than a widowedspouse, the heir could use the funds as he or shewished. The greater choices available to otherheirs could lead to pressure to ease the restric-tions on use of inherited funds by widowedspouses. These issues are discussed further inChapter Six.

If children and a widowed spouse live in sep-arate households, how would interests ofeach be accommodated?

Balancing the rights of children and widowedspouses could be an issue, as many children donot live in the same household as a survivingspouse. A study by the National Center forEducation Statistics found that in 1996 only 57percent of students in grades 1 through 12 lived


with two biological parents, while the otherslived in some other family arrangement (Nordand West, 2001). If a non-custodial parent hadremarried and then died, presumably the newspouse would inherit the deceased worker’saccount. The children in another householdmight not share in the inheritance, unless specialrules were written to apply to such cases.

How would child support requirementsaffect distribution of accounts at a parent’sdeath?

If a worker who was subject to court-orderedchild support died, would the custodial parentor child have any claim on the account of thedeceased? Under existing precedents, the custo-dial parent would have a claim against the entireestate for any past due child support at the timeof death. Settling this and other debts of theestate would occur before bequests could bemade. In most states, death ends a non-custodialparent’s future child support obligation, yetSocial Security benefits are intended to coverfuture costs. If the account is mandatory andreplaces Social Security benefits, policymakersmay wish to make arrangements to protect thechildren. For that reason, divorce decrees mightorder the non-custodial parent to take out a lifeinsurance policy for the benefit of the child.

Bequests to Heirs other thanSpouses and DependentChildren

Almost all individual account proposals allowaccount holders to bequeath their funds to heirsif death occurs before retirement. At the sametime, proposals financed with Social Securitytaxes usually limit or foreclose bequests by suchfeatures as mandatory annuitization, or manda-tory transfer of the account to a widowedspouse or, as suggested in the prior section, spe-cial inheritance rules for minor children. Theseconstraints on bequests are generally motivatedby a desire to preserve benefits that SocialSecurity now provides. An annuity mandate, forexample, ensures that retirees cannot outlive

their incomes. Mandates for joint-life annuitiesand for spousal inheritance of accounts general-ly seek to preserve Social Security protectionsfor widows, whether they are widowed afterretirement or before.

Given these constraints, bequests from accountsfinanced with Social Security taxes would resultin new payouts in various cases where benefitsare not paid under current Social Security rules.In particular, when unmarried persons (whethersingle, divorced or widowed) die before buyingan annuity, their accounts would go to theirnamed beneficiaries. Such bequests are consis-tent with a personal ownership view of theaccounts. From a strictly social insurance per-spective, however, bequests could be viewed as“leakage” that is outside the purpose of the sys-tem. To the extent that funds from SocialSecurity taxes are paid to heirs who would nototherwise be eligible for Social Security (such asnon-disabled adult children, siblings, other rela-tives, friends, or institutions), either more moneywould be needed to pay eligible beneficiaries ortheir benefits would be reduced in some way.

If policymakers wanted to protect children bymandating their inclusion as eligible beneficiar-ies, rules would need to be established thatdefine that eligibility. For instance, would therebe an age limit for eligible children? Wouldrequired bequests and required insurance pur-chases be prorated by age? If so, how wouldthe rule be enforced if someone died leavingchildren of significantly different ages? SocialSecurity currently pays benefits to non-disabledchildren under age 18 (or age 19 if still inschool).

In brief, there is a basic tradeoff between theownership model that calls for bequests, and thesocial insurance model that targets paymentsonly to eligible beneficiaries. While bequests areconsistent with a personal property view of indi-vidual accounts, they impose some new costs ona social insurance system.


Disabled Adult Children – A Profile

If proposals for financing individual accountsfrom Social Security taxes call for reducingdefined benefits on workers’ accounts, policy-makers would need to make explicit decisionsabout whether those changes in defined benefitsshould apply to disabled adult child beneficiar-ies. Methods to avoid reductions described earli-er in this chapter could be applied to thesebeneficiaries. This section offers a brief descrip-tion of the disabled adults who receive SocialSecurity benefits as the child of a worker whohas died, retired, or become disabled.

About 749,0006 adults who have been disabledsince childhood receive benefits based on theirparents’ PIAs; they receive the same types ofbenefits payable to minor children. About two-thirds of these beneficiaries (nearly 500,000) arechildren of a deceased working parent. Theiraverage benefit was $609 a month, in December2002, based on 75 percent of the deceased par-ent’s primary insurance amount (Figure 8-3).

The other third are children of retired or dis-abled parents who are entitled to 50 percent oftheir parents’ PIA. The average benefit for dis-abled adult children of retired workers was$465, while the average benefit for children of adisabled worker was $349 in December 2002(U.S. SSA, 2004a).

Disability standards for disabled adult child ben-eficiaries are the same as those for disabledworkers; that is, the individual is unable towork (to engage in substantial gainful activity).Mental retardation is the primary diagnosis forthe majority of individuals who receive SocialSecurity as disabled adult children; 60 percentare diagnosed with mental retardation, and 17percent have other mental disorders as a pri-mary diagnosis. Conditions affecting the nervoussystem or sensory organs are the next mostprevalent diagnosis, accounting for 12 percentof disabled adult child beneficiaries.

Individuals receiving benefits as disabled adultchildren range in age from young adults to sen-ior citizens. About 4 in 10 are under 40 years ofage, while another four in 10 are between 40and 54, and about 2 in 10 are age 55 or older.In nearly all cases, these individuals continue toreceive Social Security for the rest of their lives.

In some cases, disabled adults receive means-tested Supplemental Security Income (SSI) beforethey become entitled to Social Security when aparent retires, dies, or becomes disabled. If theSocial Security benefit is what causes them tolose SSI eligibility, they might lose automatic eli-gibility for Medicaid depending on where theylive.7 The disabled adult child will be eligible forMedicare after a 24-month waiting period. Thebenefit package in Medicare notably lacks long-term care coverage and the more complete pre-scription drug coverage that many stateMedicaid programs provide.

Figure 8-3. Disabled Adult Child Beneficiaries, December 2002

Distribution Average monthly benefit

Total number (in thousands) 749 $551Total percent 100Working parent deceased 66 $609Working parent retired 26 $465Working parent disabled 8 $349

Source: U.S. Social Security Administration, 2004a. Annual Statistical Supplement to the Social Security Bulletin, 2003


Most disabled adult child beneficiaries live withrelatives and many are poor or near poor. Aboutone in four (23 percent) had family incomesbelow the poverty threshold and fully six in ten(60 percent) had family incomes below twice thepoverty threshold in 1999 (U.S. SSA, 2002b).

About four in five disabled adult child benefici-aries receive benefits through a representativepayee. The Social Security Administrationassigns a payee when it determines that therecipient is unable to manage his or her ownfunds. The payee can be a relative of the benefi-ciary or a social service agency that is responsi-ble for helping the individual manage his or heraffairs. In brief, individuals who receive disabledadult child benefits from Social Security aresome of America’s most vulnerable citizens.

Summary

This chapter has explored various approaches toexempt young survivor families, children of dis-abled or retired workers, and adult beneficiariesdisabled since childhood, from reductions inscheduled defined benefits or PIAs that mightotherwise apply to retirees in mandatory indi-vidual account plans that are financed with cur-rently scheduled Social Security taxes. Withoutspecial attention to these categories of benefici-aries, proposals could inadvertently reduce someincome security protections for some of

America’s most vulnerable adults and millions ofminor children. Preserving defined benefit pro-tections for these groups would add to the costof individual account proposals that do not havethese protections.

The chapter also examines questions aboutwhether and to what extent young children ordisabled adult children would have any specialrights or claims on the individual accounts theirparents accumulate. If so, the question would bewhether these rights would be left to determina-tions of state inheritance laws, or specified infederal legislation creating the accounts.

Finally, individual account proposals might pres-ent the opportunity for some account holders tomake bequests to recipients other than spousesor children. In the eyes of many, these bequestsare desirable and consistent with property own-ership. Yet, from a social insurance perspective,such bequests could be viewed as leakage that isbeyond the purpose of the system. To the extentthat Social Security funds go to heirs who wouldnot otherwise be eligible for benefits (such asable-bodied children, relatives, or institutions)either more money would be needed to pay eli-gible beneficiaries or their benefits would belowered in some way. In designing payouts, poli-cymakers have the opportunity to weigh trade-offs between property rights and socialinsurance goals.


Chapter Eight Endnotes

1 Here “child” refers to the relationship, not toage. It is the nature of the parent/child relation-ship that places discussion of disabled adult chil-dren in this chapter.

2 This example is for a male worker with averageearnings at age 27 (with a wife also age 27 andchildren aged 0 and 2) who dies or becomes dis-abled in 2001. The value of the disability benefitsincludes benefits after normal retirement age andsurvivor benefits.

3 To be permanently insured for survivor protec-tion, workers need ten years of covered earnings.Younger workers have this family survivor pro-tection if they have worked at least one-fourth ofthe time since they were age 21. Workers who donot meet these tests would still have life insur-ance protection for their families if they workedin covered employment for at least six calendarquarters in the last 13 quarters (including thequarter in which they died). About 97 percent ofcovered workers ages 20-49 had earned survivorprotection for their children (U.S. SSA, 2003d).

4 Survivor benefits are subject to a retirement earn-ings test applicable to beneficiaries below normal

retirement age. Thus, a widowed mother orfather who is working might have his or her ben-efits fully withheld under this earnings test. Thewidowed parent’s earnings do not affect the chil-dren’s eligibility for benefits. Each child’s ownearnings would cause his or her benefit to bewithheld.

5 The earnings level for the maximum worker doesnot equal the tax maximum in the year prior toentitlement due to the historical ad hoc increasesof the tax maximum.

6 The Social Security Administration has foundthat some individuals who receive SupplementalSecurity Income (SSI) disability benefits might beeligible for Social Security disability benefits. Ithas a work in progress (known as the special dis-ability workload), which, over the next fewyears, might identify more people who are eligi-ble for Social Security benefits as disabled adultchildren.

7 States known as 209(b) states are allowed to usestricter eligibility criteria for Medicaid. Statutoryprotections to ensure continued Medicaid eligi-bility for people receiving disabled adult childrenbenefits are not applicable in 209(b) states.


Appendix: Profile of Disabled Adult Children

Figure 8-A1. Age of Disabled Adult ChildBeneficiaries, 2001

Number Percent

All Disabled Adult Children 736,553 100Under 25 68,191 925-29 62,650 930-34 75,849 1035-39 100,785 1440-44 110,263 1545-49 93,495 1350-54 71,871 1055-59 52,710 760 and older 100,739 14

Source: U.S. Social Security Administration, 2002b. AnnualStatistical Report on the Social Security Disability InsuranceProgram, 2001

Figure 8-A2. Age of Disabled Adult ChildBeneficiaries Awarded Benefits,2001

Number Percent

All Newly Awarded Disabled 37,700 100Adult Children

Under 20 3,100 820-24 9,300 2525-29 4,500 1230-34 5,600 1535-39 7,300 1940 or older 7,900 21

Source: U.S. Social Security Administration, 2002a. AnnualStatistical Supplement to the Social Security Bulletin, 2002

Figure 8-A3. Disabled Adult Child Beneficiaries by Diagnostic Group and Representative PayeeStatus, December 2001

Percent Percentage with aNumber Distribution Representative Payee

All disabled adult children 736,553 --- 79With diagnosis available 518,767 100 76Mental retardation 312,260 60 89Other mental disorders 89,858 17 64Diseases of the nervous system and sensory organs 60,578 12 51Congenital anomalies 6,395 1 66Injuries 8,599 2 38Other conditions 41,077 8 78

Source: U.S. Social Security Administration, 2002b. Annual Statistical Report on the Social Security Disability Insurance Program, 2001

Figure 8-A4. Poverty Status of Disabled Adult Child Beneficiaries, December 1999

Under 100 percent 100-199 percent 200-299 percent 300 percent Total of poverty of poverty of poverty or more

Number 625,172 144,711 228,184 151,389 100,982Percent 100% 23% 36% 24% 16%

Source: U.S. Social Security Administration, 2002b. Annual Statistical Report on the Social Security Disability Insurance Program, 2001


9Chapter


169

Certain proposals for Social Security individualaccounts would allow workers to shift part ofcurrently scheduled Social Security taxes to theseaccounts. These proposals generally providemeasures to compensate the Social Security trustfunds for the loss of this revenue, which is nec-essary to pay current benefits. This compensa-tion could take the form of additional revenue(such as transfers from general revenues,increased taxes, or dedicating specific tax rev-enue to Social Security) or reduced future bene-fits, or some of both. Proposals with mandatoryaccount participation could provide this com-pensation in a number of ways, includingacross-the-board benefit reductions that are notworker-specific (reductions that are not linkedto workers’ specific account contributions).Accounts with voluntary participation involveworker-specific offsets to treat participants andnon-participants equitably. This chapter exam-ines issues related to worker-specific offsets.

This chapter provides background about whatoffsets are intended to accomplish and examinesgeneric issues in how offsets might be designed.Also examined are issues in applying worker-specific offsets at retirement (including how the

offsets might apply to husbands and wives atretirement) and issues in applying worker-specif-ic offsets at other life events, such as divorce,disability, or death before retirement. Someadministrative and legal issues to be consideredwhen designing worker-specific offsets areexplored.

Intention Behind OffsettingBenefits

Benefit offsets arise in proposals that shift sched-uled Social Security taxes to individual accounts.A simple example illustrates this idea. Supposethat one dollar was shifted from John’s SocialSecurity taxes and deposited into his personalaccount. If there were no adjustment to John’straditional Social Security benefit, then Johnwould be made better off by one dollar. Johnwould still have the same benefit from SocialSecurity as he had before; in addition, he wouldhave one dollar in his individual account. As aresult of transferring this dollar to John’saccount, the Social Security trust fund balancewould be reduced by one dollar. (In the absenceof other changes, this would increase the unified


budget deficit by one dollar.) The deteriorationof Social Security’s long-run finances means thatthis one-dollar gain to John must eventuallyresult in a one-dollar cost to John or to othertaxpayers or beneficiaries, because traditionalbenefits will continue to be paid from the trustfunds.

Instead of having other taxpayers or beneficiar-ies pay for John’s one-dollar gain, policymakersmight reduce John’s future claim to traditionalSocial Security benefits by one dollar. In returnfor shifting one dollar of his Social Securitytaxes into his account, John could be required tocompensate the Social Security trust funds bygiving up the future benefits that this dollarwould have financed had it stayed in the tradi-tional system. This forgone future benefit is the“benefit offset” associated with the individualaccount. Box 9-1 describes further the relation-

ship between the account, investment returns,and offsets.

Compensation of the Social Security trust fundcould also be accomplished by shifting funds outof John’s individual account back into the trustfunds. Either offset method—reducing the indi-vidual account or reducing the future claim ontraditional Social Security benefits—could havethe same overall effect on John’s retirementincome and on the trust funds.

If the present value of the benefit offset (takinginto account interest over time) were equal tothe present value of the account contribution,then Social Security’s trust funds would be fullycompensated in future years. The timing ofSocial Security’s cash flows would still be affect-ed, however. In general, taking money out of thesystem in the near term would increase the sys-tem’s revenue needs in the near term, while

Box 9-1. Accounts, Investment Returns, and Offsets

If John’s account were invested in a manner similar to the investments of the Social Security trust funds(in long-term Treasury bonds), then John’s expected total future retirement income would be unchanged.Of course, whether or not John’s realized total retirement income would be higher or lower than what hewould have had without participating in the account depends on how he invested the dollar, how finan-cial markets performed prior to his retirement, and the design of the offset.1 If his account investmentsgenerated a higher rate of return, then account participation would increase his total retirement income;if his investments generated a lower rate of return, then account participation would decrease his totalretirement income. Thus, the combination of the offset rule and John’s portfolio choice would, together,determine the expected level and the riskiness of John’s total retirement income.

Thus far, this example assumes that the offset is designed so that it precisely compensates SocialSecurity’s trust funds in future years for the one-dollar account contribution. However, there is an impor-tant variation of this simple example to consider. Policymakers could design the benefit offset to be eitherhigher or lower than the level that would just compensate Social Security’s trust funds in present value. Ifthe benefit offset were less than one dollar, it is more likely that John would end up with higher totalbenefits by participating in the account. At the same time, the smaller benefit offset would also lead to adecline in Social Security’s finances – implying that some worker or beneficiary or other taxpayer wouldultimately have to make up the difference. Conversely, if the benefit offset were more than one dollar,then it is more likely that John would end up with lower total benefits by participating in the account.This case would bring a net improvement in Social Security’s finances – freeing up resources that wouldprovide gains to some worker or beneficiary.

Chapter Nine: Worker-Specific Offsets 171

reducing benefits in the future would reduce thesystem’s revenue needs at a later date.

Interactions between account contributions andthe level of traditional benefits arise explicitly(but not exclusively) in plans that use scheduledSocial Security taxes to finance the accounts.2

The important policy questions about the designof benefit offsets arise most clearly when accountcontributions are voluntary, as in proposals thatwould fit in quadrant four of Figure 9-1.

An important distinction must be made betweenbenefit reductions that arise due to the need toput Social Security into long-range fiscal balanceand reductions in traditional Social Security ben-efits that arise because individual accounts arereplacing traditional benefits. Many individualaccount proposals include both types of benefitreductions, but the term offset refers only toreductions due to the creation of individualaccounts. If a plan calls for mandatory individ-ual accounts, the offset could be accomplishedthrough a general benefit reduction applied toall workers. In this case, distinguishing betweenthe two types of benefit reductions becomesmore difficult, and perhaps less important, thanin a voluntary account system because thereduction for the purpose of offsetting theaccounts applies to everyone and need not becoordinated with contributions to the account atan individual level. If policymakers want to miti-gate benefit reductions for some categories ofbeneficiaries (for example, those who would notreceive payouts from the accounts), they couldadopt exceptions to any general benefit reduc-tions, as discussed in Chapter Seven, regardingdisabled workers, and in Chapter Eight, regard-ing children.

Offsets in a voluntary system would also need tobe carefully designed to achieve policymakers’particular goals. If the goals were to reduce ben-efits only for persons who participate in theindividual accounts, and ultimately to compen-sate the trust funds fully for lost revenue, thendesigners would seek to avoid two types ofunintended results: one would be to avoidreducing benefits of people who do not receivepayouts from the individual accounts; anotherwould be to ensure that the trust funds recoupenough to compensate for Social Security taxesthat were shifted out of the trust funds into indi-vidual accounts.

When traditional Social Security benefits are off-set due to the creation of individual accountsusing Social Security taxes, policymakers wouldface many of the same issues regarding whetherand how to apply reductions in traditional bene-fits to workers and their families, regardless ofwhether participation in the accounts weremandatory or voluntary. However, worker-spe-cific offsets would present more complex choic-es. The remainder of this chapter focuses onlyon voluntary plans with worker-specific offsets.

Design of Worker-SpecificOffsets

Individual account proposals that permit work-ers to shift Social Security taxes to personalaccounts vary widely in the way worker-specificbenefit offsets are designed. This section exam-ines two questions for policymakers. First,would the benefit offset be based on the actualindividual account balance, or would it be basedon a hypothetical balance calculated from con-tributions plus some predetermined interest rate?Second, would the offset reduce traditional


Nature of Participation Other Funds for Accounts Current Social Security Taxes Used for Accounts

Mandatory (1) (2)

Voluntary (3) (4)


Social Security benefits, or would it reduce theindividual account? To begin this discussion,Figure 9-2 illustrates some possible offsetdesigns.

Offsets Based on Actual IndividualAccount BalancesIf an offset were based on an actual individualaccount balance, an annuity could be calculatedfrom that balance at retirement (whether or notthe worker actually purchased an annuity), andthe worker’s Social Security retirement benefitcould be reduced by this monthly annuityamount.4 For example, Bill’s individual accountbalance of $20,000 would produce a single-life,inflation-indexed annuity for a 65-year old ofabout $124 per month. The retiree’s traditionalSocial Security benefit could be reduced, or off-set, by this amount for life. In this case, Bill’sretirement benefit of $900 (for example) wouldbe reduced to $776 a month because Bill hadshifted Social Security taxes to a personalaccount. He would also have the $20,000account, which he might use to buy the lifeannuity of $124. The distinguishing feature ofthis offset design is that the offset would bebased on the individual worker’s accumulatedSocial Security tax contributions and actualinvestment experience from his individualaccount.

An offset based on an actual account balancewould be consistent with a plan that does notallow individual account withdrawals prior toretirement. Two problems would arise if earlywithdrawals were allowed and the offset isbased on only the actual account balance thatremained at retirement. First, workers wouldhave an incentive to withdraw their funds beforeretirement to avoid the benefit offset. Second,the offset would not achieve its intended pur-pose – that is, to compensate the trust funds fortaxes shifted in the past from the trust fundsinto workers’ accounts.

To avoid these adverse effects, the calculationcould be modified to require that the offset bebased on what the actual account would havebeen worth if no withdrawals had been taken(even if withdrawals were taken). A mechanismmight be developed to track what the accountwould have been worth (and make assumptionsabout how it would have been invested) as if nomoney had been withdrawn.

Offsets Based on Hypothetical IndividualAccount BalancesA number of proposals that would permit work-ers to shift Social Security taxes to personalaccounts would base an offset on a hypotheticalaccount – that is, the offset would be based onthe value of Social Security taxes put into the

Figure 9-2. Examples of Worker-Specific Offset Designs3

Offset Reduces Offset ReducesIndividual Account Social Security Benefits

Offset Based on Actual Individual Account Balance (contributions (1) (2)plus actual account earnings) Rep. Shaw; HR 75 Rep. DeMint; HR 3177

Offset Based on Hypothetical (3) (4)Individual Account Balance PCSSS Models 1, 2, & 3(contributions plus predetermined Sen. Graham; S 1878 interest rate) Rep. Smith; HR 3055

Robert Pozen


actual account plus some predetermined interestrate. For example, the plans proposed by thePresident’s Commission to Strengthen SocialSecurity include a benefit offset based on theactual value of Social Security taxes put into aworker’s account, plus interest accumulated overa worker’s career at 3.5 percent, 2 percent, and2.5 percent over price inflation (for theCommission’s Models 1, 2, and 3, respectively).In these proposals, contributions and accumulat-ed interest are tracked in hypothetical or shad-ow accounts.

At retirement, the offset could be determined inone of two ways. In the first, a worker’s hypo-thetical account might be “annuitized,” with themonthly annuity amount deducted from theworker’s traditional Social Security benefit. Inthe second, the value of the shadow accountwould be divided by the present value of allexpected Social Security benefits payable basedon the worker’s earnings, and this offset ratewould be applied so as to reduce all benefitspaid from the worker’s earnings record.

The distinguishing feature of offsets of this typeis that they would not depend on the size of theactual individual account. A worker takingwithdrawals from her individual account wouldnot affect this type of offset, nor would the off-

set be affected by actual market returns earnedby the account.

Figure 9-3 illustrates what could happen toJoan’s total retirement income if her offset werecalculated from a hypothetical individualaccount using interest rates that are (1) higherthan her actual account investment earnings; (2)the same as her individual account investmentearnings; and (3) lower than her actual individ-ual account earnings.

In the example, Joan’s traditional Social Securitybenefit is $1,200 and the annuity value of heractual individual account is $300 per month. Ifthe returns on the hypothetical account exactlymatched the returns on her actual individualaccount, her offset would reduce her traditionalbenefit by $300 and her combined retirementincome would be increased by her $300 individ-ual account annuity, resulting in no change innet retirement income (scenario 2). Yet, if thehypothetical account is larger than Joan’s actualaccount (scenario 1), her offset will be largerthan the annuity value of her account and shewill end up with lower net income from tradi-tional Social Security benefits plus her individualaccount annuity. Scenario 3 shows the oppositeresult; Joan’s hypothetical account produces anoffset that is smaller than the annuity value of

Figure 9-3. Examples of Offsets by Size of Hypothetical Account Relative to Actual Individual Account Annuity

Annuity value of hypothetical account relative to actual individual account annuity

(1) Hypothetical Account (2) Hypothetical Account (3) Hypothetical AccountOffset Larger Than Actual Offset the Same as Offset Smaller Than

Account Annuity Actual Account Annuity Actual Account Annuity

Traditional Social Security $1,200 $1,200 $1,200retirement benefit

Hypothetical annuity value -450 -300 -150reduces Social Security benefit

Net traditional Social Security benefit $750 $900 $1,050

Plus actual individual account annuity +300 +300 +300

Total retirement income $1,050 $1,200 $1,350


her actual account, and she ends up with highernet retirement income.

The specified offset rate used to calculate thevalue of the hypothetical account, which differsdepending on the proposal design, is generallyset so that plan participants would have achance of achieving a better realized yield ontheir actual account investments (but that stillfully compensates the Social Security trustfunds). How an account actually performs, ofcourse, depends on a participant’s investmentchoices and on market performance.

Offsets That Reduce Traditional SocialSecurity BenefitsAs shown in Figure 9-2, proposals that permitworkers to shift Social Security taxes to individ-ual accounts generally apply offsets against tra-ditional Social Security benefits. Proposals mayalso include an offset provision when accountsare financed from other sources, such as fromunspecified general revenue as in the case of theShaw proposal. These offsets could be based onthe actual individual account balance – as inquadrant (2) – or on a hypothetical account bal-ance using the total value of the contributions tothe account plus a predetermined interest rate –as in quadrant (4).

Offsetting traditional Social Security benefits hasadvantages and disadvantages. If the intentionof the offset is to compensate the trust funds, infull or in part, for Social Security taxes shiftedto individual accounts, it may seem logical toreduce individual account holders’ future SocialSecurity benefits in exchange for their reducedparticipation in traditional Social Security. Thisoffset application creates a direct link betweenSocial Security taxes, traditional benefits, andindividual accounts. However, because SocialSecurity also provides life insurance (to survivingspouses and children of deceased workers) anddisability insurance, reducing an account hold-er’s traditional Social Security benefits couldmean that these non-retiree benefits are alsoreduced. If the offset were designed to reducetraditional Social Security benefits, policymakers

would need to decide if all the benefits payablefrom a given worker’s record should be reducedor only some of the benefits.

Offset Applied Only to the Account Holder’sBenefit

One approach would apply the offset only tothe benefits of the worker who shifted SocialSecurity taxes into a personal account and leaveunaffected the benefits of that worker’s familymembers. This result could be achieved byreducing just the account holder’s monthly bene-fit, while the spouse (or widowed spouse) andchildren of that worker (including disabled adultchildren) would have their Social Security bene-fits calculated from the worker’s unreduced pri-mary insurance amount (PIA).5 This approachwould require a larger monthly offset againstthe account holder’s retirement benefit (torecoup the value of Social Security taxes plusinterest) than if the offset applied to all benefitspaid from the worker’s earnings record, thoughthe overall offset obligation would be the same.There are no examples of current individualaccounts proposals utilizing this approach.

The primary drawback of this approach occurswhen married workers are required to purchasejoint and two-thirds survivor annuities fromtheir individual accounts at retirement. If thesurvivor’s annuity is less than the survivor’s off-set, the surviving spouse’s retirement income willbe lower than when both individuals were alive.

Offsets Applied to the Primary InsuranceAmount

If offsets were applied to the PIA, traditionalbenefits for the account holder and for all familymembers would be reduced. A proposal by Rep.Kasich (HR 5659, 106th Congress), for exam-ple, would apply an offset to the PIA, by reduc-ing traditional Social Security benefits one-thirdof 1 percent for every year of potential contribu-tions to the individual account. In this case, allworkers with the same years of potential contri-butions to the accounts would have the samepercentage reduction in their traditional SocialSecurity benefits. And, all qualifying family


members of such workers would have the samepercentage reduction in their benefits becausefamily benefits are based on the worker’s PIA.This means that any traditional Social Securitybenefits paid to spouses, ex-spouses, survivingspouses, and qualifying children (including dis-abled adult children) would be reduced whetheror not the beneficiary received a payout fromthe individual account.

If an offset were applied to the PIA, two work-ers with equal earnings and contributions totheir individual accounts might have very differ-ent expected total amounts of offset against tra-ditional benefits. For example, a worker with aspouse and children who all receive SocialSecurity benefits would expect to have a muchgreater total dollar offset than a single worker(because all family benefits would be offset),even though the accumulation in their individualaccounts would be the same.

Applying an offset to an account holder’s PIAwould also result in a lower-earning spouseabsorbing two offsets. For instance, if Joan qual-ified for Social Security retirement benefits basedon her own work history, and for a supplemen-tal benefit based on her husband’s work record,her own retirement benefit would be reduced byher own account and her spousal benefit wouldbe reduced based on her husband’s reduced PIA.

Rules could be adopted that exempt certain ben-eficiaries from the offset, as in Rep. DeMint’sSocial Security Savings Act of 2003 (HR 3177),which stipulates that retired worker and agedsurvivor benefits would be offset, but SocialSecurity benefits paid to survivors (other thansurviving spouses age 60 or older) would not besubject to an offset.

Offsets Applied to Expected Total TraditionalBenefits

A third approach would apply the offset not tothe PIA or to just the account holder’s monthlybenefit, but would calculate the offset on thetotal expected future benefits payable from theworker’s earnings record, including benefits to

qualifying spouses and children, and apply theoffset to all benefits paid. Models 2 and 3 of thePresident’s Commission to Strengthen SocialSecurity follow this approach. Steps in calculat-ing this type of offset include: (a) computing thevalue of the participating worker’s offset atretirement (based either on the hypotheticalaccount balance or the actual account balance);(b) calculating the present value of all future traditional Social Security benefits expected tobe paid from the worker’s earnings record,including benefits for a spouse, widowed spouse,and children; and (c) dividing (a) by (b) to deter-mine the offset rate to be applied to all SocialSecurity benefits paid from the account holder’searnings record. This approach results in asmaller percentage reduction when family mem-bers are eligible for traditional Social Securitybenefits, because it is intended to produce thesame expected total dollar amount of offset asfor a single worker. This concept is illustrated inthe following examples for Bachelor Bob andHusband Harold, who otherwise have similarhistories of earnings and account participation.

Bachelor Bob has a hypothetical individualaccount of $30,000. His monthly Social Securitybenefit of $900 (at age 65) is equivalent to alump sum of about $145,000. His SocialSecurity benefit would be reduced by 20.7 per-cent ($30,000/$145,000), so he would receive79.3 percent of his traditional Social Securitybenefit for the rest of his life. He would alsohave his actual account balance, which could bemore or less than $30,000, depending on hisinvestment returns.

Husband Harold’s hypothetical individualaccount is also $30,000. The present value oftraditional benefits on Harold’s earnings recordwould include benefits payable to him (roughly$145,000 like Bachelor Bob) plus the expectedvalue of benefits payable to Wanda, age 59, as awife and potential widow. Wanda’s expectedbenefits would depend on her work record aswell as her age. Assuming her benefits on hiswork record have an expected value of $55,000,total benefits payable on Harold’s earnings


record would be $200,000. The reductionapplied to all traditional Social Security benefitspaid to Harold and to Wanda from Harold’swork record would be 15 percent($30,000/$200,000). Like Bachelor Bob,Husband Harold would also have his actualindividual account balance, which could bemore or less than $30,000, depending on hisinvestment returns.

Bachelor Bob’s traditional Social Security bene-fits would be subject to a greater percentagereduction than Husband Harold’s. However, ifBachelor Bob purchased a single-life annuitywith a $30,000 account at the same time asHusband Harold purchased—with the same sizeaccount—a joint-and two-thirds survivor annu-ity for himself and his younger wife, BachelorBob would receive higher monthly annuity pay-ments (see Chapter Three).

An offset also would be calculated for Wandathat would be applied to her own worker bene-fits. Wanda’s hypothetical individual accountbalance is $10,000. The present value of expect-ed traditional benefits on her work record is$100,000. Wanda’s traditional Social Securitybenefits on her own work record would bereduced by 10 percent ($10,000/$100,000).

Using an offset rate approach to reduce SocialSecurity benefits for individual account partici-pants has advantages and disadvantages. Theprimary advantage of this approach is that itattempts to equalize the total offset amount forsingle workers and married couples. Since mar-ried couples potentially receive more benefitsfrom Social Security than single workers, theiroffset percentage (or rate) is smaller because it isapplied to all benefits payable on a given workrecord.

Any offset mechanism based on marital status atthe time of calculation would need to be flexibleenough to address the issue of divorce and pos-sibly remarriage (or multiple divorces and mar-riages). For example, if Bachelor Bob marriedafter starting to receive Social Security benefits,

his new wife would be eligible to receive tradi-tional spouse and surviving spouse benefits onBob’s work record. Would those benefits be cal-culated based on Bob’s original higher offsetrate, or would the offset rate be recalculated?Different answers to this question might pro-duce very different results for Bob and his newwife and for the trust funds. If Harold had beenmarried for more than ten years prior to hismarriage to Wanda, policymakers would need todecide whether to include the potential SocialSecurity benefits payable to his ex-wife whencalculating his offset rate. If those benefits wereincluded, it would mean that any benefits paidto Harold’s ex-wife would be reduced byHarold’s offset rate. A further discussion of thecomplexities in the case of divorce continues inthe next section.

Designers of an offset rate mechanism wouldalso need to consider the case in which thelower-earning spouse was younger than thehigher-earning spouse. For instance, if Wandawas in her 50s when Harold retired, it would bedifficult to determine how much of a dual enti-tlement (if any) she would be eligible for basedon Harold’s earnings record because her ownSocial Security benefit is based on her earningsup to her own retirement. It could be the casethat when Wanda finally retires she has earnedenough during her career that she would notqualify for a spouse benefit paid from Harold’searning record. Since Harold’s offset rate wascalculated as if Wanda would receive a dualentitlement benefit, Harold’s lower offset ratewould represent a windfall for Harold (and acorresponding shortfall to the trust funds) if hisoffset rate were not recalculated.

And, finally, an offset based on family situationat the time of retirement would reduce the pre-dictability of one’s retirement income from tradi-tional Social Security benefits, makingretirement planning more uncertain.

As a general rule, if an offset were applied totraditional Social Security monthly benefits, then


it would be consistent to calculate it as either anoffset rate or as a monthly amount.

It is important to note that even if an individualaccount program were mandatory, the interac-tions between offsetting Social Security benefitsacross-the-board and paying family benefitsresults in some of the same types of outcomes asillustrated above. In other words, an across-the-board offset applied to all workers’ SocialSecurity benefits would still raise the possibilityof all non-retiree benefits being reduced, unlessrules prohibited this. The key to the offset puz-zle lies in determining the extent to which familybenefits should be affected if an offset wereapplied to traditional Social Security benefits.

An offset could be applied to the individualaccount, avoiding the complications that arisewhen family benefits are paid from any givenworker’s Social Security earnings record (butpossibly raising other complications).

Offsets That Reduce Individual AccountsAn offset could be designed to reduce a worker’sindividual account instead of his or her tradi-tional Social Security benefit. This type of offset,commonly referred to as a clawback, wouldshift at least part of a worker’s individualaccount into the Social Security trust funds. Theoffset could be designed to reduce the individualaccount in one of two ways: the account couldbe reduced by the entire offset obligation, in alump sum, or both the offset obligation and atleast part of the individual account could beconverted to “annuities” with the reductiontaken on a month-to-month basis. The offsetfunds from the individual account would beshifted to the Social Security trust funds.

While this offset design is less common in recentindividual account proposals, applying an offsetto the individual account has a few simplifyingfeatures. Offsetting the individual accountwould avoid reducing family Social Securitybenefits paid from individual account holders’earnings records. This offset method would alsoavoid the need to convert the retiree’s potential

stream of future social insurance benefits into adollar amount to calculate an offset rate, or toconvert the individual account balance into afuture benefit stream that resembles traditionalSocial Security benefits to calculate an offsetdollar amount. While such calculations could bemade, they introduce some level of complexityand uncertainty.

If a worker-specific offset were applied to thevalue of the individual account itself, it wouldbe consistent for policymakers to make the formof the offset match the form of the payout fromthe account. For example, if a retiree could takehis individual account balance as a lump sum atretirement, then it would be consistent to calcu-late and apply the offset to the personal accountas a lump sum. If the retiree were allowed totake phased withdrawals, it would neverthelessbe straightforward and consistent to apply theoffset as a lump sum before estimating thephased withdrawals. Finally, if annuities werethe required payout from an individual accountprogram, an offset could be applied either as alump sum, before the annuity is calculated, or asa monthly offset to the annuity amount basedon the total offset obligation.

Applying Offsets at other Life Events

Policymakers would also need to explorewhether, and how, offsets would apply in situa-tions other than retirement. What issues mightarise, for example, if early withdrawals are, orare not, allowed? How would offsets apply atvarious life events? For example, how mightoffsets work in plans that permit or requireaccount assets to be divided between husbandsand wives at divorce? What rules apply whenworkers die before retirement? How would off-sets apply to disabled workers?

Offsets and Early Access to AccountsSome individual account proposals allow earlywithdrawals if account balances are of sufficientsize to keep workers above the poverty level inretirement. Some proposals might also allow


access to the accounts only for certain purpos-es—for health emergencies, home purchase,higher education, starting a business, or otherreasons—as is the case with 401(k)s and the TSP(see Chapter Five for a complete discussion).

If early withdrawals were allowed and the offsetwere based on the actual individual account bal-ance at benefit entitlement, the offset wouldneed to be designed to keep track of the finalaccount balance plus any withdrawals (includingany subsequent potential earnings), to determinethe full offset value. Without this tracking mech-anism, a worker could avoid the offset by reduc-ing his or her individual account to zero beforeretirement.

An offset based on hypothetical individualaccounts avoids this adverse incentive problem.Since hypothetical account offsets would trackthe value of Social Security taxes shifted to theaccounts plus a predetermined interest rate,these types of offsets would not depend uponthe actual individual account balance in anyway, and the offset calculation would be unaf-fected by any activity in an actual individualaccount. However, policymakers would have todecide how to deal with situations in which theremaining individual account balance or the tra-ditional benefit, depending on how the offset isapplied, is insufficient to absorb the offset.

Offsets If Accounts Are Split at Divorce Some individual account proposals would per-mit courts to divide account accumulationsbetween divorcing parties; other plans mandatesuch a division. This raises policy questionsabout how the transferred account funds wouldbe counted for offset purposes. The generalnotion described above – that workers shouldnot be allowed to deplete their accounts to mini-mize their offsets – might become less clear-cut ifworkers were required to transfer part of theiraccounts to ex-spouses.

If we think of the personal account as an“asset,” then we might consider the offset basedon that account as a “debt.” If part of the asset

transfers at divorce, should part of the debttransfer as well? If so, to what benefit would thedebt (or offset) apply? Would the offset apply tothe recipient spouse’s own retirement benefit, oronly to benefits based on the account of thedonor spouse? Should the full “debt” remainwith the person who shifted Social Securitytaxes into the personal account?

If an offset (debt) were based on a hypotheticalaccount balance, while the asset transfer is basedon the actual account balance, then the hypo-thetical account balance would presumablyreflect the transfer of contributions implied bythe actual account balance transfer at divorce. Ineach case, policymakers should consider andclarify the intended results for divorcing couples.

To illustrate, assume that Harold and Wanda aredivorcing before retirement. Both Harold andWanda had individual accounts and their offsetswere based on the value of Social Security taxesshifted to their accounts plus a predeterminedinterest rate.6 The system administrator trackedthe shifted Social Security taxes and interest inhypothetical individual accounts. At divorce,their actual individual accounts were split 50-50, but the hypothetical accounts were not(Figure 9-4).

If the intent of policymakers is to recoup thevalue of the Social Security taxes (plus interest)from each worker individually, it would not benecessary to adjust their hypothetical accountsto reflect the division at divorce; Harold wouldhave an offset calculated on his hypotheticalindividual account balance of $2,900 (while hisactual individual account would be worth only$2,200 following the divorce), and Wanda’s off-set would be calculated on her hypothetical bal-ance of $900 (while her actual individualaccount would be worth $2,200 following thedivorce).

Yet, if the intent of policymakers is to considerthe value of the actual individual account aspart of total retirement income when calculatingan offset (as in the offset rate design described


above), the individual account administratorwould need to be informed of the change in theaccounts due to divorce.

A situation could arise where an offset would belarger than the benefit payable. In the aboveexample, for instance, if Wanda’s offset werecalculated on the actual individual account bal-ance following divorce, and the offset were to beapplied to her traditional Social Security retire-ment benefit, her benefit might not be largeenough to absorb the complete offset. Yet, ifHarold’s offset were based on a hypotheticalindividual account, and the offset applied towhat was left in his actual individual account,the actual account might not be large enough toabsorb the offset.

Offset policies with regard to transferringaccounts at divorce would also need to take intoaccount the possibility that spouses may havemade different decisions on participation.Because worker-specific offsets are necessaryonly in proposals in which participation is vol-untary, it would be quite possible for one spouseto have shifted Social Security taxes into a per-sonal account while the other spouse did not. IfWanda had not participated in the individualaccount plan, but she received half of Harold’saccount at divorce, would her benefits bereduced due to the presence of the individualaccount in spite of the fact that she did not (per-sonally) shift any Social Security taxes out of thetrust funds?

Any offset design would also need to stipulatewhat would happen in the case of an ex-spousequalifying for benefits on an individual accountholder’s Social Security earnings record. Forexample, suppose Harold was married to Junefor 12 years prior to his marriage to Wanda,and prior to the start of an individual accountprogram.7 Under current Social Security law,June is entitled to the same survivor benefit asWanda (based on Harold’s work record) provid-ed she has not remarried. If an offset wereapplied to Harold’s Social Security benefit suchthat any survivor benefits were reduced, June’ssurvivor benefit would be reduced due toHarold’s individual account. June would proba-bly not receive anything from Harold’s individ-ual account because June is not a qualifiedsurvivor under current joint-and-survivor annu-ity rules. The private annuities market offersmany options beyond basic joint-and-survivorprovisions, as examined in Chapters Three andFour, and provisions for surviving divorcedspouses could be added in the future. However,purchasing an annuity that provides benefits onthree lives (Harold’s, June’s, and Wanda’s) islikely to provide lower benefits to each than anannuity payable to one annuitant and survivor.Just as policymakers would have to considerhow accounts are divided in the case of divorceand possibly remarriage, the issue of how anyoffsets would be divided is also important.

Figure 9-4. Individual Account Splitting at Divorce

Harold Wanda

Balance in individual accounts up to time of divorce $3,200 $1,200(Social Security taxes plus actual earnings on investments)

At divorce, accounts split 50-50 -1000 1,000

Balance in individual accounts after divorce $2,200 $2,200

Balance in hypothetical individual accounts up to time of divorce $2,900 $900 (Social Security taxes plus predetermined interest rate)


Offsets and Disability BenefitsIt is important to consider how an individualaccount plan, and any accompanying offsets,would affect disabled workers and their familiesthroughout the rest of their lives. Chapter Sevenoutlines six options, including five ways to pro-vide for disabled workers and their families inproposals for mandatory accounts financed withSocial Security taxes. The last of these optionsseems to raise the fewest new challenges andmight be a model for worker-specific offsets in avoluntary plan when personal accounts arefinanced with Social Security taxes.

This option has the following features: (a) dis-abled workers and their families would receivetraditional disability benefits that were not offsetbecause of the creation of the individualaccount; (b) the individual account would not beavailable until the disabled worker reached nor-mal retirement age, at which time it would beannuitized under the rules that apply to otherretirees; and (c) when the disabled workerreached normal retirement age, he or she wouldshift to a blended traditional benefit that wouldbe a weighted average of the lower retirementbenefit (after applying an offset due to the indi-vidual account) and the higher “unoffset” dis-ability benefit. The relative share of each benefit

would depend on the portion of the work lifethat the individual was disabled. For example, ifJohn had been disabled one-fourth of his poten-tial working-age years, his blended traditionalbenefit at normal retirement age would be 25percent of his “unoffset” disability benefit, plus75 percent of his offset retirement benefit, asillustrated in Figure 9-5. In addition, at retire-ment he would have his individual account oran annuity from it.

If the individual account were annuitized atretirement, and if the annuity and the offsetwere based on similar calculations, the disabledworker would experience little change in month-ly income when he or she shifted from disabilitybenefits to offset retirement benefits plus a lifeannuity.

In this policy scenario (delaying application of adisabled worker’s offset until he retired), otherquestions arise if he should die before retire-ment. Would his account then go to his heirs orestate? If so, he and his estate would have borneno offset for having shifted Social Security taxesto the personal account; the trust funds wouldabsorb a net loss, and the estate would experi-ence a net gain. This issue raises the broaderquestion of death before retirement discussedbelow.

Figure 9-5. Blended PIA for Worker Disabled One-Quarter of Working Life









Total Retirement Income** $1,150




Worker-Specific Offsets and Benefits forChildren of Deceased WorkersAs discussed in Chapter Eight, Social Securityalso pays survivor benefits to children if the chil-dren are unmarried, and under age 18 (or up toage 19 if still attending elementary or secondaryschool full time). Adults disabled since child-hood are also eligible for benefits when a parentdies. Qualifying children can receive up to 75percent of the deceased worker’s basic SocialSecurity benefit. Would these benefits for chil-dren be subject to offset? Would the answer bedifferent if the children did, or did not, inheritthe account?

Offset traditional benefits for young survivors

One option would be to offset the deceased par-ent’s basic benefit (PIA) that is used to calculatebenefits to children. In this case, the benefits forsurviving children would be lower if theirdeceased parent had chosen to set aside SocialSecurity taxes in a personal account. Under thisoption, if the parent had chosen not to shifttaxes to a personal account, then life insuranceprotection for his or her children would remainintact.

Shield young survivor families from benefitoffsets

An offset could be designed so that young sur-vivors’ benefits would not be subject to any off-set, and some individual account proposals haveadopted this rule. Under this approach, surviv-ing children would not have their benefits sub-ject to an offset. All surviving children would betreated the same, whether or not the parent hadchosen to shift Social Security taxes to an indi-vidual account.

Other Situations of Death beforeRetirementAdditional questions arise about how to equi-tably apply worker-specific offsets when workersdie before any offset has been applied to theirretirement benefits. As noted above, this couldoccur if a worker died after becoming disabled,

or died leaving minor children. In this and otherscenarios involving death of the worker beforeretirement, policymakers would have a numberof policy choices to make.

(a) First, what would happen to the individualaccount? Who would get it? Would theinheritance be offset by any amount?

(b) Second, who (if anyone) would receive lowertraditional benefits because Social Securitytaxes had been shifted from the trust fundsto the individual account? Would benefits bereduced for heirs who received the account?Would benefits be offset only if traditionalbenefits were payable from the deceasedworker’s earnings record? Would the offsetand inheritance be independent?

(c) To which benefits would an offset apply?For example, if a widow inherited anaccount, would her own benefit as a retireebe offset? Would any offset apply only tobenefits payable to survivors from thedeceased worker’s earnings record? If anable-bodied adult child or sibling (whowould not qualify for Social Security benefitsfrom the deceased worker’s earnings record)inherited the account, would an offset applyto the heir’s own retirement benefit?

(d) Under what circumstances would the trustfunds recoup any of the revenue from work-ers who chose to put Social Security taxesinto individual accounts and who then diedbefore retirement?

Policymakers might have different rules depend-ing on the different circumstances under whichworkers die and whether the deceased leftdependents who had relied on the deceasedworkers’ income. Possible scenarios include:

1. Worker dies before retirement, leaving noeligible spouse or children.

2. Worker dies leaving dependent children,either minor children or one or more dis-


abled adult children, as discussed in theprior section.

3. A worker dies leaving a widowed spouse,who inherits the account.

4. A worker dies leaving a widowed spouse,who does not inherit the account (planallows for designation of another beneficiarywith or without spousal consent).

Each of these scenarios could include cases inwhich the worker (and family, if any) had beenreceiving disability benefits before death. Thescenarios could also include cases in which noSocial Security benefits had yet been paid, butsurvivor benefits would be immediately available to the children and widowed spouse.The circumstance could be that no immediatebenefits are payable, but the widowed spousemight, in the future, become eligible for benefitsas a retiree on her or his own work record, or as a widowed spouse with traditional benefitsbased on the work record of the deceased. Inanother case, if no child or widowed spouse sur-vived the worker, the account might be willed toother relatives who would have no traditionalSocial Security benefit rights on the deceasedworker’s earnings record, in which case no rev-enue would be recouped for the Social Securitytrust funds.

Some proposals that include worker-specific offsets would offset survivor benefits only forbenefits payable to aged widowed spouses.Many current proposals exempt surviving chil-dren’s benefits from any offset, and some pro-posals exempt children and/or young widowedspouses from any offset. Many proposals wouldallow bequests to individuals not eligible for traditional Social Security survivor benefits.Typically such proposals have no explicit provi-sion for the trust funds to recoup the fundsbequeathed to heirs when workers die beforeretirement.

Offset Administrative and LegalIssues

Regardless of design, offsets raise administrativeand legal issues to be considered when integrat-ing offset provisions with other payout featuresof a proposal. This section will examine some ofthe issues triggered by benefit offsets.

When to Calculate and Apply the OffsetAn offset proposal would need to specify whatevent would trigger the calculation and applica-tion of a worker-specific offset. Two potentialtriggers for calculating and applying the offsetare: (a) when retirees claim Social Security bene-fits; or (b) when retirees first take withdrawalsfrom their individual accounts.

Claiming Social Security Benefits

Workers can claim Social Security benefits atany age after 62 and there is no advantage towaiting beyond age 70 to claim benefits. If anoffset were to reduce traditional Social Securitybenefits, it would make sense to calculate theoffset no earlier than when benefits were initial-ly claimed. This policy would avoid the possibil-ity that the value of the calculated offset wouldbe out of sync with the value of Social Securitytaxes shifted to the individual account. Forexample, if John’s offset were calculated at age62, and he continued to shift Social Securitytaxes into an individual account until age 68when he actually claimed traditional benefits,the offset would fail to reflect five years ofSocial Security taxes shifted to his account.

Similarly, if the offset were applied to traditionalSocial Security benefits, then it would need to becalculated and applied no later than when bene-fits were first claimed. This policy would avoidthe possibility that retirement benefits could bereceived without offset.

Calculating and applying an offset when tradi-tional retirement benefits were initially claimedwould ensure that individual account partici-pants would not receive any benefits from SocialSecurity without incurring the offset, and that


the offset would reflect all contributions to indi-vidual accounts before benefit receipt.

Initial Withdrawal from Individual Account

If the offset were taken as a lump sum from theindividual account (instead of a monthly reduc-tion in Social Security benefits), then somemechanism would be needed to ensure thataccounts are not depleted before an offset wereapplied. It might still be necessary to apply theoffset when the worker claims Social Security,ruling out the possibility that a worker couldreceive (non-offset) Social Security benefits,while delaying indefinitely using the money inthe individual account (and incurring an offsetfrom it).

Ultimately, the fewest issues seem to arise if anoffset were calculated and applied at the time aworker filed for Social Security benefits (and notlater than when funds were first taken from theaccount). While workers could avoid an offsetby not filing for Social Security benefits, thevalue of their traditional benefits would stay inthe trust funds and be available to pay otherworkers’ benefits.

Once an offset is calculated and applied, twoapproaches are possible for any subsequentearnings. First, policymakers could simply endthe worker’s voluntary contributions to the per-sonal account and direct all of his future SocialSecurity taxes, if any, to the trust funds. Second,in the case of an offset rate, the rate could berecomputed each year by adding any additionalaccount contributions to the original calculation.

Offsets and Mandatory AnnuityThresholdsSome proposals require that retirees annuitizetheir account balances only to the point where,when combined with traditional Social Securitybenefits, the account holder would receivemonthly payments that would keep him or herabove some specified income level (see ChapterThree). This intent would need to be coordinat-ed with the application of a worker-specific off-set. For example, if an offset would reduce

John’s traditional benefits by $300 a month,then John’s level of mandatory annuitizationfrom his personal account would be $300 amonth higher after applying the offset.

Coordinating an Offset Calculation andAnnuity Purchase Policymakers might want to consider whether toencourage (or require) that the calculation of anoffset and the purchase of an individual accountannuity are coordinated in time. As discussed inChapter Three, the timing of annuity purchase –with regard to interest rates and market per-formance – could have an impact on the month-ly annuity income an individual account wouldproduce. If an offset were based on the actualindividual account, then significant variationscould occur based on timing. Workers would allwant to purchase annuities when the market andinterest rates are high (to produce higher annu-ities) and they would like to have offsets calcu-lated when the market and interest rates are low(to produce smaller offsets).

Offset AccountabilityWhen designing an offset mechanism, the deci-sion must be made whether each individualaccount holder would be legally responsible forhis or her entire offset, or if it would be assumedthat any overall target offset amount would bemet across the entire individual account popula-tion. If the intention behind benefit offsets is torecover Social Security taxes shifted to individ-ual accounts (without affecting those who didnot participate in the accounts), the offset designneeds to address what would happen if a workerdies before his or her entire offset amount wererecovered. For instance, if John’s offset werebased on a hypothetical annuity calculated fromthe total amount of Social Security taxes con-tributed to his individual account, plus interest,the total amount of taxes and interest would bethe offset amount due from John. If John diesbefore that entire offset is recovered, otherworkers—including those who did not partici-pate in individual accounts—might be called onto make up the difference.


If a pooled approach to total offset recoverywere taken, the trust funds would get back lessof an offset from workers who die early andmore of an offset from workers who live longer.A longer-lived worker would receive more inSocial Security benefits, and more from his orher individual account annuity (if annuities wererequired), and absorb more of an offset, than ashorter-lived worker.

If each individual account holder is personallyresponsible for his or her entire offset amount,provisions would be needed to recover anyremaining offset from a worker’s estate if he orshe dies before the offset is completely recov-ered. Conversely, if a worker lives long enoughthat his or her entire offset is recovered, provi-sions would be needed to restore the worker’sbenefit to the original amount.

Minimum Benefit GuaranteeSome individual account plans, such as Rep.DeMint's “Social Security Savings Act of 2003,”guarantee that the total monthly payments avail-able from the combination of the individualaccount and traditional Social Security benefitswould not be less than some specified level, suchas the scheduled traditional benefit level in cur-rent law. If a guarantee of this type were includ-ed in the proposal, the offset mechanism wouldbe expected to apply first, with a subsequentdetermination on a case-by-case basis ofwhether additional payments were needed tomeet the guarantee.

Spousal Rights Upon Divorce or DeathIf federal law determines all issues of spousalrights in individual accounts, federal law woulddetermine the allocation of offsets. But, if theplan allows some spousal rights issues to bedetermined by state law, a separate decisionwould be needed as to whether state or federallaw would determine the division of offsets.

In a divorce, state courts divide responsibilityfor a couple’s debts as well as the couple’s assets.Would state courts have jurisdiction over a cou-ple’s individual account balances as well as their

offsets? If so, it could be possible for the courtto split the balance in an account 50-50, butdecide that no offset would accompany thetransfer to the lower-earning spouse. Conversely,the state court could decide how the accountbalances would be divided, while federal lawwould determine the division of the offsets (asfederal law determines the federal tax conse-quences of assets divided in divorces).

Allowing states to make decisions about thedivision of offsets as well as balances wouldencourage state decision-makers to consider thenet value of a share of an individual accountwhen one is divided. However, leaving decisionsabout offset allocation to state judges, lawyers,and parties—in the many cases in which partieslack legal representation—poses a risk that adecree could fail to address the offset issue or doso incorrectly. Even if rules about the allocationof offsets are set by federal law, if they are basedon decisions made at the state level about thedivision of balances, it will be necessary for fed-eral rules to anticipate a range of possible orders(such as orders reserving part of an IA for thebenefit of children) or place some constraints onthe nature of the orders. Allowing states tomake these decisions would also require a timelyand accurate reporting system to inform theindividual account system administrator of thedecision. Accurate and timely reporting could beparticularly crucial if the offset design is calcu-lated using a hypothetical individual account.

Individuals, couples, lawyers, and courts wouldneed full information about how offsets affectspousal rights in order to make informed deci-sions about whether to seek a share of an indi-vidual account during a divorce, and whether towaive the right to inherit an individual account.Depending on the offset design, assessing theimpact of an offset on the value of an accountmight require putting together information heldby account administrators and the SocialSecurity Administration. An individual accountplan should ensure access to the necessary information.


Summary

Proposals that would allow workers to choosewhether or not to shift Social Security taxes toindividual accounts typically include worker-spe-cific offsets. The reasons for such offsets are tocompensate the Social Security trust funds forthe lost revenue and to distinguish equitablybetween workers who do and who do not shiftSocial Security taxes to personal accounts. Thischapter has explored some of the wide-rangingissues for policymakers to consider when design-ing worker-specific offsets if individual accountsbecame part of federal retirement policy. Werecap some of the key questions here.

In terms of basic design, should the offset reduceSocial Security benefits, or should it reduce thesize of the worker’s individual account? Offsetsthat reduce the individual account would notimpinge upon family benefits paid by SocialSecurity. Offsets that reduce Social Security ben-efits would require policymakers to decidewhich types of benefits would be reduced (retire-ment or disability) and what family benefits(spouses, widowed spouses, disabled widowedspouses, young surviving spouses and children)would be reduced due to the worker’s participa-tion in the individual accounts.

Should the offset be based on the size of theactual individual account balance, or should itbe based on a hypothetical account balance thatis based on contributions plus some predeter-mined interest rate? Offsets based on the actualindividual account balance would be influencedby market returns. Such offsets could also beaffected by pre-retirement withdrawals or loansfrom the account, unless special rules mandatedthat such activity would not affect the offsets.Offsets based on hypothetical account balanceswould be affected only by the amount of contri-butions to the account and the assumed (orhypothetical) return on those contributions.Offsets based on hypothetical accounts mightprovide a clearer relationship between the offsetand the taxes shifted to individual accounts, butthe offset could exceed the actual account bal-

ance or traditional benefit to which it is to beapplied. In addition, accurate and timely report-ing of life events would be necessary so thathypothetical and actual individual accountsreflect comparable balances when the offset iscalculated.

If an offset applies to future traditional benefits,whose benefits should be offset? Should the off-set apply only to individual workers’ future ben-efits, or should it reduce family benefits payableon the account holder’s earnings record as well?

At retirement, what event would trigger the cal-culation and application of a worker-specificoffset? Would the trigger be when the individualfirst claims Social Security benefits, or when heor she first takes funds from the individualaccount? Applying the offset when SocialSecurity benefits are first claimed would ensurethat no retirement benefits avoid an offset. Ifretirees keep working after claiming benefits,would contributions to the account end andinstead go to the Social Security trust funds?How might this rule affect incentives withregard to timing of benefit claims?

At retirement, if an offset applies to the actualindividual account, would the offset be taken asa lump sum or as a monthly amount? It wouldbe consistent to match the form of the offset tothe form of the individual account payout.When the account is annuitized, would policyrequire that the same assumptions be used tocalculate the annuity as were used to calculatethe offset against the annuity?

If the offset were applied to traditional SocialSecurity retirement benefits, a number ofapproaches are possible. First, an offset could beapplied only to the account holder’s benefit,leaving traditional spousal benefits unaffected.Second, an offset could be applied to the work-er’s primary insurance amount, reducing bothworker and spousal benefits. Third, an offsetcould be applied to total expected retirementbenefits. Each approach would produce a differ-


ent pattern of benefits for the couple and thesurviving spouse.

The third approach—applying the offset to totalexpected retirement benefits— attempts to equal-ize the total offset amount for a single workerand a married couple, while spreading the offsetbetween both members of the couple, ratherthan applying it just to the account holder’s ben-efit as the first approach would do. However,calculating the expected benefits payable on awork record is complicated by the possibility ofdivorce and remarriage and uncertainty aboutthe future earnings of a spouse who has not yetreached retirement age. Therefore, an offsetbased on expected family benefits at the timeone member of the couple retires would reducethe predictability of traditional Social Securitybenefits. Rules for couples would also need totake account of the possibility that one spousehad chosen to shift Social Security taxes to anindividual account, while the other did not.

At divorce, if the overall proposal permits courtsto divide accounts between husbands and wives,or if it mandates such a division, some conform-ing rules might be needed for worker-specificoffsets. For example, if the personal account isviewed as an “asset” in divorce proceedings,should the offset associated with that account beviewed as a “debt?” Would part of the “debt”transfer when the account is divided? Shouldthe offset obligation remain with the originalaccount holder? The rules would need to covercases in which both parties were account hold-ers, as well as cases in which only one chose tohave an account. Questions of federal versusstate jurisdiction also arise in determining howmuch discretion courts would have to divideaccounts and the offset obligations that accom-pany them. Individuals, couples, lawyers, andcourts would need full information about howoffsets affect spousal rights in order to makeinformed decisions about whether to seek ashare of an account during marriage andwhether to waive the right to inherit an account,if such options were part of the individualaccount proposal.

At the onset of disability, worker-specific offsetpolicies could be designed to exempt disabledworkers from a worker-specific offset and toapply the offset later when the disabled workerreaches retirement age. Similarly, when a workerdies leaving minor children (or disabled adultchildren), benefits for children could be exemptedfrom application of a worker-specific offset.Policymakers will need to decide whether a work-er’s decision to shift Social Security taxes to a per-sonal account should or should not affect familylife insurance protection otherwise provided bythat worker’s earnings and contribution history.

When workers die before any offset applies, pol-icymakers have a range of choices: What shouldhappen to the individual account? Who wouldreceive it? Who, if anyone, would get lower tra-ditional benefits because Social Security taxeswere shifted to the account? Would traditionalbenefits for an aged widow be reduced even ifthe widow did not inherit the account? If theperson who inherits the account is not eligiblefor benefits on the deceased worker’s record,does that heir experience any sort of offset?Under what circumstances would the trust fundsrecoup revenues that had been shifted to person-al accounts in cases where accountholders diebefore any offsets have been applied?

In conjunction with offsets, policymakers wouldalso need to decide if each individual accountholder would be personally responsible for hisor her entire offset amount and, if the planincluded a minimum benefit guarantee, how theoffset would be applied to insure the minimum.

The application of worker-specific offsets couldresult in many different outcomes. The purposeof this chapter has been to provide a broadoverview of the kinds of questions policymakerswould need to address in designing these offsetsin a system that allows workers to shift SocialSecurity taxes to personal accounts.


Chapter Nine Endnotes

1 The expected, or average, return on stocks ishigher than the expected, or average, return onbonds. This average higher return is the market’sway of compensating investors for the fact that,due to stock’s higher volatility, investing in stocksentails more risk. Actual rates of return can behigher or lower than the expected return.

2 Generally, individual account proposals that usenew funds, either by increasing Social Securitytaxes or by using transfers from general revenue,do not include offsets because Social Security’strust funds are unaffected. Representative Shaw’sSocial Security Guarantee Plus Act of 2003(H.R. 75, 108th Congress) presents an importantvariation in which accounts financed withunspecified general revenue are used to financebenefit payments scheduled from Social Securityrather than to provide a basis for offsets againstthese benefits.

3 This taxonomy does not include all possible off-set designs. For instance, former U.S.Congressman John Kasich proposed offsettingtraditional Social Security benefits by 1/3 of apercentage point for each potential year of par-

ticipation in individual accounts (H.R. 5659,106th Congress). The offset designs listed inFigure 9-2 are some common examples.

4 Alternatively, an offset could be calculated as apercent of the actual individual account balance.Another possible offset of this type, like Rep.Shaw’s, would shift into the Social Security trustfunds a calculated amount from a worker’s indi-vidual account. Representative Shaw has notreferred to this feature of HR 75 as an offset; hisplan is mentioned here for illustrative purposesonly.

5 The primary insurance amount is the basic calcu-lation that determines the level of traditionalSocial Security benefits paid to a worker and toall eligible family members.

6 In this example, the interest rate used for thehypothetical account (upon which the offset isbased) is lower than what the couple earned intheir actual individual accounts.

7 To qualify for Social Security divorced survivingspouse benefits, the marriage must have lasted atleast ten years.


Individual Account Taxation

189

The tax treatment of individual accounts canhave a dramatic impact on the costs, participa-tion levels, forms of payout, and benefits andburdens associated with them. Tax issues arefurther complicated because one cannot under-stand how to tax payouts from individualaccounts without understanding how contribu-tions to the accounts are taxed. This is truebecause of “tax equivalences.”

In brief, policymakers can tax or exempt incomeat three different points in the saving process:they can tax deposits, investment earnings,and/or withdrawals. An income tax generallytaxes savings deposits and investment earningsbut not withdrawals. A consumption tax, how-ever, can operate in one of two ways that, undercertain circumstances, are economically equiva-lent. A consumption tax may tax deposits andexempt investment earnings and withdrawals, orit may exempt deposits and investment earningsbut tax withdrawals. Finally, it is theoreticallypossible to exempt deposits, investment earn-ings, and withdrawals but doing so can producea negative tax rate without increasing savings atall.

Based on this general situation, four current-lawmodels for taxing individual accounts are exam-ined, along with ways in which they could bemodified.1 The “normal” model for taxing sav-ings mirrors the income tax regime wheremoney that is saved is taxed when initiallyearned, and the income generated by the savingsis then taxed when it is “realized.” The tradi-tional retirement savings model mirrors the con-sumption tax regimes where income earned onqualified retirement savings is exempt from taxso that only the contributions made by workersand their employers are subject to tax. This isaccomplished either by an upfront tax deductionfor contributions or a tax exemption for with-drawals. Certain other forms of retirement sav-ings are taxed under a third model—deferral—which taxes contributions immediately and taxesincome earned on contributions upon withdraw-al. Social Security contributions and benefits aretaxed under a fourth model where the employ-ee’s half of contributions are taxed, and betweenzero and 85 percent of benefits paid are taxed,depending on the beneficiary's income level.

Each of these models can be, and in some casesis, combined with tax credits, preferential rates,

10Chapter


and/or tax penalties, all of which can have a sub-stantial effect on tax burdens. As a result, thesefour models by no means comprise a complete listof options. Other models could easily be devel-oped depending on policymakers' objectives.

Tax equivalences and the options for taxingindividual accounts are explored, as are consid-erations policymakers might bear in mind whendeciding which model or models to apply inlight of an individual account plan’s purposesand design.

Tax Equivalences

Tax issues are unique because one cannot under-stand how to tax payouts from individualaccounts without understanding how contribu-tions to the accounts are taxed. This considera-tion of the contribution phase is necessarybecause of “tax equivalences,” which summarizethe relationship between regimes that taxincome at different points in time. The govern-ment can tax (T) or exempt (E) income at threepoints in the saving process: It can tax (1)deposits, (2) investment earnings, and (3) with-drawals, in that order. An income tax generallytaxes deposits and investment earnings but notwithdrawals (summarized TTE). By contrast, aconsumption tax may operate at the individuallevel in two ways. A consumption tax may, aswith Roth individual retirement accounts(IRAs), tax deposits and exempt investment

earnings and withdrawals (summarized TEE), orit may, as with deductible IRAs, exempt depositsand investment earnings but tax withdrawals(summarized EET). Finally, it is possible toexempt deposits, investment earnings and with-drawals (summarized EEE), but this method israrely pursued for reasons described below.

Figures 10-1 through 10-3 illustrate threeimportant insights of “tax equivalences.” First,as shown in Figure 10-1, the two consumptiontax regimes (TEE and EET) are generally eco-nomically equivalent if tax rates applicable atthe time of deposit, during the accumulationperiod, and at withdrawal are the same andunchanging.2 Essentially, the taxpayer has to paytax either when money is deposited or when it iswithdrawn from the account, but not on bothoccasions. If tax is paid earlier, when funds aredeposited, it is generally equal in present valueterms to the tax that otherwise would have beenpaid upon withdrawal. This holds true only ifthe taxpayer makes the same pre-tax contribu-tion and the rate of return that the taxpayer canearn is identical under each scenario.

A second insight of tax equivalences is thatapplying the income tax regime (TTE) imposeshigher taxes on people who defer consumptionby saving than it imposes on people who receivethe same income and consume immediately.Applying the consumption tax regimes (TEE andEET) is neutral regarding the timing of con-

Figure 10-1. Equivalence of Roth IRA and Deductible IRA Models

TEE (Roth IRA) EET (Deductible IRA)

Contribution Limit 300 400Pre-Tax Contribution 400 400Tax on Contribution 100 0After-Tax Contribution 300 400Investment Earnings 478 637Tax on Investment Earnings 0 0Pre-Tax Balance at Withdrawal 778 1037Tax on Withdrawal 0 259After-Tax Withdrawal 778 778

Assumes that tax rate is 25 percent; interest rate is 10 percent; inflation is zero; and deposits are invested for 10 years.

Chapter Ten: Individual Account Taxation 191

sumption.3 This occurs because the income taxregime taxes earnings accumulated on savings,but the other regimes do not. Figure 10-2 illus-trates how the present value of consumption isidentical under the consumption tax regimes,regardless of whether a taxpayer chooses to con-sume funds immediately or save them, but thatthis is not true of the income tax and totalexemption regimes. These results cruciallyassume, however, that interest payments are notdeductible and that taxpayers do not financesaving through deductible borrowing.

If savings are in fact financed through deductibleborrowing,4 the results are quite different.Figure 10-3 demonstrates that in such circum-stances applying the consumption tax regimes

(TEE and EET) to a savings vehicle can effec-tively result in negative tax rates, and applyingthe total exemption regime (EEE) produces evengreater subsidies without necessarily increasingsavings at all. Taxpayers who borrow to partici-pate in a savings vehicle can use interest deduc-tions and the tax benefits associated with thesavings vehicle to reduce taxes they otherwisewould have paid on other income. As a result,they can extract a subsidy from the governmentwithout increasing their net savings overall.Denying any deduction for interest on borrowedfunds, including home mortgage loans, can pre-vent this opportunity for taxpayers to “game”the system. The income tax regime also preventsthis tax arbitrage. But, under the total exemp-tion regime, taxpayers can, though borrowing,

Figure 10-2. Value of Consumption for Immediate Consumption Versus Saving

TTE, TEE & EET TTE TEE EET EEE Consume Save Save Save Save

Pre-Tax Funds Available 400 400 400 400 400Tax on Consumption / Deposits 100 100 100 0 0Tax on Investment Earnings 106 0 0 0After-Tax Investment Earnings 318 478 637 637Tax on Withdrawals 0 0 259 0Net Consumption 300 618 778 778 1037Value of Consumption in Year 10 778 618 778 778 1037

Assumes that tax rate is 25 percent; interest rate for borrowing and investing is 10 percent; inflation is zero; taxpayer consumes immedi-ately or invests funds for 10 years. Consumption is shown at time of withdrawal.

Figure 10-3. Negative Rates of Tax when Saving Financed Through Borrowing

All Models TTE TEE EET EEEConsume Save Save Save Save

Amount Borrowed 400 400 400 400 400After-Tax Amount Saved6 0 400 400 533 533After-Tax Investment Earnings7 637 899 1198 1198Tax on Withdrawal 0 0 433 0After-Tax Account Balance 1037 1299 1299 1731Principal & Interest Owed 1037 1037 1037 1037 1037Value of Subsidy in Year 10 0 0 262 262 694

Assumes that tax rate is 25 percent; interest rate for borrowing and investing is 10 percent; inflation is zero; taxpayer consumes immedi-ately or invests funds for 10 years.


extract a subsidy without saving even if interestis not deductible.5

Models for the Tax Treatmentof Individual Accounts

Based on this analysis, the tax code's four cur-rent models for taxing savings can be moreclearly understood. Some models mirror the gen-eral tax regimes described above, while othersmake important modifications.

Traditional Retirement Savings ModelThe traditional retirement savings model mirrorsthe consumption tax regimes by exemptingincome generated by savings through one of twomechanisms. In Roth IRAs (equivalent to TEE),taxpayers make contributions from their after-tax earnings or savings, but accumulated incomeand withdrawals are tax-exempt. Alternatively,traditional IRAs and employer-based retirementplans (equivalent to EET) allow taxpayers todeduct contributions or exclude them fromincome, and to defer tax on all income earned onthe accounts until the funds are withdrawn. Thefull amount of the withdrawal is taxed as ordi-nary income without regard to the character ofthe earnings. Under certain conditions, these twomechanisms for exempting income earned onretirement savings are economically equivalent.

Deferral ModelA second current-law model for taxing individ-ual savings accounts, which applies to a smallernumber of retirement savings vehicles, taxesboth contributions and account earnings butdoes not tax account earnings until withdrawal.The model applies to annuities and to savings inexcess of the contribution limits in certainemployer-sponsored retirement plans.8

Effectively, tax on the earnings accumulated onafter-tax contributions is deferred under thismodel, but not eliminated as under the tradi-tional retirement savings model. The value ofthis deferral can be substantial.

“Normal” Tax Treatment of SavingsThe third option for taxing individual accountsis to follow the “normal” model for taxing sav-ings under current law, which mirrors theincome tax regime (TTE). Generally, both con-tributions to non-retirement savings accountsand income earned on such contributions aresubject to tax. Taxpayers take no deduction forincome saved for an unspecified purpose, andmust report any income generated by such sav-ings to the Internal Revenue Service (IRS). Forexample, if an individual invests in a corporatebond, tax must be paid at ordinary rates oninterest as it accrues. In contrast, if the individ-ual invests in stock, tax must be paid at lowerrates on dividends received and on capital gainswhen the stock is sold.

This model differs from the traditional retire-ment savings and deferral approaches becauseincome generated by savings is taxed. In somecases, it is taxed only when it is realized.

Social SecurityThe Social Security system presents a fourthoption for taxing individual accounts. SocialSecurity contributions and benefits are taxed atboth higher and lower rates relative to the mod-els described above, depending on the beneficia-ry’s income level.

Social Security is funded half with after-tax dol-lars (the Social Security taxes paid by workers,which they cannot deduct or exclude from theirincome) and half by pre-tax dollars (the SocialSecurity taxes paid by employers, which arededucted by employers but are not included inworkers’ taxable income). The portion of distri-butions that must be included in income riseswith income: low-income beneficiaries need notinclude any Social Security benefits in their tax-able income, while higher-income beneficiariesmust include 50 to 85 percent of benefits. Theeffective income tax rate on both the employee’sshare of Social Security taxes and the portion ofbenefits that the employee must include inincome depends on the taxpayer’s income tax


rate, which will vary from person to person andin some cases will be zero.

The Social Security model stands in contrast tothe other three models described above becausethe percentage of account earnings included inincome varies by income level at the time ofwithdrawal. Assuming constant tax rates overtime, the Social Security model is more advanta-geous than the traditional retirement models forlower-income taxpayers and is generally lessadvantageous than the traditional retirementsavings models (and in some cases, the deferralmodel) for higher-income taxpayers. Some ana-lysts view the 85 percent method as approximat-ing the deferral method, with a generalassumption that contributions are equal to 15percent of benefits under the Social Security ben-efit formula.9 However, the 85 percent methodmay understate contributions for more recent,higher-income retirees and, even if not, it is dis-advantageous for higher-income taxpayers rela-tive to the traditional methods of taxingretirement income. It may also in some cases beless advantageous than the “normal” model fortaxing savings for higher-income taxpayers.

Tax CreditsBeyond these four general models, it is possibleto alter the tax treatment of individual accountsby funding them in whole or in part with taxcredits. Under current law, the only tax creditavailable for retirement savings matches low-income taxpayers’ contributions to accounts fol-lowing the traditional retirement savings model.However, each of the models could be combinedwith tax credits to lower effective tax rates onsavings, augment contributions for some or allaccount holders, or to provide incentives or dis-incentives for certain types of withdrawals.

Tax credits provide a way to offer tax incentivesor subsidies for savings in individual accountsthat do not necessarily rise with income becausethe value of a tax credit need not depend on thetaxpayer’s tax rate. By contrast, the deferral,Social Security, and traditional retirement sav-ings models all use deductions and exclusions as

incentives and subsidies, and these are inherentlyworth less to lower-income taxpayers who aresubject to lower tax rates or, in many casesunder current law, are not required to payincome tax at all.10

In addition, tax credits can vary along differenteligibility criteria, such as income, family struc-ture, and contribution size. Tax credits may alsobe “refundable,” meaning that taxpayers whoowe no federal income taxes still receive the fullamount from the government. Refundable taxcredits are a way to provide universal tax incen-tives to save in individual accounts, and are per-haps the best incentive for workers in thebottom half of the income distribution who cur-rently do not often participate in traditional tax-preferred retirement savings vehicles.11

Summary of Tax Policy Choices

Ultimately, the tax policy decisions that must bemade when developing any model for the taxa-tion of an individual account system turn oneight major factors.12 Each of the modelsdescribed above represent one set of choicesalong these dimensions but, as illustrated byAppendix B, which summarizes the tax treat-ment rules of selected individual account pro-posals, different decisions would generate stillmore models. Eight factors that would influencethe design of any tax model for the accounts follow.

Funding SourceIndividual accounts may be funded from a vari-ety of sources, including the existing or anexpanded Social Security tax, general tax rev-enues, voluntary or mandatory individual contri-butions, employer contributions, or some otherdedicated revenue stream.

Tax Treatment of Contributions Contributions may be made on a pre-tax basis,in which the account holder typically receives adeduction or is allowed to exclude contributionsmade on his behalf, or contributions mayinstead come from after-tax income. As illustrat-


ed by the tax equivalence discussion, policymak-ers also have the option to tax contributionswhen account funds are withdrawn.

Tax Treatment of Account Earnings Income tax on the earnings from individualaccount contributions may be imposed at differ-ent times and at different rates. In particular,income generated by individual accounts, suchas interest, dividends, and capital gains, may betax exempt or taxed at different stages: asaccrued, when realized, when withdrawn, or atsome other time. Preferential rates may apply toall account earnings or certain types of earnings.If earnings are taxed when withdrawn or distrib-uted, different methods may be used to deter-mine what portion of each withdrawal isattributable to contributions as opposed toaccount earnings.

Eligibility RequirementsEligibility to participate in any individualaccount program, or to receive certain tax bene-fits associated with the program, may berestricted through various means. For example,policymakers might establish contribution capsor participation might be restricted to workerswith income below a certain threshold.

Tax CreditsTax credits may be used to match contributionsor earnings on account balances, to persuadeemployers to contribute to accounts, or toencourage financial institutions to market andadminister the accounts.

Differential Tax Treatment ofWithdrawals Account holders may be subject to penalties orincentives for certain withdrawals. For example,some current retirement savings vehicles penalizewithdrawals prior to age 591/2 unless with-drawals are for approved purposes, such as edu-cation or a first-time home purchase. Penaltiesmay also be imposed if accounts are not annu-itized or if a portion of the account is not with-drawn each year during retirement (to preventtaxpayers from using individual accounts as

estate planning devices). Sufficiently high penal-ties may function as a disincentive for certainuses of individual accounts that policymakersdeem undesirable. Conversely, tax deductions,exclusions, credits, and preferential tax ratesmay be used to create incentives for certainwithdrawals.

Transfer Tax TreatmentEstate and gift taxes might also be used to fur-ther tax policy goals with respect to individualaccounts. Workers may have to pay gift tax, forexample, if they transfer their rights to anaccount. If withdrawals are taxed, a bequest orgift could also trigger recognition of income as ifaccount assets were withdrawn.

Implicit Taxes Finally, implicit taxes may be levied by publicand private sector means-tested programs (forexample, scholarship policies that consider per-sonal accounts in setting scholarship awards),and by involuntary distributions resulting from,for example, third-party creditor claims.

Potential Designs of IndividualAccounts

Lawmakers will likely be influenced by the basicdesign of an individual account system whenmaking decisions about tax policy. Of particularimportance for tax purposes is how the accountswould be funded and what rules would governwithdrawals. There are generally five ways offunding individual accounts, depending on boththe source of revenue and the rules of participa-tion (Figure 10-4). Currently scheduled SocialSecurity taxes might be used to fund either amandatory account system, where some portionof each worker’s current Social Security contri-butions would be shifted to an individualaccount, or a voluntary system where workerscould choose to shift a portion of their currentSocial Security contributions to individualaccounts.13 Alternatively, policymakers mightlook to new, earmarked contribution sourcesfrom workers, their employers, and/or the feder-


al government (for example, through increasedSocial Security taxes or general revenue spend-ing). Again, participation in and contributionsto a program funded with new revenues couldbe mandatory or voluntary. Finally, policymak-ers might fund the accounts with unspecifiedgeneral revenues.

In addition to the funding source, the design ofan individual account system depends on whetherand how account funds are withdrawn bothbefore retirement (if permitted) and at retirement.Two general payout methods are the most rele-vant for tax purposes. An account holder (andpotentially his or her spouse) might elect toreceive one or more lump sums and might trans-fer undistributed amounts as a gift or bequest.Life annuities, in contrast, would guarantee annu-itized payments for the remainder of the accountholder’s (and potentially a spouse's) life.14

These methods of funding and paying out fundsfrom individual accounts are not mutuallyexclusive. For example, all workers might berequired to annuitize base accounts fundedthrough mandatory contributions but also havean option to supplement their accounts throughvoluntary contributions not subject to an annu-itization requirement. This flexibility may be avirtue, but it may further complicate the ques-tion of how the accounts should be treated fortax purposes.

Implications of the Design andPurpose of Individual Accounts

Having laid out the potential designs of an indi-vidual account system, this final section exam-

ines considerations policymakers may wish totake into account in deciding which tax treat-ment model to apply, and in understanding whateach tax model practically entails. We examineeach potential design separately,15 but in manycases the considerations of earlier designs will berelevant to the designs examined subsequently.

Mandatory Accounts Funded withCurrent Social Security TaxesIf a portion of all workers’ current SocialSecurity taxes were automatically shifted to indi-vidual accounts, some aspects of account taxa-tion would be pre-determined: there would beno eligibility or contribution limits, and contri-butions would come from the Social Securitytax. However, the broad questions of how toaddress any distributional concerns and whetherto provide incentives or penalties for certainforms of withdrawals would remain.

In addition, the question would arise whethercontributions were deemed to come from theemployer or employee share of Social Securitytaxes. If contributions were split between theemployer’s share and the employees’ share, halfwould be after-tax and half would be pre-tax.Alternatively, if all the contributions came solelyfrom the worker’s share of Social Security tax,the entire amount would be after-tax. But in thiscircumstance, it might be necessary to adjust thetax treatment of traditional Social Security bene-fits because the current tax treatment assumesthat they are funded half with pre-tax and halfwith after-tax contributions. Accordingly, forsimplicity, the following discussion generallyassumes that individual accounts would be fund-


New Earmarked Currently Scheduled Unspecified GeneralNature of Contributions for Social Security RevenuesParticipation the Accounts Taxes for Accounts for Accounts

Mandatory Participation (1) (2)

(5)Voluntary Participation (3) (4)


ed half with the employer share and half withthe employee share of Social Security taxes.

Addressing Distributional Concerns throughAccount Taxation

In general, the distribution of benefits amongdifferent demographic groups under a mandato-ry individual account program funded with cur-rent Social Security taxes may be affected in fourways: through the benefit formula governing theremaining defined-benefit portion of the SocialSecurity system, through the method for allocat-ing shifted Social Security taxes to individualaccounts, through offsetting savings in othervehicles, or through the tax treatment ofaccount earnings and distributions.16

Under any combination of these mechanisms, itis difficult or impossible to design individualaccounts in such a way that, together with tradi-tional Social Security benefits, they provide thesame benefit coverage and exactly compensatethe Social Security trust funds for lost revenues,even disregarding transition costs.17 Nonethe-less, these mechanisms might be used toapproach (or modify) the distributional effectsof the current Social Security system. This chap-ter, however, focuses on designing the accounts’tax treatment and what administrative anddesign issues each tax treatment model wouldraise. The actual choice of which tax treatmentmodel to adopt will likely be driven not just bydistributional objectives, but also by revenuecosts and administrability concerns.18

Applying the Social Security Model

If the purchase of life annuities were mandatoryand the accounts followed the Social Securitymodel in their tax treatment, the treatmentwould depend on whether the contributionscame from employer or employee taxes. Forexample, if half of the contributions came fromeach, withdrawals should be taxed as if theywere Social Security benefits, with zero percentto 85 percent of the withdrawal included inincome, depending on the beneficiary’s incomelevel. In this case, the main differences betweennew individual accounts and Social Security

would be that individual accounts would likelyhave a different distributional effect than SocialSecurity, retirement income would be pre-funded(to the extent of individual accounts), workersmight have some degree of property rights totheir accounts, and workers might have authori-ty to determine how their accounts are invested.Workers would still likely have little or no con-trol, however, over when and how they receivetheir account funds.

Assuming that half of contributions are fromemployers and half from employees, one poten-tial advantage of following the Social Securitymodel is simplicity for beneficiaries: becauseSocial Security benefits are the only form ofretirement income for roughly 20 percent ofretirees, many people would need to understandonly one taxing regime. However, if fundingwere not half from employers’ and employees’shares of Social Security taxes or, as discussedbelow, if annuitization were elective, followingthe Social Security model would be far morecomplicated.

Another potential advantage of following theSocial Security model is that it eliminates onesource of distributional and revenue differencesbetween individual accounts and Social Securitybenefits. Stated differently, if an individualaccount program seeks to fully compensate theSocial Security trust fund for lost revenue andmaintain the distributional profile of the SocialSecurity system as a whole—and is somehowable to accomplish this result though offsets ona pre-tax basis—presumably policymakerswould want to follow the Social Security modelfor taxing the accounts to maintain this resulton a post-tax basis as well.19 It will, however, bedifficult, if not impossible, to create an individ-ual account program that is distributionally neu-tral and revenue neutral given issues likevariable investment returns, and the complexentitlements that the Social Security system cre-ates for children, former spouses, survivingspouses, and disabled adults.


Applying the Traditional Retirement SavingsModel

If mandatory annuitized accounts instead aim tofollow the traditional model for retirement sav-ings, the treatment would also depend onwhether the contributions came from employeror employee taxes. For example, if half of thecontributions came from each, 50 percent ofwithdrawals should be taxed as ordinaryincome, reflecting the 50 percent of accountcontributions that would have come from thepre-tax dollars of employer Social Securitytaxes.20 In effect, the income earned on theaccounts would then be exempt from tax. Oneadvantage of this model is that it is relativelysimple and intuitive if people understand thathalf of their contributions were not taxed. Itwould, however, alter the distributional profileof the current Social Security system because theportion of distributions subject to tax would notvary by income, as is the case with SocialSecurity benefits today.

Applying the “Normal” Model for TaxingSavings

If the accounts were to follow the “normal”model for taxing savings, workers would betaxed currently on earnings from the accountsand, if the model were followed strictly and con-tributions were funded half from the employers’share of Social Security taxes, workers wouldalso have to include the employers’ share oftheir Social Security tax contributions to theirindividual accounts in taxable income. Whilesuch tax treatment might address distributionalconcerns, its distributional impact would dependon the percentage of income that differentdemographic groups contribute, and the levelsof capital income tax rates applicable to differ-ent taxpayers relative to their ordinary incometax rates. This model would also be relativelycomplicated, especially for taxpayers who wouldnot otherwise report capital income. Accountholders would have to keep records of accountearnings and annually report and pay taxes onsuch income at varying rates depending on boththeir level of income and the form of the invest-ment income. A system would also have to be

put in place to apportion their after-tax contri-butions across annuity payments.

Applying the Deferral Model

Alternatively, the accounts might follow thedeferral model, which would tax all SocialSecurity tax contributions to individual accountscurrently, but would not tax account earningsuntil they are withdrawn. At that point, theentire amount of each annuity payment wouldbe included in income to the extent that itexceeded the allocated share of Social Securitytax contributions, which would have alreadybeen taxed. This option would avoid the needfor annual reporting of account earnings underthe “normal” model for taxing savings.However, this model could potentially be moreadministratively difficult and costly because tax-payers or account administrators would need tokeep records of contributions accumulated overtheir entire careers, and possibly at multiplefinancial institutions. Like the “normal” modelfor taxing savings, if contributions were fundedhalf from the employer’s share of Social Securitytaxes, workers would also have to include theemployer share of their Social Security tax con-tributions to their individual accounts in taxableincome.

Other Options

Other simpler options that do not fit within anyof the existing models are also possible. Forexample, workers might be required to include,or be permitted to exclude, 100 percent ofaccount withdrawals from income.21 Alterna-tively, all withdrawals could be taxed as SocialSecurity benefits. These one-size-fits-allapproaches would avoid both the liquidity prob-lems of the “normal” model and the complexityof the deferral approach. These alternativeapproaches would, however, alter the distribu-tional profile of Social Security. The first optioncould be criticized as involving double taxationif contributions came in part from the after-taxemployee share of Social Security taxes. As dis-cussed in the section on tax equivalences, thesecond and third options could be criticized asresulting in a subsidy without resulting in any


new saving if contributions came from the pre-tax employer share of Social Security taxes or ifindividuals financed their accounts through bor-rowing. Many tax experts believe that individualaccounts should be taxed at least once (either bytaxing contributions or withdrawals) except tothe extent that the accounts are intended to be amechanism for redistribution to low-incomeworkers.

Considerations If Accounts Are NotAnnuitizedIf workers are not required to annuitize theiraccounts, additional questions arise with respectto their tax treatment. Under elective annuitiza-tion, in contrast to Social Security, some propos-als would permit workers to withdraw theirfunds before retirement, receive lump-sum distri-butions (or periodic withdrawals) at retirement,and/or transfer their assets to third parties. As aresult, it would become necessary to determinehow to treat transfers of the accounts to thirdparties for tax purposes and whether to applyminimum payout rules. Many of the current tax-favored retirement savings vehicles require thataccount holders withdraw (and include inincome) a portion of their accounts each yearafter they reach a certain age or retire to ensurethat the accounts are used for retirement incomerather than estate planning purposes.

If annuities were elective, the tax system couldalso be used to create incentives to annuitizethrough tax deductions, exclusions, or credits,or through penalty taxes on non-annuitizedwithdrawals. Further, elective annuitizationwould also create design challenges in applyingthe Social Security, traditional retirement sav-ings, or deferral taxation models because work-ers might be subject to high tax brackets anddegrees of inclusion relative to their actualfinancial position. For example, if a workerreceived a large lump-sum withdrawal, theSocial Security approach of requiring taxpayersto include benefit payments in taxable incomebased on their income level might result in ahigh proportion of the lump sum being subjectto tax—and at a high rate—even though the

worker is generally low-income. To addressthese problems, it would be necessary to imputean annuity or provide for some other form ofincome averaging when determining the degreeof income inclusion and tax rate. Either of theseapproaches would introduce further complexity.

Finally, if annuitization were elective, the ques-tion would arise whether to permit, and how totreat, withdrawals prior to retirement. Thisquestion is addressed generally in Chapter Five.As that chapter suggests, the answer depends onthe goals of an individual account system andwhether and how policymakers seek to affectbehavior. For instance, if the accounts werefunded from the same revenue stream as SocialSecurity, workers might be prohibited from, orpenalized22 for, withdrawing funds prior toretirement, in order to promote retirement secu-rity. Alternatively, withdrawals might be restrict-ed to uses deemed socially productive.

Voluntary Accounts Funded with CurrentSocial Security TaxesMany of the implications for taxation describedabove would remain the same if accounts fund-ed with current Social Security taxes were volun-tary, rather than mandatory. The amount ofcontributions would likely be fixed or limited toa percentage of Social Security taxes. If annuiti-zation were mandatory, the accounts could betaxed according to any of the four existing mod-els, or the other models outlined above, withsimilar implications to those described above(for example, taxing 50 percent of withdrawalsas ordinary income in the case of the traditionalretirement savings model if 50 percent of thecontributions were pre-tax). And, elective annu-itization would raise the same questions aboutwhether minimum distribution rules shouldapply, whether and how to provide tax incen-tives to annuitize, how to tax transfers andlump-sum withdrawals, and whether workersshould be subject to penalties or prohibitions forcertain pre-retirement withdrawals.


Addressing Distributional Concerns

Two significant new taxation issues would ariseif the accounts were voluntary. First, achievingany distributional objectives would becomemore complicated because it is likely that thosewho might benefit more from the individualaccount system than from the traditional SocialSecurity system would disproportionately electto participate. If participants were dispropor-tionately higher-income, this would reduceSocial Security tax revenues available to achieveany distributional objectives through the remain-ing Social Security system. One way that policy-makers could try to avoid this distributionaleffect and revenue loss is by using offsets. If theobjective is to render the individual account pro-gram distributionally and revenue neutral on apre-tax basis (and this could somehow beaccomplished through offsets), policymakerswould probably want to apply the SocialSecurity tax treatment model to individualaccounts (including requiring contributions to behalf pre-tax and half post-tax) in order to main-tain this result on a post-tax basis. In addition,policymakers might adjust the distributionaleffects of individual accounts through mecha-nisms such as restricting eligibility for individualaccounts, disproportionately allocating shiftedSocial Security taxes to the accounts of lower-income workers, providing more favorable taxrules for the withdrawals of lower-income work-ers, or applying offsets to the individualaccounts that effectively return a portion of theindividual accounts to the Social Security trustfund.

Effect on Participation Rates

A second, related issue that arises with volun-tary accounts is whether to use taxes to createincentives (or disincentives) to shift SocialSecurity taxes into an individual account pro-gram, putting aside distributional objectives.Voluntary accounts necessarily introduce furthercomplexity into the retirement savings systembecause workers must choose whether to partici-pate by projecting which option is best for themin the future. Absent incentives or disincentives,it is unclear whether and to what extent workers

would elect to create individual accounts. Forexample, a risk-averse person might shy awayfrom the accounts given their uncertain returns.By contrast, some workers might be attracted toindividual accounts as a means of potentiallyshielding their future retirement income frompolitical whims and budget constraints. Further,people with different priorities may disagreeabout whether workers should be encouraged ordiscouraged to create individual accounts inplace of a portion of their Social Security bene-fits as a policy matter. The uncertain returnsfrom the accounts may increase the number ofretirees that the government must support dur-ing retirement; yet, the accounts might provideworkers with more freedom and autonomy.

Depending on the actual and preferred level ofparticipation in a voluntary individual accountprogram, some additional tax incentives to cre-ate the accounts may be desirable. Taxationmodels that exclude a larger percentage of con-tributions or withdrawals from income wouldtend to offer participation incentives, whilethose that include a larger portion of contribu-tions and withdrawals in income would tend todiscourage participation. In addition, policy-makers would need to pay particular attentionto incentives or disincentives created by theinteraction between the tax treatment of theaccounts and any benefit offsets. For instance, ifSocial Security benefits were offset by the valueof a hypothetical annuity purchased with thepre-tax account balance at retirement, and theaccounts were taxed more favorably than SocialSecurity benefits, this could create a participa-tion incentive.

Mandatory Accounts Funded with NewRevenues or Unspecified GeneralRevenues If individual accounts were mandatory andfunded with new revenues, many of the tax con-siderations would be similar to the two designsdiscussed above where contributions were madefrom Social Security taxes. The key differencewould be that the tax treatment would nolonger be constrained by the fact that funding


came from current Social Security taxes. As aresult, policymakers might, for example, com-bine tax credits with any of the models toaddress distributional concerns.23 When contri-butions come from a non-Social Security taxsource, several new issues arise.24

Applying the Social Security Model

First, applying the Social Security model tomandatory accounts funded from non-SocialSecurity revenues would be relatively complicat-ed. Perhaps 50 percent of contributions could betaxable25 and withdrawals, if annuitized, couldbe included in income as if they were SocialSecurity benefits. But, as described above, apply-ing the Social Security model would be compli-cated if annuitization were not mandatory.

Applying the Traditional Retirement SavingsModel: Roth IRAs v. Deductible IRAs

Second, if taxation of the accounts aimed to fol-low the traditional retirement savings model,policymakers would need to decide whether toemploy the deductible IRA or Roth IRAmethod. Under the deductible IRA method, thefull value of accounts funded with pre-tax con-tributions would be taxed as ordinary incomewhen withdrawn, while under the Roth IRAmethod accounts funded with after-tax contribu-tions would be tax exempt when withdrawn.The Roth IRA method would have the benefit ofproviding a more certain stream of retirementincome to workers because workers would notneed to plan for possible changes in the tax ratesthat would otherwise apply to their retirementincome. It is also generally more favorable toworkers who are currently low- and middle-income and do not benefit from current deduc-tions. Yet, by providing its tax benefits up front,the deductible IRA method might achieve higherlevels of participation for a given cost if, as dis-cussed below, the accounts were voluntary.

The political risk and revenue implications ofthe deductible IRA and Roth IRA methods dif-fer. Under the deductible IRA method, there maybe pressure not to tax withdrawals later on,resulting in a negative tax rate relative to an

income tax, while under the Roth IRA methodthere may be pressure to tax future distributionsto upper-income taxpayers. At the same time,the impact on future government revenues fromdeductible and Roth IRAs depends on future taxrates. For example, the Roth IRA model mightpotentially exacerbate the long-term fiscal gap(relative to the deductible IRA model) becausethe present value of the associated revenue losswould be greater if tax rates are expected to riseover time, and because much of the associatedrevenue loss would be outside the five or ten-year budget window and is less likely to betaken into account by legislators.

Administrative and Complexity ConcernsRaised by Other Models

Finally, if the accounts were mandatory andfunded with non-Social Security revenues, tech-nical and administrative complications wouldarise if taxation of the accounts aimed to followthe deferral model or the “normal” model fortaxing savings, as discussed above.Contributions would have to be after-tax, tax-payers or account administrators would need tokeep records of total contributions and accountearnings, and a system would have to be devel-oped for allocating basis (the portion of theaccount already taxed) over multiple with-drawals. In the case of the “normal” model fortaxing savings, taxpayers or account administra-tors would also need to keep records of theirbasis in individual assets held in their accounts,and the tax treatment would vary depending onthe type of assets in the accounts.

Voluntary Accounts Funded with NewRevenues or Unspecified GeneralRevenuesThe final individual account design—voluntaryaccounts funded with new revenues—raisesmany of the questions outlined above whoseanswers again depend on the accounts’ goalswith regard to distributional results, simplicity,and revenue. This design also raises several newissues.


As a preliminary matter, the taxation optionsdepend on the chosen contribution method. It isunlikely that new voluntary accounts would befunded through Social Security taxes if theaccounts were not a part of the Social Securitysystem because that arrangement would imposecomplicated withholding requirements on smallbusinesses, requiring them to track whether eachof their employees had elected to participate inthe program, and if so, at what level. Also, eventhough voluntary accounts would not mandateuniversal participation, the system could onlymake individuals universally eligible to partici-pate by separating contributions from theemployment-based Social Security tax system.As a result, contributions would likely comedirectly from individuals or employers electingto participate in the program.

Once the method for contributions is deter-mined, policymakers could choose among theexisting taxation models or could create a differ-ent taxation system through exemptions and/orcredits. As discussed above, the Social Securitymodel would be difficult to apply if contribu-tions were made outside the Social Security taxsystem. It would also make little sense to followthe “normal” model for taxing savings, standingalone, because taxpayers would then have noincentive to participate relative to other tax-pre-ferred forms of retirement saving. The remainingoptions are (1) the traditional retirement savingsmodel; (2) the deferral model; (3) exemptingcontributions and taxing withdrawals like SocialSecurity benefits; or (4) supplementing one ofthese options, or the “normal” model, with taxcredits.

Use of Tax Credits

The extensive use of tax credits to stimulate par-ticipation might raise new concerns, especiallyabout abuse. A voluntary individual accountprogram funded with new revenues might relymore extensively on tax credits than other indi-vidual account designs, both because tax creditscould be used to alter the distribution of taxbenefits in a system funded with new revenues,and because refundable tax credits would be

necessary if universal tax incentives to partici-pate were an important goal. If tax credits wereused to match contributions, and there were norestrictions in pre-retirement withdrawals, tax-payers might make contributions and immedi-ately withdraw them, in order to receive the taxcredits. Even if penalties were imposed on pre-retirement withdrawals, this “churning” couldbe advantageous if the match rate were highenough relative to the penalties. Accordingly, tothe extent that tax credits were a significant ele-ment of an individual account program that per-mitted pre-retirement withdrawals, anti-abuserules would likely be necessary.

Ultimately, the decision about which of themodels to apply would depend largely on thenormative objectives of the individual accountprogram. For instance, if the program sought toredistribute or to ensure a base level of retire-ment income or assets for all, refundable taxcredits, and perhaps eligibility or contributionlimits, would be needed. In contrast, if the pro-gram sought to eliminate disincentives for sav-ings in the tax system, irrespective of thedistribution of benefits, the traditional retire-ment savings model might be advisable. Finally,if the accounts aimed to increase national sav-ings overall, there is a robust debate aboutwhether incentives to save that are directed atthe less affluent (for example, refundable taxcredits) or the more affluent (such as, exemptingincome earned on the accounts from tax) wouldbe more effective, and it would be importantthat the new savings not be funded by increasedgovernment borrowing.

Summary

This chapter has explored various ways in whichindividual accounts could be treated for tax pur-poses. The tax treatment of individual accountscan have dramatic consequences for the cost ofany individual account program, the distribu-tional aspects of the system, and (if the accountswere voluntary) participation rates. Some of thekey tax policy decisions policymakers will need


to make with respect to any individual accountssystem follow.

How would the accounts be funded?Possibilities include existing or expanded SocialSecurity taxes, general tax revenues, individualcontributions, employer contributions, or someother dedicated revenue stream. The accounts'revenue cost, distributional profile, and size willdepend on their funding source and on whetherthe accounts are voluntary or mandatory.

Who would be eligible for any tax benefits asso-ciated with the accounts and at what contribu-tion level? For example, tax benefits could berestricted to certain low- and moderate-incomeworkers, or to contributions below a certainlevel. Both of these policies would affect the dis-tributional impact of the accounts and reducetheir cost.

Would tax credits be used to supplement theaccounts of certain workers, or to create incen-tives to establish accounts, contribute toaccounts, or to make certain types of with-drawals? For instance, tax credits might be usedto match individual or employer contributions,persuade financial institutions to administer theaccounts, or to encourage participants to annu-itize account balances. If the individual accountprogram were voluntary, refundable tax creditsare one way to provide universal incentives toparticipate.

How would the tax treatment of individualaccounts affect the taxation of traditional SocialSecurity benefits? If an individual account planis funded out of existing Social Security taxesbut is not funded equally from the employers’and employees’ shares, the creation of individualaccounts may raise the question whether adjust-ments are appropriate to the taxation of tradi-tional Social Security benefits.

How would account contributions, investmentearnings, and withdrawals be taxed? In general,current law provides four models.

Under the “normal” model for taxing savings,contributions to savings vehicles are taxed andthen income earned on such savings is taxed assoon as it is accrued or realized, often at differ-ent rates depending on the type of income. Forexample, interest income is taxed at ordinaryrates, while dividends and capital gains aretaxed at preferential rates. Withdrawals are nottaxed except to the extent that they result inincome earned on the savings being realized.

Under the traditional model for taxing retire-ment savings, either (1) contributions are taxed,but income earned on such savings and all with-drawals are tax exempt, or (2) contributions andincome generated by the contributions are nottaxed, but all withdrawals are taxed at ordinaryrates. Penalty rates typically apply to pre-retire-ment withdrawals. Under certain conditions,these two methods are economically equivalent.

Under the Social Security model, 50 percent ofcontributions are subject to income taxation,and a portion of each withdrawal is included intaxable income, ranging from zero percent forlow-income workers to 85 percent for higher-income workers.

Finally, under a deferral model, contributionsare taxed and income generated by such savingsis taxed, but such income (and only suchincome) is taxed when withdrawn.

Ultimately, the choice between these four mod-els, and the answers to the other questionsraised by this chapter, will depend on distribu-tional objectives, revenue concerns, administra-bility considerations, and how policymakerswould like to affect different individuals' incen-tives to participate.


Appendix A: Current Rules forTaxing Savings

The current law on the taxation of savings is amonument to mind-numbing and ever-changingcomplexity, and a detailed description couldconsume volumes. What follows is a briefoverview that illustrates the options that mightbe appropriate in the context of individualaccounts.

The Primary Alternatives Under CurrentLawThe “Normal” Model for Taxing Savings:Saving with After-Tax Dollars and CurrentTaxation of Income

Absent special provisions, a worker pays currentincome tax on earnings and has only after-taxdollars to put aside for savings. Interest on thosesavings (unless tax exempt) is taxed currently atordinary rates, but dividends are taxed at prefer-ential rates and, if the savings are invested in acapital asset, gain on that asset is also taxed atpreferential rates only when that asset is sold.The “normal” model imposes no restrictions onwithdrawals or distributions, and the owner ofthe assets has control over transfers of the assetsto third parties.

Social Security

The tax treatment of savings in Social Securitydiffers radically from the “normal” model forsavings. Contributions are mandatory and madethrough the Social Security tax, half of which ispaid by employers and half by employees.Employer contributions are excluded from theemployee’s gross income, while the employee’scontributions are not. Accordingly, workers gen-erally save through Social Security with half pre-tax and half post-tax dollars.

Workers may receive Social Security benefitsonly when certain events, such as retirement ordisability, trigger eligibility. The progressive ben-efit formula gives lower-income beneficiarieshigher replacement rates for their wages thanthose available to higher-income beneficiaries. A

portion of benefits received may also be taxed atordinary rates, but the specific percentage sub-ject to tax varies from zero percent for low-income beneficiaries to 85 percent for higher-income beneficiaries. This percentage bears nonecessary relation to the earnings that wouldhave accumulated on the worker’s contributions.

The Traditional Retirement Savings Model I:Saving with Pre-Tax Dollars and TaxingDistributions

As a partial supplement to Social Security,numerous provisions in the tax law allow tax-payers to save for retirement using pre-tax dol-lars. So, individuals may exclude such savingsfrom their income (such as contributions to401(k)s and other qualified plans) or deduct con-tributions to tax-preferred savings vehicles (forexample, traditional IRAs, SEP, and SIMPLEIRAs). Income on these savings (such as interestand dividends), and any gain on the sale of assetsheld in these accounts, is not taxed. Rather, allwithdrawals are taxed at ordinary rates as thosewithdrawals occur (including those representingcapital gains and dividends that would otherwisebe taxed at preferential rates). In effect, theseprovisions provide two tax benefits for retire-ment savings: they allow taxpayers to save withpre-tax dollars and to defer the taxation ofincome that might otherwise be taxed.

Eligibility and contribution limits for such tax-preferred savings generally follow two models:(1) employer-based programs (for example,401(k)s, qualified plans) and (2) individual-based programs (such as traditional IRAs). Theemployer-based model seeks to achieve broadcoverage of employees through nondiscrimina-tion requirements. Employer-based programsalso have relatively high caps on contributionamounts, and they permit a fair amount of flexi-bility in deciding whether to make contributionsmandatory and in determining the relative per-centages of contributions from employers andemployees. Individual-based programs, in con-trast, do not require employer sponsorship.These programs seek to alter the distribution ofthe tax benefits through eligibility restrictions


and contribution limits. Individuals voluntarilymake contributions.

These programs usually prohibit withdrawalsbefore retirement, as in defined-benefit plans, orrestrict such withdrawals to certain uses such asfirst-time home buying, education, or pursuantto Qualified Domestic Relations Orders in casesof divorce. If pre-retirement withdrawals areprohibited, the plans often impose restrictionson transfers to third parties, such as through ajoint-and-survivor annuity requirement. If pre-retirement withdrawals are not prohibited, non-qualified withdrawals generally are subject to a10 percent tax penalty on top of the normallyapplicable tax rate. This penalty is intendedboth to recover the benefit of tax deferral thatthe account holder has received, and to deternon-qualified withdrawals.

The Traditional Retirement Savings Model II:Saving with After-Tax Dollars and ExemptingDistributions from Tax

As an alternative to saving with pre-tax dollars,a number of provisions in the tax law follow thetraditional model for taxing retirement savingsby permitting taxpayers to fund tax-favoredaccounts with after-tax dollars, and exemptingall income earned on the accounts and with-drawals from tax (for instance, Roth IRAs,Coverdell Education Savings Accounts, Section529 Plans, and life insurance26). All else beingequal, the tax implications are the same underthis model as under the previous model wherecontributions were pre-tax.27 Of course, all elseis seldom equal because the worker is rarelytaxed at the same rate when he withdraws fundsas he was when he made the contributions. Aworker currently taxed at low rates, who antici-pates facing higher rates in retirement, wouldprefer to save with after-tax dollars and receivetax-free withdrawals, while a worker currentlytaxed at high rates, who anticipates being taxedat lower rates in retirement, would prefer tosave with pre-tax dollars and pay tax on with-drawals. And, setting rates aside, workers mayprefer the cash flow certainty that comes withnon-taxed withdrawals under this model.

Like individual-based accounts funded with pre-tax contributions, some of these provisions (forexample, Roth IRAs and Coverdell EducationSavings Accounts) restrict eligibility and havecontribution limits, presumably to reduce thecost and ensure that the tax benefits do not dis-proportionately flow to more affluent savers.However, Section 529 plans, which are regulatedat the state-level, have relatively high contribu-tion limits and are open to all.

Unlike some accounts funded with pre-tax con-tributions (such as defined-benefit pensions),plans that exempt account earnings typicallyimpose no prohibitions on non-qualified with-drawals. And when non-qualified withdrawalsare penalized, such withdrawals are taxed onlyto the extent that the account holder withdrawsmore funds from the account for a non-qualifieduse than were contributed, in part because con-tributions are made after taxes. In such cases,the excess generally is taxed as ordinary incomeplus a 10 percent penalty.

The Deferral Model

In addition to the traditional retirement savingsoptions described above, a smaller number ofvehicles tax both contributions and incomeearned on contributions, but in the latter caseonly when the income is withdrawn. This treat-ment applies, for example, to annuities and tosavings that exceed the contribution threshold in401(k) plans that elect to receive after-tax con-tributions, which generally are treated under theannuity rules if annuitized.

Under the annuity rules, if an annuity pays fixedsums for a period of years, basis (the worker’stotal after-tax contribution) is recovered prorata. Stated differently, if the worker receives tenequal payments, upon each distribution he willpay tax on one-tenth of the amount by whichthe total payments due exceed his total contribu-tions. If an annuity does not have a fixed dura-tion (a life annuity), the annuitant can excludefrom income a portion of each payment equal tothe basis divided by the number of payments heis expected to receive based on mortality tables.


In effect, tax on both of these forms of annuitiesis deferred because the annuitant pays tax whenincome is received as an annuity rather thanwhen it accrues or is realized.

Like traditional IRAs and 401(k)s, distributionsfrom annuities and savings vehicles followingthe deferral model are taxed as ordinary income(including any portion that otherwise would betaxed at preferential rates) to the extent theyexceed contributions. But, in contrast to tradi-tional IRAs and 401(k)s, contributions to annu-ities and savings vehicles following the deferralmodel are not deductible or excludable fromincome. Indeed, annuity taxation may be lessfavorable than taxation under the “normal”model for taxing savings if the spread betweenthe ordinary and capital gains rates is largeenough and the annuity contributions accumu-late income over relatively few years or if theincome is mostly unrealized capital gains

While there are no contribution or eligibilitylimits on purchasing annuities, non-qualifiedwithdrawals are subject to a 10 percent taxpenalty and treated as income to the extent thatthe amount received exceeds basis. In addition,the definition of a qualified withdrawal is rela-tively limited, excluding, for example, educationand first-time home buying expenses. The 10percent tax penalty also applies to other savingsvehicles following the deferral model, but thesesavings vehicles must be sponsored by anemployer and impose fewer restrictions on withdrawals.

Credit-Based Saving Subsidies

Finally, while far less common, several currentand proposed measures provide saving incen-tives or subsidies in the form of tax credits,

which may be refundable and/or automatic.28

Examples of these measures include the Section25B Saver’s Credit, Retirement Savings Accounts(RSAs), and Universal Savings Accounts (USAs).These arrangements may be viewed as enhancedversions of savings with pre-tax dollars, wherethe dollar value of the credit for low-incometaxpayers is comparable to, or greater than, thevalue of a tax deduction for higher-income tax-payers (for example, traditional IRA holders eli-gible for the Saver’s Credit). Alternately, taxcredits can have the potential effect of exempt-ing from tax both earned income and incomefrom capital (for example, Roth IRA holders eli-gible for the Saver’s Credit).

The Saver’s Credit piggybacks on existing savingsvehicles that exempt from tax income earned onaccounts. Low-income savers receive a non-refundable tax credit (funded from foregone gen-eral revenues) for up to 50 percent of theirsavings in these vehicles. The credit reduces aworker’s tax liability, and the worker has com-plete discretion over whether to spend the creditor contribute it to the savings vehicle. In con-trast, some proposals that incorporate tax credits(such as USAs) would automatically depositrefundable tax credits into individual accountsfor low-income workers regardless of whetherthey have saved, and/or automatically deposittax credits in individual accounts that match allor a portion of workers’ voluntary contributions.

These credits might be thought of as grants,which may be taxable or nontaxable dependingon the overall tax treatment desired.


Appendix B: Tax Treatment in Account Proposals

Figure 10-B. Tax Treatment Rules in Selected Individual Account Proposals

Proposal Tax Treatment Provisions


ACSS (Gramlich): Individual • Accounts would be funded with new worker Social Security tax Account Plan, 1996 contributions; tax treatment is not specified.

• Accounts would be annuitized but tax treatment of withdrawals is unclear.

Committee on Economic Development, • Accounts would be funded with increased Social Security tax 1997 contributions from both workers and employers. Both shares would be

excluded from employee income. • Contribution limits for other tax-preferred savings vehicles would not be

changed.• Withdrawals would be taxed as ordinary income. • Account balances would be included in a worker's estate if the

individual died before retirement.Mandatory Accounts Funded with Scheduled Social Security Taxes

ACSS (Schieber & Weaver): Personal • Accounts would be funded with current Social Security taxes. Security Accounts, 1996 • Other details of tax treatment unclear.

National Commission on Retirement Policy, • Mandatory accounts would be funded with current Social Security1999 taxes. Tax treatment is not specified.

• Tax treatment of withdrawals unclear.• Additional voluntary contributions of up to $2,000 (net of IRA

contributions) could be made to accounts. • Voluntary contributions would be after-tax and earnings accumulated

on voluntary contributions would be taxed when withdrawn.

Reps. Kolbe & Stenholm's 21st Century • Mandatory accounts would be funded with current Social SecurityRetirement Security Act (H.R. 2771, taxes. Tax treatment is not specified. 107th Congress) • Additional after-tax voluntary contributions of up to $5,000 could be

made to accounts.• Refundable tax credit would match low-income workers' voluntary

contributions ($150 for the first dollar, and 50 percent thereafter up to a maximum match of $600). An additional credit would match voluntary contributions from Earned Income Tax Credit refunds.

• Withdrawals of mandatory contributions would be taxed like Social Security benefits. Withdrawals of voluntary contributions and matching tax credits (and earnings thereon) would be excluded from income.


Clinton's Retirement Savings Accounts, • Workers earning under $80,000 could contribute up to $2,000 to2000 accounts on an after-tax basis.

• Refundable tax credit would match 200 percent of the first $200 that a low-income couple contributed, and 100 percent of the next $1,800 in contributions. Match rate would phase out at higher incomes. Tax credit must be deposited to an account if not used to offset tax liability.

• Account earnings would be taxed, but only upon distribution.


R.M. Ball, Social Security Plus, 2003 • Workers could elect to have employer deduct 2 percent of wages (on top of current Social Security taxes) and pay into an account.

• Workers could receive deduction for contributions.• Withdrawals included in ordinary income.• At death, account balance would be included in estate.

Savings for Working Families Act of 2003 • Up to $1,500 per year could be deposited in IDAs offered by qualified(S. 476, 108th Congress) financial institutions to low-income taxpayers.

• Financial institutions would match up to $500 of participant contributions annually on a one-to-one basis, and would receive a tax credit to offset such matches. Financial institutions would also receive an annual $50 credit per account to maintain the account and provide financial education. The tax credits could be transferred.

• Participant's contributions and account earnings on such contributions would be taxed.

• Matching contributions and earnings on such contributions would not be taxed when deposited or, in the case of a qualified withdrawal, when withdrawn.


PCSSS Model 2, 2001 • Worker could elect to redirect 4 percent of wages (up to $1,000) from Social Security taxes to an account. Tax treatment is not specified.

• Tax treatment of withdrawals unclear.

PCSSS Model 3, 2001 • Worker could elect to redirect 2.5 percent of wages (up to $1,000) from Social Security taxes to an account if they voluntarily contributed 1 percent of wages in an account. Tax treatment is not specified.

• Low-income taxpayers could receive a refundable tax credit to offset a portion of voluntary contributions.

• Tax treatment of withdrawals unclear.

Sens. Moynihan & Kerrey's Social Security • Workers could elect to have employee payroll tax reduced by 1 Solvency Act of 1999 (S. 21, 106th percentage point, or to have 2 percent of wages (1 percent employee Congress) share, 1 percent employer share) deposited in an account. Tax

treatment is not specified. • Tax treatment of withdrawals unclear.


Rep. Shaw's Social Security Guarantee • Workers would receive a refundable tax credit of 4 percent of wagesPlus Act of 2003 (H.R. 75, 108th each year (up to $1,000, indexed to average wages).Congress) • Account earnings would accumulate tax-free.

• Distributions would be taxed to the same extent as the worker's Social Security benefits. If the worker died before benefit entitlement, account distributions would not be taxable as benefits and would not be subject to the estate tax.

Figure 10-B. Tax Treatment Rules in Selected Individual Account Proposals (continued)

Proposal Tax Treatment Provisions


Chapter Ten Endnotes

1 Details on these models and others can be foundin Appendix A.

2 Because personal circumstances change, individu-als change tax brackets throughout their lives,and future tax rates are unpredictable, thisassumption is unlikely to hold true. The directionof the difference, however, is uncertain.

3 The total exemption regime is not applied toconsumption in this analysis because if it wereapplied generally to both saving and consump-tion the government would raise no revenue.

4 A taxpayer could finance saving throughdeductible borrowing by, for example, taking outa home equity loan on the appreciated value ofthe house, deducting interest paid on the loan,and using the borrowed proceeds to invest in thesavings vehicle.

5 If interest is not deductible, the after-tax accountbalance under the total exemption regime wouldbe $1,382 in Figure 10-3, and the present valueof consumption would be $345, also implying anegative tax rate. Under the consumption taxregimes, the after-tax account balance would be$1,037 and the present value of consumptionwould be zero, implying zero taxes and zero sub-sidies.

6 These figures assume that the taxpayer increasescontributions to compensate for the value ofdeductions and exclusions claimed. Stated differ-ently, it assumes that the taxpayer’s after-taxincome available for consumption is unchangedrelative to the taxpayer’s position prior to invest-ing in the savings vehicle.

7 These figures assume that taxpayers annually addto their accounts the value of interest deductionsclaimed on the borrowed funds.

8 Annuities and after-tax employer-sponsoredretirement plans tax contributions at ordinaryrates when deposited and earnings on contribu-tions at ordinary rates when withdrawn. Thismeans that when the income is mostly in theform of unrealized capital gains, the deferralmethod may be disadvantageous in comparison

to the “normal” method of taxing income.Theoretically, it is possible to tax contributionsat ordinary rates, but to tax earnings on contri-butions at ordinary or preferential rates depend-ing on the character of the earnings.

9 See, for example, Yvonne Hinson & DanielMurphy, “Is the 85-Percent Social SecurityInclusion Ratio High Enough?” 59 Tax Notes571 (Apr. 26, 1993).

10 Approximately 35 percent of all tax units in agiven year are not required to pay federal incometax after expenses, exclusions, exemptions,deductions and credits. See Tax Policy Center,Federal Tax Liability By AGI (citing JointCommittee on Taxation 1999 data), available athttp://www.taxpolicycenter.org/TaxFacts/overview/liability.cfm. Over time, however, farmore tax units face a positive federal income taxliability.

11 Employer matches are economically similar torefundable credits but are far from universal.

12 While state tax considerations are certainly rele-vant to how individual accounts should be taxed,such issues are beyond the scope of this chapter.As a general matter, some states piggyback onfederal rules, some do not have an income tax,and some vary from the federal regime in impor-tant respects.

13 As discussed in Chapter One, some individualaccount proposals fund the accounts with currentSocial Security taxes but use general revenues orother revenues to compensate the Social SecurityTrust Fund for the taxes that have been shifted.

14 As discussed in Chapter Three, annuities canprovide payments in a variety of ways, for exam-ple for a set number of years or for life with aminimum total payment.

15 We do not address category 5 separately fromcategories 2 and 4.

16 As discussed in Chapters Six and Seven, theintra-family distributional concerns that currentlyinform Social Security might be addressed inindividual accounts by requiring joint-and-sur-vivor annuities, or splitting payroll contributions


into an account for each spouse if a worker filesjointly.

17 Chapter Nine discusses some issues in achievingthis goal in the context of offsets.

18 In this regard, it is worth noting that technicallyindividual accounts would only be funded out ofcurrent Social Security taxes if their tax treat-ment followed the Social Security model becauseall of the other models involve more or lessamounts of foregone revenue than the currentSocial Security system.

19 Theoretically, offsets could be calculated on apost-tax basis, however this approach wouldrequire detailed information about each benefi-ciary's specific tax circumstances.

20 Some individual account proposals have deemedcontributions to come only from employee SocialSecurity taxes. In this case, the traditional retire-ment savings model would imply excluding with-drawals from income. However, as discussedabove, it might then be necessary to adjust thetax treatment of traditional Social Security bene-fits to reflect that as much as 100 percent of thecontributions might be pre-tax. If policymakersalso wanted to apply the traditional retirementsavings model to traditional Social Security bene-fits and the portion of Social Security benefitsshifted was large enough, this might entailincluding all traditional Social Security benefitsin taxable income. In addition, the possibility offunding individual accounts solely with eitheremployee or employer Social Security taxes raisesother issues discussed in the section below enti-tled “Applying the Traditional RetirementSavings Model: Roth IRAs v. Deductible IRAs.”

21 A version of the 100 percent exclusion option isthe tax treatment afforded to Health SavingsAccounts (“HSAs”), a special provision initiatedby Congress to respond to particular issues ofcost control in the health care context. In certaincircumstances, 100 percent of both contributionsto, and withdrawals from, HSAs can be excludedfrom income. This would imply funding theaccounts solely with employer Social Securitytaxes, which are pre-tax. We do not take thetreatment afforded to HSAs as a precedent for

retirement savings or income and, in this sense, itwould be a new model.

22 The traditional retirement savings model imposesa 10 percent additional tax penalty on most earlywithdrawals.

23 The distributional impact of the accounts couldalso be adjusted by taxing different proportionsof withdrawals depending on the taxpayer’sincome (similar to the Social Security model);funding the accounts through a revenue sourcewith a different distributional burden; changingthe formulas used to determine benefits or con-tributions; or altering savings in other vehicles. Inparticular, policymakers disagree about whethermandatory individual accounts funded with newrevenues would result in reduced or increasedsavings (including, but not limited to, in employ-er-sponsored retirement plans) and, if so, amongwhich workers.

24 Theoretically, if the accounts were universal,applying any of the tax treatment models thatinvolve funding the accounts on a pre-tax basismight present liquidity problems. For example, ifthe government deposited a fixed annual amountin each person’s account, an individual with noincome in a given year might be subject to taxfor contributions made on his or her behalf fromgeneral revenues. In practice, this is unlikely tobe a problem given the standard deduction andpersonal and dependent exemptions.

25 It is debatable what “after-tax” means in thecontext of accounts funded with income tax rev-enues. Most likely all contributions would thenbe treated as after-tax.

26 Life insurance only follows this model if benefitsare paid upon the death of the policyholder. Ifthe policyholder instead cashes out the policy, heis taxed on his gain at that time, in line with thedeferral model. It is important to bear in mindthat the pure insurance element of life insurancecontracts (the amount the policyholder wouldpay for pure term life insurance) is sometimesquite small.

27 To see how the two arrangements are equivalent,it is useful to put the calculation in algebraicterms. If T is the tax rate, D is the amount

210

deposited, i is the rate of return, and y is the

years over which the deposit accumulates, the

taxation of contributions but not of earnings on

the account can be represented as (T*D)*((i)y).

The taxation of account earnings, but not of con-

tributions, can be represented as T*(D((i)y)). And

(T*D)*((i)y) = T*(D((i)y)). In general, each of

these arrangements can be viewed as exemptingfrom tax the income from capital.

28 As noted above, this overview of the currentoptions for taxing savings is by no meansexhaustive. Many other types of savings receiveother forms of favorable tax treatment (such ashome ownership).


211

Glossary

Adverse selection: A term used in the insurancefield to describe the situation whereby thosemore likely to use or benefit from the insuranceare more likely to purchase it. For example,adverse selection occurs when individuals withabove average life expectancy are more likely tobuy life annuities. The buyers’ self-selection isadverse to the insurer.

Aid to Families with Dependent Children(AFDC): A state-based federal assistance pro-gram created as part of the Social Security Actof 1935 that provided cash assistance to familiesin need of monetary aid who met specificincome requirements. The Temporary Assistanceto Needy Families (TANF) program was enactedin 1996 to replace AFDC.

Annuitization (life): The process of convertingfunds in a person’s retirement account intomonthly (or other periodic) income that is paidfor the rest of the person’s life; the purchase of alife annuity.

Annuity (life): A guaranteed periodic (usuallymonthly) income that is paid for the life of theannuitant. Annuity contracts are sold by lifeinsurance companies. The insurance companyreceives premiums, either in a lump sum or aseries of payments, from an annuity buyer and,in return, has a contractual obligation to pay aguaranteed income to the annuitant for the restof his or her life.

Consumer Price Index (CPI): An index formu-lated and published by the Bureau of LaborStatistics of the U.S. Department of Labor thatmeasures average changes in the prices of goodsand services.

Consumer Price Index for Urban Wage Earnersand Clerical Workers (CPI-W): An index calcu-lated and published by the Bureau of LaborStatistics of the Department of Labor that isused to annually adjust Social Security benefitsto keep pace with inflation.

Contingent joint-life annuity: An annuity thatpays a lower amount to a widowed secondaryannuitant than to a widowed primary annuitant.The primary annuitant’s payment is not reducedif he or she is widowed. If the secondary annui-tant is widowed, the payment could be 75 per-cent, 67 percent, 50 percent, or any otherfraction of the amount previously paid to theprimary annuitant.

Contributions: Payments into a retirement plan.Social Security taxes on wages are also some-times called contributions.

Corporate bond: An IOU issued by a corpora-tion. By selling the bond, the corporation bor-rows money from the investors who purchasethe bonds. Corporate bonds are also referred toas corporate debt instruments. Most bonds payinterest at regular intervals until they mature, atwhich point investors get their principal back.Alternatively, some bonds are sold at a discountto their face value – for example, $800 for a$1,000 bond – and do not pay interest at regu-lar intervals. In this case, the investor gets$1,000 when the bond matures, receiving boththe interest and principal repayment in a lumpsum.

Cost-of-living adjustment (COLA): Periodicincreases in Social Security benefits (in effectsince 1975) to keep pace with inflation as meas-ured by the Consumer Price Index for UrbanWage Earners and Clerical Workers (CPI-W).

Civil Service Retirement System (CSRS):Originated in 1920, the system provides retire-ment, disability and survivor benefits for civilianemployees of the federal government. It is adefined-benefit retirement program funded byemployee and government contributions.Employees covered under the CSRS are not cov-ered by Social Security. The CSRS continues tocover federal employees who were hired before1984 and did not elect to shift to the newFederal Employees Retirement System (FERS).CSRS-covered employees can make contribu-tions to the Thrift Savings Plan, but they do not


receive matching contributions from the government.

Disabled adult child benefits: Social Securitybenefits paid to adults who have been disabledsince childhood (before age 22) and are eligibleas the children of insured workers who havedied, retired, or become disabled.

Deferred annuity: A tax-favored investmentproduct that, unlike a life annuity, does notguarantee payments for life. The account holderhas the option to later use the proceeds to buy alife annuity. The product is used mainly as amechanism for tax deferral during fund accumu-lation.

Defined-benefit plan: A retirement plan thatpromises to pay the participant a specificmonthly benefit for life. The monthly benefitsare often calculated through a formula that con-siders the participant’s salary and work history.The plan sponsor is responsible for having suffi-cient funds to pay the promised benefits. TheSocial Security program is an example of adefined-benefit plan.

Defined-contribution plan: A retirement planthat provides an individual account to each par-ticipant. The funds available at retirementdepend on contributions to the account, invest-ment gains or losses, administrative expenses,and whether funds were withdrawn beforeretirement.

Employee Retirement Income Security Act(ERISA): Enacted in 1974, ERISA was designedto secure the benefits for participants in privatepension plans by setting federal rules regardingparticipation, vesting, funding, reporting, anddisclosure and establishing the Pension BenefitGuaranty Corporation.

Electronic Transfer Accounts (ETA): A low-costaccount designed by the Treasury Department toensure that individuals who are required toreceive federal payments electronically haveaccess to an account at a reasonable cost andwith the same consumer protections available to

other account holders at the same financial institution.

Equity: An ownership share in a corporation,also called a stock.

Federal Employees Retirement System (FERS):The retirement system for federal employeeswho were hired after 1983 and are automatical-ly covered by Social Security. FERS also coversemployee hired before 1984 who elected to becovered by Social Security and FERS. FERS par-ticipants are eligible for defined benefits thatsupplement Social Security and receive govern-ment matching contributions to the ThriftSavings Plan.

Fixed life annuity: An annuity that pays a flatdollar amount (usually monthly) for the life ofthe annuitant.

Fixed term annuity: A contract that promisesspecified payments for a given term (for exam-ple, five or ten years). The annuity providerbears no mortality risk.

Federal government debt: The federal debt isthe total of all the Treasury bonds, bills andnotes representing obligations of the federal gov-ernment to repay the holders of these instru-ments. These bonds, bills and notes are held bythe public and by various federal trust funds,including the Social Security trust funds. Thedebt is the net accumulation of past annual fed-eral budget deficits and surpluses.

Federal Retirement Thrift Investment Board(FRTIB): An independent government agencycreated by the Federal Employees’ RetirementSystem Act of 1986 that oversees and adminis-ters the Thrift Savings Plan (TSP).

Gross domestic product (GDP): The total dollarvalue of all final goods and services produced ina year by labor and property located in theUnited States, regardless of who supplies thelabor.

Glossary 213

Health Savings Account (HSA): Accounts creat-ed by the Medicare Prescription Drug,Improvement and Modernization Act of 2003that permit individuals to save on a tax-freebasis for future qualified medical and retireehealth expenses.

Illustrative average (low, high, or maximum)earner: See steady earner and scaled earner.

Individual development account (IDA): A typeof savings account that is subsidized withmatching funds for low- and moderate-incomeindividuals. The funds can be used for specificasset- and wealth-building purposes, such as thepurchase of a home, post-secondary education,or business start-up costs. The accounts areoperated by community-based nonprofit organi-zations with funds from private foundations andfederal funds under the 1998 Assets forIndependence Act.

Inflation-indexed life annuity: A life annuitythat is adjusted each year to keep pace withinflation as measured by the consumer priceindex.

Internal Revenue Code (IRC): Title 26 of theUnited States Code that contains the tax laws ofthe United States.

Internal Revenue Service (IRS): The agency ofthe U.S. Treasury Department that is responsiblefor implementing the tax rules of the UnitedStates.

Joint-and-survivor or joint-life annuity: Anannuity that pays a regular monthly income forlives of two people: the primary annuitant and asecondary annuitant, or annuity partner (oftenthe primary annuitant’s spouse). A joint-lifeannuity pays until both annuitants have died.

Long-term disability insurance (LTDI):Insurance that provides a partial replacement ofa worker’s earnings that are lost due to a seriousillness or injury that renders the worker unableto work. LTDI policies vary in the duration andamount of payout.

Means test: A feature of eligibility criteria forassistance programs. Assistance programs maystipulate that only individuals with incomebelow certain thresholds (an income test) andassets below certain amounts (an asset test) areeligible for the assistance.

National Organization of Life and HealthInsurance Guarantee Associations (NOLHGA):An organization created in 1983 to help statescoordinate guaranty funds to cover insurancecompany insolvencies that involve three or morestates.

Offset: A term of art as used in this report, anoffset is a feature of proposals that shift sched-uled Social Security taxes to personally heldindividual accounts. The offset is a reduction inthe account holder’s future Social Security bene-fit or individual account that is intended to com-pensate the Social Security trust funds in full orin part for the value of Social Security taxesshifted to individual accounts.

Participating variable life annuity: A variablelife annuity in which the risk of changes in lifeexpectancy is shared between annuitants and theannuity provider.

Pension Benefit Guaranty Corporation (PBGC):A federal corporation created by the EmployeeRetirement Income Security Act of 1974 to pro-vide pension benefit insurance for participants in private defined-benefit pension plans. PBGCmaintains separate insurance programs for single-employer defined-benefit pension plansand for multi-employer defined-benefit pensionplans. Plan sponsors must pay premiums toPBGC based on the number of participants inthe plan.

Primary Insurance Amount (PIA): The SocialSecurity benefit an individual receives if he orshe elects to begin receiving benefits at normalretirement age. The PIA is calculated by a for-mula in law that is based on the worker’s aver-age indexed monthly earnings in Social Securitycovered employment. All Social Security benefits


for family members are based on a worker’sPIA.

Qualified Domestic Relations Order (QDRO):A judgment, decree or order that establishes oracknowledges the existence of an alternatepayee's right to receive, or assigns to an alter-nate payee the right to receive, all or a portionof the benefits payable with respect to a partici-pant under an ERISA-qualified employee benefitplan.

Real annuity, inflation-indexed life annuity: Alife annuity that is automatically adjusted tokeep pace with inflation as measured by theconsumer price index.

Real wage growth: Wage growth in excess ofinflation.

Refund of premium annuity: An annuity thatguarantees payments equal to the nominal pur-chase price of the annuity.

Rising life annuity: An annuity that paysamounts that increased at a prescribed rate (forexample, 3 percent a year) for the life of theannuitant.

Scaled earners: Illustrative workers with lifetimeearnings patterns that vary by age to reflect typi-cal age-earnings profiles. Created by the Officeof the Chief Actuary of the Social SecurityAdministration, the scaled earners are illustratedat three earnings levels: (1) a scaled mediumearner (whose career average indexed earningsare equal to the Social Security average wageindex); (2) a scaled low earner (whose lifetimecareer average indexed earnings are equal to 45percent of the Social Security average wageindex); (3) a scaled high earner (whose careeraverage indexed earnings are equal to 160 per-cent of the Social Security average wage index).

Single-life annuity: An annuity that pays regularmonthly income for the life of one person.

Steady earners: Illustrative workers with life-time earnings that are a constant fraction of the

average wage of all U.S. workers. Created by theOffice of the Chief Actuary of the SocialSecurity Administration, the steady earners areillustrated at four earnings levels: (1) a steadyaverage earner (who always earned the averagewage of all workers); (2) a steady low earner(who always earned 45 percent of the averagewage); (3) a steady high earner (who alwaysearned 160 percent of the average wage); and(4) a steady maximum earner (who alwaysearned the maximum amount that is taxed andcounted for Social Security purposes).

Stock: An ownership share in a corporation;also called an equity. Some stocks pay periodicdividends (a share of the company’s profits) totheir owners. A stock can be sold at a pricehigher or lower than was originally paid.

Supplemental Security Income (SSI): A federalassistance program administered by the SocialSecurity Administration that pays monthly bene-fits to aged and disabled individuals who havelow incomes and limited financial resources.

Survey of Consumer Finances (SCF): A surveyof U.S. households conducted by the FederalReserve Board every three years (since 1983) ofthe balance sheet, pension, income, and otherdemographic characteristics of U.S. families. Thesurvey also gathers information on the use offinancial institutions. The SCF oversampleshigher income families.

Survey of Income and Program Participation(SIPP): A survey of U.S. households conductedby the U.S. Census Bureau of the demographicand economic characteristics of persons andfamilies. It provides detailed information aboutsources and amounts of monthly income, taxes,assets, liabilities, and participation in govern-ment benefit programs. The SIPP oversampleslower income families.

Symmetric joint-life annuity: An annuity thatpays the same amount to a widowed primaryannuitant as would be paid to a widowed sec-ondary annuitant. The payment to the longer-lived person could be 100 percent, 75 percent,

Glossary 215

67 percent, or any other fraction of the amountpaid while both were alive.

Temporary Assistance to Needy Families(TANF): Created by the Personal Responsibilityand Work Opportunity Reconciliation Act of1996, TANF replaced Aid to Families withDependent Children (AFDC) and the JobOpportunities and Basic Skills Training (JOBS)programs. TANF provides assistance and workopportunities to low-income families by grantingstates federal funds and wide flexibility to devel-op and implement welfare programs.

Ten-year certain life annuity: A type of period-certain life annuity that provides payments forten years, even if the annuitant dies within tenyears. Period-certain life annuities guaranteepayments for a specific amount of time even ifthe annuitant dies before the period has elapsed.

Treasury Bond: An IOU issued by the federalgovernment; also called debt. By selling a bond,the federal government borrows money from theinvestor who purchases it and has a legal con-tract to pay it back with interest. Treasury alsosells short-term debt, called notes and bills.

Treasury Inflation-Indexed Securities (TIPS): Aspecial class of Treasury securities in whichinterest and redemption payments are tied toinflation. Like other Treasury securities, TIPSmake interest payments every six months toholders and pay back the principal when thesecurity matures. Unlike other securities, theTreasury Department adjusts the principal valueof TIPS daily, based on the consumer priceindex.

Thrift Savings Plan (TSP): A retirement savingsand investment plan for federal employees,established in 1986 as part of the FederalEmployees' Retirement System Act and adminis-tered by the Federal Retirement ThriftInvestment Board (FRTIB). The TSP is a tax-deferred defined-contribution plan similar to aprivate 401(k) plan.

Variable life annuity: An annuity that is tied tothe performance of a particular investment port-folio, such as corporate stocks or bonds.Payments can go down as well as up. The annui-tant bears all or part of the investment risk.


217

References

Advisory Council on Social Security (ACSS).1996. Individual Account Plan of the AdvisoryCouncil on Social Security, 1994-1996;Chairman Edward M. Gramlich.

Ameriks, John. 1999. “The Retirement Patternsand Annuity Decisions of a Cohort of TIAA-CREF Participants.” Research Dialogue, Issue60. New York, NY: TIAA-CREF Institute.

————. 2002. “Recent Trends in the Selectionof Retirement Income Streams Among TIAA-CREF Participants.” Research Dialogue, Issue74. New York, NY: TIAA-CREF Institute.

Ball, Robert M. 2002. Personal communication.

————. 2003. Social Security Plus.

Barr, Michael S. 2004. “Banking the Poor.” YaleJournal on Regulation 21, pp. 121-237. NewHaven, CT: Yale Law School.

Benna, R. Theodore. 1997. Helping EmployeesAchieve Retirement Security: Features Answersto the Most Often Asked 401(k) Questions.Washington Depot, CT: Investors Press, Inc.

Bitsberger, Timothy. 2004. Treasury DebtManagement, TIPS Slide Show (June). Availableat: http://www.treas.gov/offices/domestic-finance/key-initiatives/tips-presentation.pdf.

Blake, David and William Burrows. 2001.“Survivor Bonds: Helping to Hedge MortalityRisk.” The Journal of Risk and Insurance 68(2): 339-348.

Board of Trustees. 2003. Annual Report of theBoard of Trustees of the Federal Old-Age andSurvivors Insurance and Disability InsuranceTrust Funds. Washington, DC: Social SecurityAdministration.

Board of Trustees. 2004. Annual Report of theBoard of Trustees of the Federal Old-Age andSurvivors Insurance and Disability Insurance

Trust Funds. Washington, DC: Social SecurityAdministration.

Brown, Jeffrey R. 2002. “Differential Mortalityand the Value of Individual Account RetirementAnnuities.” In M. Feldstein and J. Liebman, TheDistributional Effects of Social Security Reform.Chicago, IL: University of Chicago Press.

————. 2003. “Redistribution and Insurance:Mandatory Annuitization with MortalityHeterogeneity.” The Journal of Risk andInsurance 70 (1): 17-41.

Brown, Jeffrey R., Olivia S. Mitchell, and JamesM. Poterba. 2000. “Inflation Risk, MortalityRisk and Annuity Products.” NBER WorkingPaper no. w7812. Cambridge, MA: NationalBureau of Economic Research. (See also forth-coming Z. Bodie et al., eds., FinancialInnovations in Retirement Income. Philadelphia,PA: The Pension Research Council of theWharton School of the University ofPennsylvania.)

————. 2001. “The Role of Real Annuitiesand Indexed Bonds in an Individual AccountRetirement System.” In Campbell, John andMartin Feldstein, eds., Risk Aspects ofInvestment Based Social Security Reform.Chicago, IL: University of Chicago Press.

————. 2002. “Mortality Risk, Inflation Risk,and Annuity Products,” in O. Mitchell, Z.Bodie, B. Hammond, and S. Zeldes, Innovationsin Retirement Financing, pp. 175-197.Philadelphia, PA: University of PennsylvaniaPress.

Brown, Jeffrey R. and Mark J. Warshawsky.2001. “Longevity-Insured RetirementDistributions from Pension Plans: Market andRegulatory Issues.” NBER Working Paper no.w8064. Cambridge, MA: National Bureau ofEconomic Research.

Burman, Leonard E., William Gale, MatthewHall and Peter R. Orszag. 2004. “DistributionalEffects of Defined Contribution Plans and


Individual Retirement Accounts.” TPCDiscussion Paper No. 16. Washington, DC: TaxPolicy Center.

Carney, Stacie and William Gale. 1999. “AssetAccumulation Among Low-IncomeHouseholds.” Washington, DC: The BrookingsInstitution.

Caskey, John P. 2001. “Reaching Out to theUnbanked.” Proceedings: April, pp. 81-92.Chicago: The Federal Reserve Bank of Chicago.

Charles Schwab and Company, Inc. 2004.Retirement Accounts for Individuals. Availableat: http://www.schwab.com.

Choi, James J., David Laibson, BrigitteMadrian, and Andrew Metrick. 2002. “DefinedContribution Pensions: Plan Rules, ParticipantDecisions, and the Path of Least Resistance.”The Pension Research Council Working Paper2002-3. Philadelphia, PA: The Pension ResearchCouncil of the Wharton School of the Universityof Pennsylvania.

Citro, Constance F. and Robert T. Michael.1995. Measuring Poverty: A New Approach.Washington, DC: National Academies Press.

City of Los Angeles Department of Water &Power v. Manhart, 435 U.S. 702 (1978).

Committee for Economic Development (CED).1997. Fixing Social Security: A Statement by theResearch and Policy Committee of theCommittee for Economic Development.Washington, DC: Committee for EconomicDevelopment.

Conservation & Liquidation Office. 2002.Serving the California Insurance Commissioner.Executive Life Insurance Company. Available at:http://www.caclo.org.

Copeland, Craig. 2001. “Pension Participation,February 2001.” EBRI Notes 22 (12).Washington, DC: Employee Benefit ResearchInstitute.

———. 2003a. “Individual Account RetirementPlans: An Analysis of the 2001 Survey ofConsumer Finances.” EBRI Issue Brief #259.Washington, DC: Employee Benefit ResearchInstitute.

———. 2003b. Unpublished tabulations fromthe Employee Benefit Research Institute basedon data from the 2001 Survey of ConsumerFinances.

Cronqvist, Henrik and Richard H. Thaler. 2004.“Design Choices in Privatized Social SecuritySystems: Learning from the SwedishExperience.” The American Economic Review94 (2): 424-428.

Davidoff, Tom, Jeffrey R. Brown, and Peter A.Diamond. 2003. “Annuities and IndividualWelfare.” NBER Working Paper no. w9714.Cambridge, MA: National Bureau of EconomicResearch.

Dawsey, Amanda and Lawrence Ausubel. 2002.“Informal Bankruptcy.” Twelfth Annual UtahWinter Finance Conference.

Diamond, Peter A. 2004. “Social Security.” TheAmerican Economic Review 94 (1): 1-24.

Dowd, Kevin. 2003. “Survivor Bonds: AComment on Blake and Burrows.” The Journalof Risk and Insurance 70 (2): 339-348.

Draut, Tamara and Javier Silva. 2003.“Borrowing to Make Ends Meet: The Growthof Credit Card Debt in the ‘90s.” New York,NY: D_mos: A Network for Ideas and Action.

Dukeminier, Jesse and James E. Krier. 2002.Property (Casebook). New York, NY: AspenPublishers, Inc.

Dunham, Constance and Timothy Bates. 2003.“Introduction to Focus Issue: Use of FinancialServices by Low Income Households.”Economic Development Quarterly 17 (1).

References 219

Executive Life Action Network (ELAN). 2004.Take Action. Available at: http://www.execu-tivelife.org/action.htm.

Favreault, Melissa and Frank J. Sammartino.2002. “The Impact of Social Security Reform onLow-Income and Older Women.” Public PolicyInstitute, #2002-11. Washington, DC: AARP.

Federal Retirement Thrift Investment Board(FRTIB). 2001. Summary of the Thrift SavingsPlan for Federal Employees. Available at:http://www.tsp.gov.

Federal Retirement Thrift Investment Board(FRTIB). 2002. Withdrawing Your TSP AccountAfter Leaving Federal Service. Office ofPersonnel Management. Washington, DC:Government Printing Office.

Fetherstonhaugh, David and Lee Ross.“Framing Effects and Income Flow Preferencesin Decisions about Social Security.” In Aaron,Henry J., ed., Behavioral Dimensions ofRetirement Economics. Washington, DC: TheBrookings Institution and New York, NY:Russell Sage Foundation.

Fidelity Investments. 1999. Building Futures:How American Companies Are Helping TheirEmployees Retire—A Report on CorporateDefined-Contribution Plans. FMR Corporation.

Fierst, Edith and Nancy Duff Campbell, eds.1988. Earnings Sharing in Social Security: AModel for Reform. Technical Committee onEarnings Sharing. Washington, DC: Center forWomen’s Policy Studies.

Finkelstein, Amy and James M. Poterba. 2000.“Adverse Selection in Insurance Markets:Policyholder Evidence from the U.S. AnnuityMarket.” NBER Working Paper no. w8045.Cambridge, MA: National Bureau of EconomicResearch.

Fleming v. Nestor, 363 U.S. 603 (1960).

Fronstin, Paul and Virginia P. Reno. 2001.“Recent Trends in Retiree Health Benefits andthe Role of COBRA Coverage.” Health andIncome Security Brief #4. Washington, DC:National Academy of Social Insurance.

Goss, Stephen C. 2003. “Estimated OASDIFinancial Effects for Two Provisions that WouldImprove Social Security Financing Plus aBalancing Tax-Rate Increase.” Memo to RobertM. Ball, October 10, 2003.

Goss, Stephen C. and Alice H. Wade. 2002.“Estimates of Financial Effects for Three ModelsDeveloped by the President’ Commission toStrengthen Social Security.” Office of the ChiefActuary. Baltimore, MD: Social SecurityAdministration.

Greacen, John M. 2002a. Self-RepresentedLitigants and Court and Legal ServicesResponses to Their Needs: What We Know, p.4,citing Office of the State Court Administrator,Florida, 1999. Report to the Florida Legislatureon Family Court Self-Help Programs.

————. 2002b. Self-Represented Litigants andCourt and Legal Services Responses to TheirNeeds: What We Know, p. 6, citing JudicialServices Division, Administrative Office of theCourts, 2001. An Analysis of Pro Se Litigants inWashington State, 1995-2000.

Graetz, Michael J. and Jerry L. Mashaw. 1999.True Security: Rethinking American SocialInsurance. New Haven, CT and London: YaleUniversity Press.

Haveman, Robert and Edward N. Wolff. 2001.“Who Are the Asset Poor? Levels, Trends andComposition, 1983-1998.” Discussion Paper no.1227-01. University of Wisconsin: Institute forResearch on Poverty.

Hewitt Associates. 2003. “Survey Findings:Trends and Experience in 401(k) Plans 2003.”p. 71. Lincolnshire, IL: Hewitt Associates.


Hill, Catherine and Virginia P. Reno. 2003.“Children’s Stake in Social Security.” SocialSecurity Brief No.14. Washington, DC: NationalAcademy of Social Insurance.

Hinson, Yvonne and Daniel Murphy. 1993. “Isthe 85-Percent Social Security Inclusion RationHigh Enough?” 59 Tax Notes 571; April 26.

Hisquierdo v. Hisquierdo, 439 U.S. 572 (1979).

Holden, Karen C. and Cathleen Zick. 1998.“Insuring Against the Consequences ofWidowhood in a Reformed Social SecuritySystem.” In Arnold, R. Douglas, Michael J.Graetz and Alicia Munnell, eds. Framing theSocial Security Debate: Values, Politics andEconomics. (Conference of the NationalAcademy of Social Insurance.) Washington, DC:National Academy of Social Insurance.

Holden, Sarah and Jack VanDerhei. 2003.“401(k) Plan Asset Allocation, AccountBalances, and Loan Activity in 2002.” EBRIIssue Brief # 261. Washington, DC: EmployeeBenefit Research Institute.

Hungerford, Thomas L. 1999. “Saving For aRainy Day: Does Pre-Retirement Access toRetirement Savings Increase Retirement Saving.”Draft Paper, May 24, 1999.

Insure.com. 2004. “Federal Option forInsurance Regulation Proposed.” Available at:http://www.insurance.com.

Iowa Insurance Division. 2004. “Selling YourLife Insurance Policy: Understanding ViaticalSettlements.” Available at: http://www.iid.state.ia.us.

Jocoby, Melissa B., Teresa A. Sullivan andElizabeth Warren. 2000. “Medical Problems andBankruptcy Filings.” Norton’s BankruptcyAdvisor. Eagan, MN: Thomson/West.

Johnson, E., Hershey, J., Miszaros, J., andKunreuther, H. 1993. “Framing, ProbabilityDistortions and Decisions.” Journal of Risk and

Uncertainty 7: 35-51. Cambridge, MA: KluwerAcademic Publishers.

Joint Committee on Taxation. 1999. Analysis ofIssues Relating to Social Security IndividualAccounts. Senate Committee on Finance, JCX-14-99.

Judicial Council of California, Task Force onUnrepresented Litigants. 2004. Statewide ActionPlan for Serving Self-Represented Litigants, p.2.

Kahneman, Daniel. 1999. Comment by DanielKahneman in Aaron, Henry J., ed., BehavioralDimensions of Retirement Economics.Washington, DC: The Brookings Institution andNew York, NY: Russell Sage Foundation.

King, Francis P. 1996. “Trends in the Selectionof TIAA-CREF Life-Annuity Income Options,1978-1994.” Research Dialogues, Issue 48. NewYork, NY: TIAA-CREF Institute.

Krueger, Alan B. 2004. “Economic Scene: SomeLessons From Sweden on the Pros and Cons ofPrivatizing Social Security.” The New YorkTimes, February 5, Section C, p. 2.

Langbein, John H. 2004. “Social Security andthe Private Pension System.” In Ghilarducci,Teresa, et al., In Search of Retirement Security:The Changing Mix of Social Insurance,Employee Benefits, and IndividualResponsibility. New York, NY: CenturyFoundation Press/National Academy of SocialInsurance, forthcoming January 2005.

Langbein, John H. and Bruce A. Wolk. 2000.Pension and Employee Benefit Law. ThirdEdition, University Casebook Series; p. 973.New York, NY: Foundation Press.

Larin, Kathy and Robert Greenstein. 1998.“Social Security Plans that Reduce SocialSecurity Retirement Benefits Substantially AreLikely to Cut Disability and Survivors Benefitsas Well.” Washington, DC: Center on Budgetand Policy Priorities.

References 221

Liebman, Jeffrey. 2002. The Role of Annuities ina Reformed U.S. Social Security System. AARPReport 2002-17. Project Managers Jules H.Lichtenstein and Alison M. Shelton.Washington, DC: AARP Public Policy Institute.

Life Insurance Marketing Research Association(LIMRA) International. 2004. Unpublished dataprovided in private correspondence to DouglasElliott.

Loewenstein, George. 1999. “Is More ChoiceAlways Better?” Social Security Brief No. 7.Washington, DC: National Academy of SocialInsurance.

Mackenzie, G.A. 2002. The Distribution Phaseof an Individual Accounts Reform of SocialSecurity: The Potential Role for Private SectorAnnuities. Working Paper, March 19.Washington, DC: The Urban Institute.

Madrian, Brigitte and D.F. Shea. 2001. “ThePower of Suggestion: Inertia in 401(k)Participation and Savings Behavior.” QuarterlyJournal of Economics 116 (4): 1149-1187.

Mashaw, Jerry L. and Virginia P. Reno. 1996a.“Vocational Rehabilitation and Return to WorkServices.” Table 6-4, in Balancing Security andOpportunity: The Challenge of DisabilityIncome Policy. Report of the National Academyof Social Insurance’s Disability Policy Panel.Washington, DC: National Academy of SocialInsurance.

————. 1996b. “The Fit Between Disabilityand Retirement Policy.” In Balancing Securityand Opportunity: The Challenge of DisabilityIncome Policy. Report of the National Academyof Social Insurance’s Disability Policy Panel.Washington, DC: National Academy of SocialInsurance.

McCarty v. McCarty, 453 U.S. 210 (1981).

Mills, Gregory, Rhiannon Patterson, Larry Orr,and Donna DeMarco. 2004. Evaluation of theAmerican Dream Demonstration: Final

Evaluation Report. Cambridge, MA: AbtAssociates.

Mishel, Lawrence, Jared Bernstein and HeatherBoushey. 2003. The State of Working America,2002/2003. Ithaca, NY: Cornell UniversityPress; p. 279.

Mitchell, Olivia S., James M. Poterba, Mark J.Warshawsky and Jeffrey R. Brown. 1999. “NewEvidence on the Money’s Worth of IndividualAnnuities.” The American Economic Review 89(5): 1299-1318.

Munnell, Alicia H. 2003. “The Declining Roleof Social Security.” Just the Facts on RetirementIssues, no. 6. Boston, MA: Center forRetirement Research, Boston College.

Munnell, Alicia H., Annika Sunden, andCatherine Taylor. 2003. “What Determines401(k) Participation and Contributions?” SocialSecurity Bulletin Perspectives 64 (3): 64-75.Washington, DC: Social Security Administration.

National Academy of Social Insurance (NASI).1996. Social Insurance Update, December.Washington, DC: National Academy of SocialInsurance.

National Commission on Retirement Policy(NCRP). 1999. The 21st Century RetirementSecurity Plan: Final Report of the NationalCommission on Retirement Policy. Washington,DC: Center for Strategic and InternationalStudies.

Nord, Christine Winquist and Jerry West. 2001.“Fathers’ and Mothers’ Involvement in TheirChildren’s Schools by Family Type and ResidentStatus.” Education Statistics Quarterly 3 (2).Washington, DC: National Center for EducationStatistics. Available at: http://nces.ed.gov.

Office of Community Services. 2001. IndividualDevelopment Account (IDA) Frequently AskedQuestions. Administration for Children andFamilies, U.S. Department of Health and

´


Human Services. Available at:http://www.acf.hhs.gov.

Orszag, Peter R. 2003. Unpublished tabulationsof data from the Brookings Institution, TaxPolicy Center, Washington, DC.

President’s Commission to Strengthen SocialSecurity (PCSSS). 2001. Report of the President’sCommission to Strengthen Social Security.Strengthening Social Security and CreatingPersonal Wealth for All Americans. Washington,DC: Government Printing Office.

Schreiner, Mark, Margaret Clancy and MichaelSherraden. 2002. Final Report: SavingPerformance in the American DreamDemonstration, A National Demonstration ofIndividual Development Accounts. St. Louis,MO: Center for Social Development, GeorgeWarren Brown School of Social Work,Washington University.

Shammas, Carole, Marylynn Salmon, andMichel Dahlin. 1987. Inheritance in America:From Colonial Times to the Present. NewBrunswick and London: Rutgers UniversityPress.

Sherraden, Michael and Michael Barr. 2004.“Institutions and Inclusion in Saving Policy.”Paper presented at Building Assets, BuildingCredit Symposium at the Joint Center forHousing Studies, Boston, MA (HarvardUniversity) November 2003; revised March2004.

Sondergeld, Eric T. 1998. Project Design Series,Immediate Annuities: A 1998 Report. Hartford,CT: LIMRA International, Inc.

Stegman, Michael A. 2003. “Banking theUnbanked: Connecting Residents of SocialHousing to the Financial Mainstream.” InHousing and Social Change: An InternationalPerspective. London: Routledge.

Sunden, Annika. 2004. “Pension Reform inSweden.” In Ghilarducci, Teresa, et al., In

Search of Retirement Security: The ChangingMix of Social Insurance, Employee Benefits, andIndividual Responsibility. New York, NY:Century Foundation Press/National Academy ofSocial Insurance, forthcoming January 2005.

Svahn, John A. and Mary Ross. 1983. “SocialSecurity Amendments of 1983: LegislativeHistory and Summary of Provisions.” SocialSecurity Bulletin, July, pp. 3-48.

U.S. Census Bureau. 2001. Statistical Abstract ofthe United States. Vital Statistics, Table 117.Washington, DC: Government Printing Office.

————. 2002. Statistical Abstract of theUnited States. Washington, DC: GovernmentPrinting Office.

————. 2003. Statistical Abstract of theUnited States. Washington, DC: GovernmentPrinting Office.

U.S. Department of Labor, Bureau of LaborStatistics (BLS). 2003. National CompensationSurvey: Employee Benefits in Private Industry inthe United States, 2000. Washington, DC:Bureau of Labor Statistics.

————. 2004. National CompensationSurvey: Employee Benefits in Private Industry inthe United States, 2003. Washington, DC:Bureau of Labor Statistics.

U.S. Department of Labor, Pension and WelfareBenefits Administration. 2001–2002. Abstract of1998 Form 5550 Annual Reports. PrivatePensions Plan Bulletin, No. 11, pp. 81-82.Washington, DC: Department of Labor.

U.S. Department of the Treasury. 2003. Bureauof the Public Debt. Treasury Inflation-IndexedSecurities (TIPS). Available at: http://www.publicdebt.treas.gov.

U.S. General Accounting Office (GAO). 1992.Insurer Failures: Regulators Failed to Respondin Timely and Forceful Manner in Four LargeLife Insurer Failures, Testimony Before the

´

References 223

Subcommittee on Energy and Commerce, Houseof Representatives, Statement of Richard L.Fogel, Assistant Comptroller General, September9. GAO/T-GGD-92-43. Washington, DC:Government Printing Office.

————. 1997. 401(k) Pension Plans: LoanProvisions Enhance Participation But MayAffect Income Security for Some. GAO/HEHS-98-5. Washington, DC: Government PrintingOffice.

————. 2002. Electronic Transfers: Use byFederal Payment Recipients Has Increased butObstacles to Greater Participation Remain.GAO-02-913. Washington, DC: GovernmentPrinting Office.

————. 2004. Social Security: Distribution ofBenefits and Taxes Relative to Earnings Level.GAO-04-747. Washington, DC: GovernmentPrinting Office.

U.S. Office of Management and Budget. 2004.Budget of the United States, FY 2004. Table 6-1,6-3. Washington, DC: Government PrintingOffice.

U.S. Social Security Administration (SSA). 1979.Income of the Population 55 or Older, 1976.Office of Policy, Office of Research, Evaluationand Statistics. Washington, DC: Social SecurityAdministration.

————. 1982. “Gifts to the Trust Funds.”Social Security Bulletin, February, pp. 34-45.Office of Policy, Office of Research, Evaluationand Statistics. Washington, DC: Social SecurityAdministration.

————. 2001a. Income of Disabled-WorkerBeneficiaries. Office of Policy, Office ofResearch, Evaluation and Statistics. Washington,DC: Social Security Administration.

————. 2001b. “Present Values of Benefits toIllustrative Survivors and Disability Cases.”Memorandum to Stephen Goss, Chief Actuary,

July 23. Office of the Chief Actuary. Baltimore,MD: Social Security Administration.

————. 2002a. Annual Statistical Supplementto the Social Security Bulletin, 2002. Office ofPolicy, Office of Research, Evaluation andStatistics. Washington, DC: Social SecurityAdministration.

————. 2002b. Annual Statistical Report onthe Social Security Disability InsuranceProgram, 2001. Office of Policy, Office ofResearch, Evaluation and Statistics. Washington,DC: Social Security Administration.

————. 2003a. Unpublished calculations fromthe Office of the Chief Actuary, August 2003.

————. 2003b. Office of Policy, Office ofResearch, Evaluation and Statistics. Unpublisheddata on number of persons dually entitled asretired worker and divorced spouse or survivingdivorce spouse.

————. 2003c. “Utility of Older ReinstatedWages from the Earnings Suspense File.” Officeof the Inspector General. Audit Report #A-03-02-22076. Available at: http://www.ssa.gov.

————. 2003d. “Fact Sheet on the Old-Age,Survivors, and Disability Insurance Program.”Available at: http://www.ssa.gov.

————. 2004a. Annual Statistical Supplementto the Social Security Bulletin, 2003. Office ofPolicy, Office of Research, Evaluation andStatistics. Washington, DC: Social SecurityAdministration.

————. 2004b. Selected solvency memorandafrom the Office of the Chief Actuary. Baltimore,MD: Social Security Administration.

————. 2004c. Automatic Increases, Cost-of-Living Adjustments. Updated October 19, 2004.Available at: http://www.ssa.gov.

————. 2004d. Personal correspondence withthe Office of the Chief Actuary, 2004.


————. (Forthcoming). Income of thePopulation 55 or Older, 2002. Office of Policy,Office of Research, Evaluation and Statistics.Washington, DC: Social Security Administration.

Watson Wyatt. 1998. “Choosey EmployeesChoose Lump Sums!” Watson Wyatt Insider 8(4). Washington, DC: Watson Wyatt Worldwide.

————. 2003. Impaired Life Annuity Survey.Washington, DC: Watson Wyatt Worldwide.

Williams, Cecili Thompson, Virginia P. Reno,and John F. Burton, Jr. 2003. Workers’Compensation: Benefits, Coverage, and Costs,2001. Washington, D.C.: National Academy ofSocial Insurance.

Yen, Hope. 2004. “Supreme Court PondersIRA-Bankruptcy Case.” Seattle Post-Intelligencer, December 1.

NATIONAL ACADEMY OF SOCIAL INSURANCE1776 Massachusetts Avenue, NW

Suite 615Fax: 202/452-8111

Washington, DC 20036

202/452-8097

www.nasi.org

$29.95

insert UNION BUG

Date post:	27-Aug-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Uncharted Waters - National Academy of Social Insurance (NASI) · 2017. 1. 10. · Study Panel...

Documents