“The expansion of
consumer credit
has created a
windfall of bene-
fits to both
economies and
consumers, but
that growth has
paralleled a sub-
stantial increase
in the financial
insecurity of
Americans.”
M e t r o p o l i t a n P o l i c y P r o g r a m
The Brookings Institution
Credit Scores, Reports, andGetting Ahead in AmericaMatt Fellowes
FindingsConsumer credit reports and scores play a growing role in the ability of families to getahead, now influencing prices for loans and insurance and efforts to get jobs and rentapartments. An analysis of a quarterly sample of 25 million anonymous consumer creditreports and scores for every U.S. county between 1999 and 2004 reveals that:
■ Consumer credit scores widely vary across counties, with the South having thehighest concentration of consumers with weak credit scores. In 2004, among all con-sumers, the average score on a credit score index maintained by one of the majorbureaus was 656, out of a scale that ranges from 350 to over 850. Meanwhile, the aver-age credit score in the South was 635, and more than one in five borrowers in a typicalSouthern county have scores that suggest they are very risky borrowers.
■ Between 1999 and 2004, most counties with weak consumer credit scores sawdeclines in the average consumer credit score, while counties with strong scoresgenerally experienced modest gains. Nationwide, credit scores only modestly fell dur-ing this period, but the average Southern county experienced a larger decrease.
■ Counties with relatively high proportions of racial and ethnic minorities are morelikely to have lower average credit scores. This evidence does not suggest that a biasexists, or that there is a causal relationship between race and credit scores, raising questionsfor future research.
■ High homeownership rates and county per capita income are strongly associatedwith high consumer credit scores. The average county with a low, mean credit scorehad a per capita income of $26,636 and a homeownership rate of 63 percent in 2000.Meanwhile, the typical county with high average credit scores had higher per capitaincomes ($40,941) and higher shares of homeowners (73 percent).
■ Financial insecurity, primarily measured by the frequency of loan delinquencies,rose between 1999 and 2004. Over those five years, the proportion of mortgage bor-rowers 60 or more days late in their mortgage payments increased by 108 percent, fromone out of every 106 borrowers to one out of every 51. About one out of every 21 bor-rowers had at least one credit-bearing account 60 or more days past due in 2004.
Consumer credit reports and scores are playing a growing role in the economic mobility ofconsumers today. But rising consumer debt and loan delinquencies mandate that govern-ment leaders, with their private sector partners, pursue a series of reforms to increaseconsumer education and responsibility, market accountability, and accuracy.
MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES 1
Introduction
Everyday, more than 27,000employees in the creditbureau industry walk intoover 1,000 locations around
the country and process over 66 mil-lion items of information.1 Out of thismassive churning of activity, creditbureaus produce consumer creditreports and scores, two of the mostpowerful determinants of modernAmerican consumer life.2
Among their many applications,credit reports and scores now helpdetermine if a family can borrowmoney to buy major necessities likehomes and cars; they affect the pricesbusinesses charge for such products asmortgages and auto insurance; reportsand scores are used by an increasingnumber of employers to assess jobapplicants; they are used by landlordsto evaluate prospective renters; and agrowing number of utilities are usingcredit reports and scores to pricedeposits for numerous services.3 Inshort, both the access and terms ofaccess to an increasing array of basicnecessities, including jobs, housing,insurance, energy, and communica-tions, are now influenced by anindividual’s consumer credit reportand scores.4
Among other effects, the growingavailability of credit has contributed tothe recent surge in consumer debt(Figure 1). Businesses have substan-tially expanded access to loan productsamong consumers as a result of thecapacity credit reports and scores givebusinesses to predict lending risks.5
Among other benefits, this hasincreased access to assets like houses,given consumers more choices aboutmarket products, and spurred eco-nomic development in neighborhoodsonce ignored by creditors and insur-ance companies.
At the same time, consumer creditinformation has provided businesseswith a very sophisticated marketingtool, creating countless new opportu-nities to target consumer segments
with individually tailored products.Pre-approved credit card offers, forinstance, are made possible by thereadily available information in con-sumer credit reports.6
While this expansion of consumerdebt has created a windfall of benefitsto both economies and consumers, itsgrowth has paralleled a substantialincrease in the financial insecurity ofAmerican consumers. As one indica-tion of this, between 1980 and 2004the personal bankruptcy rate skyrock-eted from about one out of every1,000 individuals to about five out of
every 1,000 individuals (Figure 2).7 Atthe same time, bankruptcy is affectinga broader cross section of Americans,becoming, some experts say, a “middleclass phenomenon.”8
Among the numerous factors con-tributing to this growing financialinsecurity, consumers frequently pointto the proliferation of credit offers andto their inability to understand all ofthe many choices they are now con-fronted with in the financial servicesmarket.9 Not understanding how toresponsibly manage debt has becomeincreasingly costly for families as debt
2 MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Source: Author’s analysis of data from the Bureau of Economic Analysis, Bureau of Labor Statistics,
and the Federal Reserve
Figure 1. Consumer Debt Has Substantially Increased Over Time
$2,500
$2,000
$1,500
$1,000
$500
0
(in
billi
ons
)
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
Total Consumer Debt(2005 dollars)
25%
20%
15%
10%
5%
0
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
Consumer Debt as a Percentage of Personal Income
has become more widely available.Expanding uses of consumer credit
reports and scores have consequentlyspurred benefits and costs to Americanconsumers and local economies. But,the geographic distribution of theseeffects is not widely known, stuntingthe development of appropriate, localpolicy responses to these market prod-ucts. Many consumers likely needmore information about credit reportand scores, for instance, as they grap-ple with the increasing number ofchoices made possible by creditreports. Similarly, consumer awareness
about their role overseeing the accu-racy of their credit report is now moreimportant than ever.
More generally, the lack of visibilityof these issues has meant that aresearch agenda geared toward under-standing credit reports and the waysthey are used to set prices and informmarket decisions is underdevelopedand under-funded. On the most ele-mental issue about the accuracy ofcredit reports, for instance, there is awidely held sense that the bill paymentinformation not collected by bureausdrives down the credit scores of mil-
lions of borrowers, and drives-up thecost of credit associated with thosescores.10 But, empirical evidence thatspeaks to that point, and the generaldistributional effects of nontraditionaldata on consumer credit scores, isthin.11 There are also very differentassessments about the accuracy ofinformation currently collected by thebureaus.12
Also there are important questionsleft unanswered about the appropri-ateness of market responses to creditscores. There is no public data avail-able, for instance, that speaks to theoptimal level of mortgage price fluctu-ation across different levels of risk.Put differently, the price-point wherehigher prices for mortgage borrowerswith low credit scores becomes price-gouging rather than just cost-coveringis not clear. In a market that functionsperfectly, competition would driveprices to that point. But evidence thatconsumers are under-informed cer-tainly creates an incentive forover-charging.13 On the other hand, itis not clear that government can (orshould) set such a price point.
In short, the amount of informationabout credit reports, scores, and themarket applications of both, is out ofstep with the importance these marketproducts now play in the lives of con-sumers.
To begin to address the need for thisinformation, this paper analyzes howinformation in credit reports, and oneof the credit scores that information isused to calculate, varies across thecountry. We also examine the relation-ship between those scores anddelinquencies and socioeconomic fac-tors. To do this, we use a uniquedatabase based on a sample of 25 mil-lion consumer reports and scores inevery quarter between 1999 and 2004.This information is used to build pro-files of consumers in every county inthe United States.
We also develop a policy agenda forpolitical and business leaders torespond to the fundamental role thatcredit reports and scores now play in
3MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Source: Author’s analysis of data from the American Bankruptcy Institute
Figure 2. Personal Bankruptcy Filings Have SubstantiallyIncreased Over Time
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
(in
mill
ions
)
1980
1981
1982
1983
1984
1985
1986
1986
1986
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
6
5
4
3
2
1
0
(Num
ber
of B
ankr
uptc
ies
per
1,00
0 pe
opl
e)
1980
1981
1982
1983
1984
1985
1986
1986
1986
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
Total Number of Personal Bankruptcy Filings
Bankruptcy Rate
the lives of American consumers andthe economy. It is time for governmentleaders, with their private sector part-ners, to pursue a series of reforms toincrease consumer education andresponsibility, market accountability,and accuracy.
Methodology
About the DataData for this project was obtainedfrom TransUnion’s trend database.TransUnion is one of the three majorcredit bureaus that collect financialinformation on nearly every consumerin the U.S. that has some type ofcredit account. All available data inthe trend data were aggregated fromdepersonalized consumer creditreports.14
For this analysis, we use an anony-mous sample of American consumersthat had a credit report on file withTransUnion between 1999 and 2004.In each of the 24 quarters duringthese six years, a random sample ofapproximately 25 million borrowerswas extracted from the population ofAmerican borrowers with a creditreport. This sample was used to createnational, quarterly estimates related tocredit scores and delinquencies inevery quarter between 1999 and2004.15
Importantly, some estimates inTransUnion’s trend database differfrom more readily available estimates.For instance, John M. Barron and hiscolleagues report that 1998 Tran-sUnion estimates of bankcarddelinquencies differ from estimatesavailable from the American BankersAssociation, even if the metrics are notidentical.16 Similarly, our own 2004estimates of bankcard debt differ fromother, recent estimates. To date, a fullaccount explaining those differences isnot available, although the bureau’smuch larger population of reportinginstitutions may play a role in explain-ing these differences. There are alsodifferences in the metrics used to
measure variables. For these reasons,the numbers reported in this analysisshould be interpreted as estimates.
Besides the national estimates, wealso report estimates for every countyin the country. For instance, we reportthe average consumer credit score in acounty, the proportion of consumers ina county that are in different creditscore ranges, and the proportion ofborrowers in a county who are delin-quent on different types of loans.
These county level data allow us tolook at how credit scores and informa-tion in credit reports varies across thecountry. At the same time, these datacan be linked with socioeconomicinformation from counties, whichgives us the opportunity to analyzehow information in credit reportsvaries with different county socioeco-nomic profiles. Also, county level datais just the type of local data policymak-ers need to understand the importanceof credit scores in the lives of theirconstituents.17
About Credit Reports and ScoresCredit reports contain four generaltypes of information. First, there isinformation related to the identity of aconsumer, including a consumer’sname, address, social security number,and date of birth, among other similarinformation. Second, credit reportscontain information related to an indi-vidual’s use of credit-based products.For most consumers, this includesinformation related to mortgages, creditcards, retail credit cards, and autoloans, among others. A growing numberof consumers also have their utility pay-ment histories reported and chronicledby credit bureaus, along with nontradi-tional loans, like payday loans.18 Third,there is an inquiry history of applica-tions for credit. Finally, public recordinformation related to a consumer’sfinancial health is tracked by creditbureaus. This includes declarations ofbankruptcy, along with a range of otherrelated records that are publicly avail-able. Together, this informationcomprises a consumer’s credit report.
Among the numerous uses of thisinformation, bureaus create creditscores, or sell this information to otherinstitutions that calculate their owncredit scores. Credit scores are calcu-lated differently for different marketapplications and for different compa-nies. In general, though, scores are afunction of numerous factors relatedto the financial life of a consumer,including an individual’s payment his-tory, debt-to-equity ratio, length ofcredit history, extant types of extendedcredit, and numerous additional vari-ables related to recent transactions.19
Although the weights assigned to eachof these general classes of variablesare available for some credit scores,the specific variables and the specificweight assigned to these variablesremains the private property of theinstitutions that develop them.20
Typically, consumer credit scoresare scaled for ease of use to rangebetween 350 and 850, where highernumbers represent lower levels ofrisks—or a lower probability of a delin-quency or default—and lowernumbers indicate a higher level of risk.Nonetheless, each bureau, along withthe company that pools together thebureau’s reports into a single score(myFICO), has a modestly uniquerange. The TransUnion Auto Modelranges between values of 300-900, forinstance. For this analysis, we rely onTransUnion’s credit score in the trenddatabase, which ranges between 350and over 850.
About the AnalysisTo account for why consumer creditscores vary so widely across the coun-try, we first consider how thisinformation varies across the majorCensus regions and 10 divisions. NewEngland consists of counties in Maine,New Hampshire, Vermont, Massachu-setts, Rhode Island, and Connecticut;the Middle Atlantic includes countiesin New York, Pennsylvania, New Jer-sey; the South Atlantic includes theDistrict of Columbia and counties inDelaware, Maryland, West Virginia,
4 MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Virginia, North Carolina, South Car-olina, Georgia, and Florida; the EastSouth Central division includes coun-ties in Kentucky, Tennessee, Alabama,and Mississippi; the West South Cen-tral division includes counties inTexas, Oklahoma, Arkansas, andLouisiana; the East North Centralincludes counties in Ohio, Michigan,Indiana, Illinois, and Wisconsin; theWest North Central includes countiesin Minnesota, Iowa, Missouri, NorthDakota, South Dakota; the Mountaindivision includes Montana, Wyoming,Colorado, Utah, Idaho, Nevada, Ari-zona, and New Mexico; and the Pacificdivision includes Washington, Oregon,California, Alaska, and Hawaii.
Because we have county-level data,we make inferences about consumersin these regions and divisions by tak-ing population-weighted averages of
counties within these areas. It wouldhave been more ideal to weight theseaverages by the number of borrowersin each county, but that informationwas not available. We assume that thedistribution of people across countiesclosely resembles the overall distribu-tion of borrowers across counties.
Data for county populations isbased on the U.S. Census Bureaucounty population projections. Theseestimates are based on a methodreferred to as the “administrativerecords component of populationchange” method. This techniqueessentially uses a wide range of annu-ally available administrative records tomake inferences about populationchange over time.21
Next, we consider how consumercredit scores vary across counties todetermine the different levels of finan-
cial insecurity across the country. Weconcentrate on delinquencies as ameasure of financial insecurity, includ-ing the proportion of consumers in acounty with any loan that is more than60 days past due, along with the pro-portion of consumers that aredelinquent on their mortgages andbankcards. All three measures arecompiled by the same credit bureauthat collected the consumer creditscores analyzed in this paper.
We also assess the relationshipbetween numerous socioeconomiccharacteristics and both consumercredit scores and financial insecurity.The major characteristics we considerare: median county income, the pro-portion of black county residents, theproportion of Hispanic residents, andthe proportion of homeowners in acounty. The most recent data on these
5MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Source: Author’s analysis of data in TransUnion’s trend database
Note: All available data in the trend database were aggregated from depersonalized consumer credit reports. Data are displayed by country and in quintiles.
Figure 3. Credit Scores Widely Vary Across U.S. Counties
Average Consumer Credit Score (2004)Lower Risk Area (Average credit score greater than 690)
Higher Risk Area (Average credit score less than 624)
characteristics for the entire popula-tion of U.S. counties is the long formsurvey included in the 2000 U.S. Cen-sus.22 This form was completed byapproximately one out of every sixAmerican households in 2000, orabout 19 million different housingunits. Using this information, Censusweighted the response to approximatethe population in each county of thecountry, creating the richest resourcecurrently available to analyze socioeco-nomic characteristics across counties.
Finally, this study includes an analy-sis of specific types of loans in a creditreport, which includes revolving loans,non-revolving loans, and mortgages.Revolving loans, such as those avail-able through credit cards or homeequity loans, are lines of credit thatprovide a continuous source of creditwithin some predetermined limit; non-
revolving loans are one-time lines ofcredit, such as automobile or educa-tion loans, that usually close once theprincipal and any interest is paid off;and mortgage loans are all loanssecured by a home.
Findings
A. Consumer credit scores widelyvary across counties, with theSouth having the highest concen-tration of consumers with weakcredit scores. Out of a possible range between 350and over 850 (where higher numbersindicate lower risks for underwriting)the average consumer credit score was656 in 2004. Around this mean, thereis a fairly flat distribution: about 55percent of the population has scores
between 600 and 800 on this scale.23
On either side of this central tendency,about 20 percent of the populationhas scores less than 600, and aboutanother 25 percent have scores above800. While most consumers in theUnited States are clustered around theaverage score, the “high” and “low”risk consumers are concentratedtogether in fairly systematic waysacross the country.
In general, consumers in the typicalSouthern county have much lowercredit scores than elsewhere in thecountry. While the typical consumer inthe Western, Midwest, and Northeastcounties had credit scores that rangedbetween 660 and 675, the averageamong Southern counties was 635 in2004. Because this scale is constantacross the country, this indicates thatthe average consumer in a Southern
6 MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Source: Author’s analysis of data in TransUnion’s trend database
Note: All available data in the trend database were aggregated from depersonalized consumer credit reports. Data are displayed by country and in quintiles;
low consumer credit scores are less than 492, bottom 10 percent range of scores in the validation sample.
Figure 4. Southern Counties Have High Proportions of Consumers with Very Weak Credit Scores
Proportion of Low Consumer Credit Scores to All Scores (2004)Low (Less than 13.00%)
High (More than 24.14%)
county appears more of a credit riskthan the average consumer in otherareas of the country. This carriesimportant implications for the cost ofcredit across different areas of thecountry.
To see this, Figure 5 illustrates therelationship between consumer creditscores, interest rates, and the annualcost of a $150,000 mortgage.24 Theseconsumer credit scores are averageFICO scores, based on the threeFICO-branded scores estimated byeach of the three major bureaus.According to the company, FICOscores are the “credit scores mostlenders use to estimate risk.”25 Thecompany illustrates on its web pagethe relationship between interest ratesand credit score categories—a practicealso used by creditors. Here, interestrates range from a low of 5.3 percentcharged to consumers with scoresbetween 760–850 to a high of 9.3 per-cent charged to consumers with scoresbetween 500 and 559. That differenceadds up to nearly $5,000 every year inextra payments that are charged toconsumers with scores in the lowerrange.
Certainly a high-risk borrower maybe more than happy to pay an extra$5,000 to qualify for a mortgage thatthey may have been turned down for
in the past. Their credit report indi-cates that they are a higher risk thanother consumers, and financial institu-tions now can rationally pass on thathigher risk through higher prices,whereas in the past there may nothave been an offer of credit extended.Still, those higher costs are not with-out costs that both consumers andcommunities should take seriously.Most importantly, higher prices takemoney off the table for other types ofinvestments that can help families getahead. Similarly, when high-risk bor-rowers are clustered together, theeffects of these higher costs may spillover into the community by drainingconsumer spending away from retail,homeownership, home improvements,and educations. For both families andleaders, it is therefore important tounderstand consumer credit scores,and the strategies needed to improvescores.
Credit report data do point to clus-ters of extremely high risk andextremely low risk areas of the country(Figure 4). To illustrate this we con-sider the eight credit score intervalsreported in TransUnion’s trend data-base: scores less than 421, between422–492, 493–594, 595–700,701–795, 796–839, 840–850, andgreater then 850. Together, the pro-
portions of consumers in these cate-gories represent the universe of allborrowers in each county in the coun-try. To assess counties with unusuallyhigh and low proportions of borrowerson either end of this credit score dis-tribution, we look at the bottom andtop two intervals. In particular, theproportion of a county’s residents withvery weak credit scores is measured asthe proportion of borrowers that havescores below 492, which includes thelowest two categories tracked by ourdata source. The proportion of acounty’s residents with very strongcredit scores is measured as the pro-portion of borrowers that have scoresabove 840, which includes the twohighest categories tracked by our datasource.26
Among the Southern counties inour analysis, an average of over 22percent of borrowers in a county hasvery weak credit scores, or scoresbelow 492. That suggests more thanone out of every five borrowers in atypical Southern county may not haveaccess to additional credit or, at thevery least, pays a substantial premiumfor it. In contrast, about 16 percent ofcurrent borrowers in a typical countylocated in other major regions hadscores under 492 in 2004.
Why do Southern counties havesuch large proportions of borrowerswith low credit scores?
Most clearly, the information inthese borrowers’ credit reports tends toshow more risky behavior, because ofhigher delinquency rates, higher debtto equity loads, fewer lines of opencredit, and so on. But these data donot explain what is driving this morerisky behavior. Why are Southern bor-rowers, for instance, more likely to fallbehind on payments than borrowers inother regions? Is it something to dowith systematic differences in their liv-ing expenses or wages? Or, does ithave something to do with the types ofbusinesses selling products in thisregion? That such large proportions ofborrowers in this region stand out forhaving such weak credit scores begs
7MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Source: Author’s analysis of data from Fair Isaac.
Figure 5. Credit Scores are Strongly Related to Mortgage Terms
(Credit Score Range)
500–559 560–619 620–639 640–659 660–679 680–699 700–759 760–850
$14,856$13,884
$11,832$11,184
$10,680 $10,440 $10,236 $9,984
10%
9%
8%
7%
6%
5%
4%
3%
2%
1%
0%
Annual Cost of a $150,000, 30 year, fixed rate mortgage
Interest Rate
further research in the future. With-out that attention, large proportions ofconsumers in Southern counties willlikely continue to have credit scoresthat limit access to credit products oronly qualify them for very high-pricedmortgages, auto loans, and all of theother credit-based products familiesrely on today.
On the other hand, none of the U.S.regions stand out as having an unusu-ally high number of consumers withextremely strong credit scores, orscores in the highest two categoriestracked by TransUnion’s trend data-base. Among all U.S. counties in 2004,an average of 12 percent of consumersin a county had scores above 840.Across the four major regions, theaverage county in each region withconsumers in this extremely low riskcategory varied between 11 and 14 per-cent of all consumers, suggesting thatthe typical county across each of theregions had about the same proportionof extremely low-risk consumers.
B. Over time, most counties withweak consumer credit scores sawsharp declines in the averageconsumer credit score, whilecounties with strong scores gener-ally experienced modest gains.27
This trend is illustrated in Figure 6.The average consumer credit score in1999 for every U.S. county is listed onthe horizontal axis; the vertical axisdisplays the absolute change in theaverage consumer credit scorebetween 1999 and 2004, by county.Organizing the data in this way illus-trates how consumer credit scoreschange over time in counties with low,average, and higher consumer creditscores, relative to the rest of the coun-try. The trend is unmistakable.
Counties that had low consumercredit scores in 1999 relative to thenational average generally saw thataverage drop in value by 2004. In par-ticular, counties with average creditscores lower than 90 percent of theother counties in the country in 1999saw that average drop by an average of
17 points by 2004, or from 611 in1999 to 594 by 2004.
But counties with average con-sumer credit scores in 1999 showednearly no discernable trend duringthis same period. Moreover, countiesthat had very high average consumercredit scores in 1999 had modestlyhigher values by 2004. In particular,counties that had higher averagecredit scores than 90 percent of theother counties in the country in 1999saw that average modestly increase bytwo points by 2004, or from 708 in1999 to 710 in 2004.
Some of the sharpest declines inaverage, county credit scores occurredin the South. Between 1999 and2004, the average credit score inSouthern counties fell by an averageof 10 points, compared to a nation-wide decrease of just two points. Evenmore dramatically, in both 1999 and2004, 93 percent of the counties withaverage credit scores lower than 90percent of the other counties in thecountry were located in the South.
At least when looked at from thisaggregated level, these trends suggest
consumer credit scores are somewhatpath dependent, with particularly seri-ous consequences for Southernconsumers. Areas of the country withhigher average lending risks amongconsumers saw that average lendingrisk increase over time, whereas areaswith lower average lending risks sawthat risk drop even lower over time.
This trend points to a potentiallyruinous fiscal cycle for consumerswith low credit scores, recently exam-ined in more detail by Dean S. Karlenand Jonathan Zinman.29 While creditscores create an opportunity to under-write high-risk consumers thatopportunity comes with a price, suchas higher-priced mortgage loans. Inturn, higher prices may make high-risk consumers more likely to miss billpayments than consumers with strongcredit scores. Gaining access tohigher priced credit, in other words,may not always be a wise financialdecision, particularly for families whohave trouble paying bills on time. Thisrisk is likely pronounced when lowerincome households are the high-riskborrowers.
8 MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Source: Author’s analysis of data in TransUnion’s trend database
Note: All available data in the TransUnion trend database was aggregated from depersonalized
consumer credit reports
Figure 6. Counties with Strong Credit Scores Tend to ImproveOver Time; Counties with Weak Scores Decline
This trend may be reinforced bysome questionable business practicesused for high-risk borrowers. In themortgage industry, for instance, somelenders sell negatively amortizing linesof credit to high-risk clients, most ofwhich really should not buy this prod-uct. Negative amortization on a line ofcredit means that the monthly pay-ment for that credit does not accountfor the full amount of monthly inter-est charged for the credit. This maymake sense for high net-worth indi-viduals, or for individuals who onlytemporarily need to make small pay-ments. But, for high-risk clients whocan only afford a loan by accumulat-ing more debt on their principalbalance, this may lock their creditscore into a downward path.
Along those same lines, universaldefault policies on credit cards alsomay contribute to this trend. A univer-sal default policy means that a creditcard company automatically booststhe APR (annual percentage rate) onan individual’s credit card if theircredit report shows that the borrowerhas recently missed a payment onanother line of credit. This policymeans that one late payment can addup to higher payments on numerouslines of credit, increasing the chancesfor further missed or late payments.In this way, high-risk borrowers mayspiral into lower and lower creditscore categories.
Meanwhile, it is important to pointout again that credit scores have madeunderwriting many high-risk con-sumers possible, where in the past itmay not have been even considered.
C. Counties with relatively highproportions of racial and ethnicminorities are more likely to havelower average credit scores.Credit scores are an assessment aboutthe level of numerous types of riskposed by consumers, such as to a cred-itor or an employer. The models thatcalculate credit scores do not includea person’s race, so it is unclear howthese scores could discriminate
against a borrower, a problem that thelending and insurance industry hasgrappled with in the past.30 The scoresalso do not include any informationabout the neighborhood that the bor-rower lives in, which makes it difficultto see how scores can redline commu-nities, another problem that hasplagued these industries.
For these reasons, credit scores are
a substantial improvement over theoften subjective measures used toevaluate risk in the past. Still, we dofind that that county racial profiles areassociated with the average creditscore of a consumer. In other words,the higher the concentration of racialor ethnic minorities in a county, themore likely the county’s average creditscore will be low. This does not suggest
9MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Table 1. Race and County Credit Score Profiles
The average racial profile of counties in each risk category
Proportion Proportion
Credit Score Risk Categories Black Hispanic
850–720 (Very Low Risk) 1% 4%
700–719 5% 5%
675–699 5% 8%
620–674 12% 14%
560–619 28% 19%
500–559 (Very High Risk) 26% 23%
Source: Author’s analysis of data from TransUnion’s trend database and the U.S. Census Bureau.
Note: All available data in the TransUnion trend database was aggregated from depersonalized con-
sumer credit reports. Credit score categories correspond with the credit score risk categories reported
by Fair Issac in October 2005.
Table 2. Wealth, Homeownership Rates, and County CreditScore Profiles
Mean Terms for
a 30 year fixed-rate,
$150,000
County Wealth mortgage
Annual
Credit Score Per-Capita Homeowner Unemployment Interest Payments
Risk Categories Income Rate Rate Rate Due
850–720 (Very Low Risk) $41,384 75% 4% 5.7% $10,404
700–719 $40,946 73% 4% 5.8% $10,536
675–699 $36,280 69% 5% 6.3% $11,160
620–674 $29,391 65% 6% 7.5% $12,552
560–619 $26,636 63% 7% 8.5% $13,884
500–559 (Very High Risk) $17,136 72% 10% 9.3% $14,856
Source: Author’s analysis data in TransUnion’s trend database and the U.S. Census Bureau.
Note: All available data in the TransUnion trend database was aggregated from depersonalized con-
sumer credit reports. Credit score categories correspond with the credit score risk categories reported
by Fair Issac in October 2005.
that a bias exists, or that there is acausal relationship between race andcredit scores. Instead, this associationreflects the numerous, historical dis-parities between races in the access toand availability of high quality educa-tion, well-paying jobs, and access toloans, among other factors. But thepresence of this relationship does raiseimportant questions that should beexplored through further research,particularly in instances where infor-mation in reports are being used innontraditional, under-researched mar-ket applications, like screening jobapplicants and pricing insurance.
To illustrate this relationship, firstconsider the simple associationsbetween a county’s racial and con-sumer credit score profile in Table 1.Here, we use the credit score cate-gories automatically generated by thebureau to assess this relationship.
Take the counties on either side ofthe central tendency: the approxi-mately 552 U.S. counties that havevery low average consumer creditscores between 560–619, and the 270counties that have very high averagescores between 700–719. In the coun-ties with a very low typical score,about 19 percent of the population isLatino and another 28 percent isblack. On the other hand, the countiesthat have higher typical credit scorestend to be essentially all white coun-ties. In particular, in the 270 countieswith average credit scores between700–719 only about 5.1 percent of thepopulation is Latino and just 1.1 per-cent is black.
This evidence does not suggest thatracial differences between countiescause these differences in scores.Instead, these data point to an associa-tion, which frankly is not very wellunderstood, and requires more rigor-ous analysis. For instance, the strongassociation in Table 1 could be over-stated because we have not accountedfor the variables factored into creditscores, like an individual’s paymenthistory, debt-to-equity ratio, length ofcredit history, extant types of extended
credit, and numerous additional vari-ables related to recent transactions.Once these variables are accountedfor, there may not be an independenteffect associated with a county racialprofile.31
Along those lines, there is evidencethat the depth of an individual’sknowledge about credit scores, and thesignificance of these scores in a fam-ily’s financial life, is strongly related tomany socioeconomic characteristics.32
Less information about this productmay lead to lower scores.33 Similarly,more risky credit behavior may bedriven by systematic differencesbetween borrowers with thin and fullcredit files, the effects of unscrupu-lous lenders of credit, and systematicdifferences in the labor market.
There also may be an additional,albeit far more technical, force at workhere obscuring the true relationshipbetween race and credit scores. Arecent analysis by economists at theFederal Reserve looked at problemsassociated with selective reporting ofaccount information to the bureaus.Their analysis suggests that the use ofa borrower’s highest balance on arevolving line of credit as a substitutefor the credit limit (which affects thecredit utilization component used tocalculate credit scores) may systemati-cally drive down credit scores for thosewho “are at the margins of credit wor-thiness.”34 Traditionally, that marginhas been occupied by more vulnerablesectors of the economy.
D. High homeownership ratesand county per capita income arestrongly associated with high con-sumer credit scores. Homeownership rates and county percapita income increase as the typicalconsumer credit score in a countyincreases in value. These findingspoint to an important opportunityleaders have to have families get aheadby boosting their creditworthiness.
Take the counties on either side ofthe central tendency that we exam-ined in the previous section (Table 2).
In the 552 counties with a very lowtypical score, the per capita income is$26,636, about 6.7 percent of thelabor force is unemployed, and about63 percent of the households own thehomes they live in. On the otherhand, the counties that have highertypical credit scores tend to havemuch more wealth and higheremployment rates. In particular, inthe 270 counties with average creditscores between 700–719, the percapita income is $40,946, or 54 per-cent higher than the counties withvery low average credit scores. About4.2 percent of the labor force isunemployed, which is quite a bitlower than the national average. And,about 73 percent of the householdsown the homes they live in, againhigher than the national average.
This evidence points to the strong,underlining association between thetypical consumer credit profile of acommunity and the per capita incomeand homeownership rates in a com-munity. Counties with good consumercredit scores have more homeownersand more personal income.
Why is this so? A big reason here isthat credit scores act as a gatekeeperfor creditors, insurers, even potentialemployers, in deciding if they shouldextend an offer to an individual. Incounties with lower average creditscores, a higher proportion of peoplemay be turned down for an applicationfor a loan or insurance than in acounty with a higher proportion ofpeople with higher average creditscores.35 When added up across all ofthe consumers in a county, we seemajor differences in the comparativehomeownership rates of communities.In the example above, for instance, thecommunities with a very low averagecredit score have a homeownershiprate 16 percent lower than the com-munities with a very high averagecredit score.
Differences in credit scores alsohave important implications for theprices consumers are charged forthese necessities. In the counties with
10 MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
very high average scores, for instance,the annual payments for a $150,000mortgage would be about $10,536,while the payments for the same loanin the counties with lower averagescores would be $14,856 every year, a41 percent price premium. Thesehigher prices make homeownershippossible for many high risk consumers.But, that considerable benefit must beweighed against the effects of takingmoney off the table for other types ofinvestments in getting ahead. A con-sumer may very well be better off, forinstance, investing that money andrepairing their credit scores beforegoing ahead and buying a higher-priced mortgage.
Together, these findings point to animportant opportunity leaders have tohelp families get ahead. Helping fam-ilies boost their creditworthiness incommunities with low credit scoreswould give more families that oppor-tunity to invest in assets, like houses.It also would likely lower the pricesthey pay for a number of basic neces-sities, including financial andinsurance products. In turn, thisfrees up money for savings, retail pur-chases, or other investments, likeeducation or retirement.
To do this, leaders will need toaddress the rising rate of delinquen-cies, which we discuss in the nextsection. By design, this has a system-atic effect on credit scores. But toboost credit scores, leaders also willneed to look more closely at the rea-sons why consumers are deemed to bea high risk by creditors, insurers,employers, utility companies, and anyof the various other institutions thatutilize these data.
E. Financial insecurity, primarilymeasured by the frequency ofloan delinquencies, rose between1999 and 2004 The status of payments made on out-standing loans is a critical variablefactored directly into a credit score. Ina nearly linear manner, consumercredit scores decrease in value when
delinquencies increase in frequency.This then has an influence on theprice for loans and insurance, andconsumer efforts to get jobs and rentapartments. Since information aboutloan delinquencies is collected bycredit bureaus, it provides a very richglimpse at the financial security ofAmerican across the country, alongwith very detailed information aboutthe performance of different types ofloans, including mortgages and revolv-ing lines of credit. The unmistakableimpression from these data is thatfinancial insecurity has increased inrecent years, though the South andparticular lines of credit are largelydriving this trend.
Overall Financial InsecurityThe ratio of borrowers who are latepaying their credit-bearing accountshas increased in recent years. In 1999about one out of every 23 borrowerswas over 60 or more days behind intheir payments. In 2004, that numberhad grown to about one out of every21 borrowers, about a 9 percentincrease.
By design, falling behind on pay-ments has a substantial impact onconsumer credit scores. To illustratethese effects, Figure 7 charts the rela-tionship between the proportion ofborrowers with delinquencies in acounty and the average credit score ina county. In a nearly linear manner,credit scores decrease in value as theincidence of delinquencies increasesin a county.36 This carries importantimplications for the price of manynecessities along with access to aneven larger number of necessities.
We consider these effects by analyz-ing either end of the delinquencydistribution, or counties with thehighest and lowest proportion of bor-rowers with a delinquent loan. At thelow end, this includes the countiesthat have a higher proportion of delin-quent borrowers than 90 percent ofthe other counties in the country, agroup we refer to as counties withvery high delinquency rates.37 At thehigh end, this includes the countiesthat have a lower proportion of delin-quent borrowers than 90 percent ofthe other counties in the countries, a
1 1MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Source: Author’s analysis of data in TrendUnion’s trend database
Note: All available data in the TrendUnion database was aggregated from depersonalized consumer
credit reports
Figure 7. By Design, Credit Scores are Strongly related to Delinquincies
20%
18%
16%
14%
12%
10%
8%
6%
4%
2%
0
525 550 575 600 625 650 675 700 725 750
Pro
po
rtio
n o
f P
ast-
Du
e B
orr
ower
s, b
y co
un
ty
Average Credit Score, by county
group we refer to as having very lowdelinquency rates.
In the counties with very high delin-quency rates, the average credit scorein 2004 was 597. In contrast, amongthe counties with very low delinquencyrates, the average credit score in 2004was 702. One of the important conse-quences of these differences is thehigher prices charged for many neces-sities when credit scores are low.When average county credit scores arelow the average price of many of thesenecessities is likely high. Capturing aportion of those extra costs in commu-nities with severe delinquencies couldcreate extra spending power amongconsumers in these areas. To theextent that these delinquencies arebehavioral, lowering delinquency ratesamong consumers might very well rep-resent a significant economic
development opportunity.38
Just as average credit scores widelyvary between U.S. counties, delin-quency rates, and the effects they holdfor consumers and local economies,also systematically vary across thecountry (Figure 8). Standing out againfrom the other regions of the country,Southern counties have a higher pro-portion of delinquent borrowers thanborrowers in other regions of thecountry. In 2004, nearly six out ofevery 100 borrowers in the typicalSouthern county had a delinquentloan, compared to about five borrow-ers out of every 100 in every otherregion of the country.
Even more telling, delinquencyrates have been rising in the Southfaster than in other regions of thecountry. Counties in the Southeaststates experienced the fastest recent
growth in consumer delinquency ratesout of any other census division,growing by nearly 23 percent between1999 and 2004. The next fastestgrowth rates during this time periodwere in Midwestern counties, whichsaw the average delinquency rate growby about 15 percent during thisperiod. But the average delinquencyrate is much lower in the Midwestthan the national average, suggestingthat there is a lower overall level offinancial insecurity, even if it isincreasing at a faster pace than therest of the country.
Meanwhile, counties in the Westactually saw nearly no change indelinquency rates between 1999 and2004, increasing by just about twopercent during this period—a trendlargely driven by counties in thePacific. Counties in the Northeast
12 MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Source: Author’s analysis of data in TransUnion’s trend database
Note: All available data in the trend database were aggregated from depersonalized consumer credit reports. Data are displayed by country and in quintiles;
past-due borrowers are 60+ days past due.
Figure 8. Southern Counties Have Higher Consumer Delinquency Rates Than Other Areasof the Country
Ratio of Past-Due Borrowers to All Borrowers (2004)Low (Less than 3.30%)
High (More than 6.96%)
area of the country also saw moremodest increases in delinquency ratesduring this period compared to thenational average, increasing by aboutsix percent.
These clear regional differences indelinquency rates illustrate an impor-tant, and widely overlooked, spatialdimension of financial insecurity inthe United States. While rates ofdelinquencies have increased nearlyeverywhere in recent years, counties inthe South stand out as having theswiftest increases in delinquency ratesin the country.
What explains these higher delin-quency rates in the South? It’scertainly true that the South tends tohave a much lower median incomethan other areas of the country, whichmay mean that consumers have lessmoney available to make paymentsthan consumers elsewhere. But hous-ing prices and other major costs ofliving tend to be much lower in theSouth than in other areas of the coun-try, which should depress thesignificance of the earnings gap. At thesame time, the unemployment rate inthe South is lower than in both theMidwest and the West. For these rea-sons, it’s not at all clear that basiceconomic differences between theregions explain the particularly highdelinquency rates in the South. Futureresearch will need to explore this vex-ing issue.
Home Loan InsecurityOverall measures of financial insecu-rity in a county paint a broad pictureof growing financial insecurity throughout the country. Still, this overallimpression masks the underliningfinancial commitments made by con-sumers to create this impression. Oneof the most important of these under-lining components is the growingpropensity of homeowners to fallbehind in their mortgage payments.
Throughout the 1990s, homeowner-ship surged across the country, addinga substantial asset to the lives of mil-lions of Americans. By 2004, about
seven out of every 10 householdsowned at least one home, up by abouteight percent from 1992. Not only isthis the highest homeownership raterecorded since the Census Bureaubegan tracking this statistic in 1965,the recent jump in homeownership isalso the largest sustained increase onrecord.39
But, the promise of homeownershiphas turned into too large of a financialburden for a rapidly growing numberof homeowners (Figure 9). This isreflected by the large number ofhomeowners that are delinquent ontheir mortgage payments. On thiscount, it is important to point out thatthere are many different sources ofdata on mortgage delinquencies, fromeach of the three credit bureaus, tosurveys of financial establishments likethat administered by the MortgageBankers Association, to surveys of bor-rowers like the Survey of ConsumerFinances. The estimates across all thesedifferent sources are often different, andit is still not clear why this is the case.Although our estimates are based onan enormous sample of consumers, itshould be treated as an estimate, justlike all of these other estimates.40
Our data indicate that about oneout every 50 homeowners with anactive mortgage was delinquent ontheir mortgage payments in 2004.That certainly does not mean thateach of these borrowers will lose theirhomes. But, because of the effectscredit scores have on several majornecessities, the costs of living for thesehomeowners will likely increase,which may make them more likely tolose their homes in the future.
Meanwhile, the rate that mortgagedelinquencies increased in recentyears has substantially outpacedincreases in overall delinquency rates.Between 1999 and 2004, the propor-tion of homeowners behind on theirmortgage payments jumped 115 per-cent, increasing from one out of every100 mortgage borrowers to about oneout of 50. This remarkable leap inmortgage delinquencies means that a
growing number of American home-owners are finding the costs ofhomeownership too financially bur-densome, throwing in jeopardy whatfor most families is their primary asset.But, as with the overall delinquencyrate, this trend is reflected unevenlyacross the country.
Once again, consumers in Southerncounties face a risk that is substan-tially larger than consumers who liveelsewhere in the country, even thoughthe housing stock in the South is theleast expensive in the country. Notonly are there higher proportions ofhomeowners with delinquent mort-gages in this region, their ranks havealso swelled at a clip unmatched bynearly any other area of the country.Leading the pack, counties in theAlabama, Kentucky, Mississippi, andTennessee Southeast division saw theaverage mortgage delinquency rate ina county nearly triple between 1999and 2004, including about one ofevery 32 mortgage borrowers by 2004.
In stark contrast, the average West-ern Pacific county actually saw only a14 percent increase in the number ofdelinquent mortgages during that timeperiod. So, while most counties saw asurge in the number of borrowersunable to pay their mortgages on time,there are clearly regions of the countrythat have emerged as leaders and lag-gards in this phenomenon. Althoughhome prices have generally surged wellahead of income increases in recentyears, it is actually where homes areleast expensive that we find the high-est incidences of homeowners fallingbehind in payments.
Revolving Debt InsecurityMuch has recently been written aboutthe surge in revolving debt, includingdebt held in credit cards and homeequity loans.41 This is certainly true.According to our data, the total realvalue of revolving debt held per bor-rower was valued at about $8,900 in2004, a 46 increase from 1992.42
But, the rate of delinquencies onrevolving loan accounts has actually
13MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
decreased during this period. Accord-ing to TransUnion’s trend database, anaverage of one out of every 40 borrow-ers with a revolving line of credit wasdelinquent on their payments in 2004,a 10 percent decrease from 1999.Once again, though, these trendsplayed out unevenly across the country(Figure 8).
Proportions of borrowers withrevolving lines of credit in the Pacificdivision of the country saw thesharpest drops in the rates of delin-quencies between 1999 and 2004,falling by about 19 percent. On theother hand, delinquency rates in fourCensus divisions—New England,South Atlantic, West North Central,and the Southeast—saw delinquencyrates on revolving lines of credit fall bytwo percent or less during this period.Meanwhile, the counties in the Moun-
tain division of the country have thehighest rates of delinquency rates onrevolving loans in the country, includ-ing more than one out of every 33borrowers with this line of credit.Right behind these counties, delin-quency rates on revolving lines ofcredit include about one out of every31 borrowers in the South Atlanticcounties that carry this form of debt.
This evidence again speaks to thespatial implications of these nationaltrends. The South, home to some ofthe least expensive places to live in thecountry, also is home to some of thehighest proportions of borrowers withrevolving debt who can not meet theirpayments on time. Because the effectthese late payments have on creditscores, consumers in these areas willend up paying higher prices for a num-ber of major necessities.
So, even while the proportion ofdelinquent borrowers with revolvinglines of credit remain generally lessthan found in the mortgage market,many more people own revolving linesof credit than mortgages, meaning thatthe insecurity associated with theselate payments, including the effects oncredit scores, affects many more peo-ple. At the same time, the smallervalue of payments typically made tomaintain revolving lines of credit maymean that borrowers have an easiertime meeting these payments thanmuch higher mortgage payments.
14 MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
Source: Author’s analysis of data in TransUnion’s trend database
Notes: All available data in the trend database were aggregated from depersonalized consumer credit reports. Data are displayed by country and in quintiles;
delinquent mortgages are 60+ days past due.
Figure 9. Southern Counties Have Higher Mortgage Delinquency Rates Than Other Areas of the Country
Ratio of Past-Due Mortgage Borrowers to all Mortgage Borrowers (2004)Low (Less than 1.30%)
High (More than 3.43%)
Research and Policy Implications
Credit reports and scoresaffect both the access andterms of access to a largeand growing array of basic
necessities, including jobs, housing,energy, and communications. Besidesformally separating issues of class,race, and gender from underwritingdecisions, information in credit reportsis also used to substantially expand therange of market products available toconsumers. Among other benefits, thishas increased access to assets likehouses, given consumers more choicesabout market products, and spurredeconomic development in neighbor-hoods once ignored by creditors andinsurance companies. At the sametime, consumer credit reports andscores have spurred a dramatic surgein the availability of consumer debt,which has improved the lifestyle ofcountless Americans.43
With these benefits, however, therehave come some costs. Consumers areconfronted with an array of choicesthat they do not fully understand.Applications of these products to mar-ket decisions are not well researched.And, consumers have amassed a sub-stantial amount of debt. As onepossible outcome, we found the delin-quencies across a range of differentlines of credit substantially increasedin recent years. In turn, growing finan-cial insecurity drives down creditscores, which affects the access andterms of access to a broad array ofbasic necessities and undermineshousehold goals for savings and wealthaccumulation.
Many of these trends vary in sys-tematic ways across the country, evendividing the country into clear geo-graphical areas that lead or lag thenation. Most strikingly, counties in theSouth are sharply distinguished fromthe rest of the country by the highrates of delinquency in the typicalcounty, which drives the high propor-tion of consumers who have extremely
low credit scores. We also found that some of the fac-
tors that credit scores have replacedfrom underwriting decisions—likerace—are strongly associated withcredit scores. Some of this may haveto do with the fact that the depth of anindividual’s knowledge about creditscores, and the significance of thesescores in a family’s financial life, isstrongly related to many socioeco-nomic characteristics.44
What implications do these findingsraise for policymakers? Most impor-tantly, leaders must take moreseriously the complexity of participat-ing in the modern Americanmarketplace. The need for a sustainedfinancial education has never beengreater. Policy and business leadersalso need to continue asking if theseassessments of consumers are alwaysaccurate, while also assessing if themany market responses to these riskassessments are reasonable. Weexpand on all of these implicationsbelow:
Provide a Financial Education forConsumersAlthough credit reports and scoresplay a fundamental role in the finan-cial life of individuals, few peopleunderstand that role. How many peo-ple know, for instance, that creditreports affect the access and terms ofaccess to financial services, insurance,telecommunications, apartments, andeven jobs?
In fact, a recent survey by ProvidianFinancial and the Consumer Federa-tion of America indicates that fewpeople understand credit scores, oneof the primary uses of credit reports.45
Although 93 percent of consumersindicated that they knew credit scoresare affected by missing payments, only27 percent knew what credit scoresactually measure. Moreover, thescarce knowledge about the impor-tance of credit scores that does existsystematically varies with an individ-ual’s income and educationalattainment.46 Only 56 percent of the
respondents with a low educationalattainment, and 64 percent of respon-dents with a low income, indicatedthat they knew that their credit ratingweakened when they missed a creditcard payment.
Even while there is limited publicunderstanding of credit reports andscores, there has never been moreinformation about both. Hundreds ofcredit counseling agencies now pro-vide information about credit reportsand scores, consumer finance TV per-sonalities constantly advertiseinformation, and there are thousandsof documents posted on the Internetabout both.47 Lack of information,then, is not driving this low publicunderstanding.
Instead, there is likely a low under-standing of credit reports, and thesignificance they play in people’s lives,because people are not accessing thisinformation. Unless an individual is infinancial trouble, there are no routineways that people are exposed to creditreports, and their many applications.This absolutely vital ingredient to aperson’s ability to get ahead is reallyonly recognized after someone is infinancial trouble. This makes creditreports both an asset that goes unuti-lized, while at the same serving as anunnecessary roadblock for others.
The solution to building awarenessand knowledge of credit reports lies,then, in promoting routine ways thatgive consumers access to this informa-tion. For this reason, leaders shouldimplement financial education forconsumers. This can be incorporatedinto existing curriculum at the K-12level. Banks can integrate basic infor-mation about credit reports when theysign their clients up for service. Theycould also make access to a variety ofbudgeting tools part of their customerservice operations. Consumer organi-zations could provide feedbackinformation about credit counselingagencies in local markets, and investin web-based financial educationcourses. Insurance companies canshare with customers the impact their
15MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
scores are having on their rates; insur-ance regulators could provideinformation about the different waysthat companies they regulate use thisinformation.48 And, utility andtelecommunications companies caninform their customers about the rolescores play in their access to services.Through all of these ways, consumerscan receive the information they needto become proficient in negotiatingthrough this tool, or barrier, to gettingahead.
Pass a Credit Bureau Disclosure ActPolicymakers should pass a CreditBureau Disclosure Act (CBDA) thatrequires the credit bureaus, rather than policymakers, to report theirassessments of the accuracy of theirinformation. The bureaus already regularly assess the accuracy of theirmodel predictions because their business depends on it. But, thesesassessments are not shared with poli-cymakers, or the public.
Is this necessary? Various independ-ent researchers have tried to assess theaccuracy of the information in thereports, along with the accuracy of therisk estimates. But sample sizes areoften small, time bound, and based oninformation from a single bureau. Theresult: there is a lot of conflicting evi-dence and uncertainty about theaccuracy of these market products.This fuels ad hominem opinions aboutthe accuracy of these products, whichis unfair to the bureaus, the busi-nesses that rely on their data, andconsumers.
To address just this type of uncer-tainty, the Fair and Accurate CreditTransactions Act of 2003 (FACT Act)required the bureaus to make one freecredit report available to consumersevery year, which gives consumers theability to more easily assess the accu-racy of the information contained inthe report.49 But, the law did notrequire the bureaus to report a) thenumber of queries they receive or b)information about the outcome ofthose queries. The law also did not
give consumers the option to receive afree credit score estimate from thebureaus or MyFICO, which makes itmore difficult to link reports withscore predictions. All of this meansthat the FACT act made some impor-tant steps forward, but more needs tobe done. For those households whohave not received news of this law, notacted on it, or cannot afford to buytheir credit score, it is uncertain howtheir interest is being served by thisnew policy.
To address this limitation, a CBDAshould require that bureaus annuallysubmit reports to federal policymakersthat speak to the accuracy of theirinformation. Most importantly,bureaus should report the results ofconsumer inquires they receive intothe accuracy of their information.Such an act would preserve the rightsof these businesses to maintain theirintellectual property, while also safe-guarding the privacy of consumers.Just as important, it would create anincentive for a long overdue assess-ment of how the accuracy of bothcredit scores, and the reports they arebased on, can be improved.
Research market responses to creditreport informationThis report has illustrated how pricesrespond to changes in credit scoresthrough an analysis of pricing deci-sions made in the mortgage market.When credit scores are low, mortgageborrowers pay more—in our example,thousands of dollars more every year—for their loan than borrowers who havehigh credit scores. But, this examplereally just scratches the surface of howmarkets now respond to credit reportsand scores. We also know that pricesfor and access to other loans, likehome equity or auto loans, areaffected by credit scores. So are pricesand access to auto and home insur-ance, along with access to jobs andapartments.
Oftentimes, credit scores have sub-stantially expanded the opportunity ofhigh-risk households to access credit,
as we have discussed. There also mayhave been beneficial effects associatedwith non-traditional applications, likeinsurance pricing and landlord deci-sion-making.
Still, the rising delinquency ratesmake it important to now ask if creditreports and scores are always beingused in a way that is beneficial to allconsumers. What costs, in otherwords, are created along with the sub-stantial and widely-recognized benefitsassociated with these products, andhow are those costs distributed acrossboth place and people?
To start with, there may be an overreliance on credit scores if other infor-mation that speaks to the true risk aperson represents is not considered. AsPeter McCorkell of Fair Isaac andCompany has noted, “ignoring otherrelevant information in the mortgagedecision process is not in the bestinterests of either borrowers orlenders….[But] during the mortgageand refinancing boom there was cer-tainly an economic motivation to moveon to the easy cases rather than spendextra time on the difficult ones.”50 Inthis case, credit scores are being reliedon too heavily because they are notthe only factor that should be used toassess risk.
Similarly, there is no rule or even ageneral guide that specifies what aproper market response in any of thesecases should be to credit scores. Thereis no fixed cap, for instance, that sug-gests a broker cannot systematicallyadd 2, 5, or even 10 percent to thetotal value of a loan in fees and higherinterest rates when lending to some-one with a low credit score. Aside fromconcerns addressed in the EqualCredit Opportunity Act, no govern-ment policy specifies what is fair andwhat crosses the line.
This raises a key, albeit very compli-cated, question for policymakers:What is the optimal level of price fluc-tuation across different levels of risk?In other words, at what point dohigher prices for borrowers with lowcredit scores stop covering the costs of
16 MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
the predicted higher level of risk, andstart becoming price-gouging marketproducts?
In theory, an invisible cap exists inthe market that curbs the extent towhich prices vary across and withinmarkets. Competition between busi-nesses, in this ideal world, drives thiscap to its most efficient point. But, wedon’t live in an ideal world. Unfortu-nately, it is currently difficult to knowwhat an appropriate cap would looklike, or whether a cap is even a properpolicy response, without more informa-tion. We know that markets areresponding to these market products,but we don’t always have a very goodsense of how they respond, particu-larly in non-traditional applications ofthese products. For this reason, policy-makers need to begin systematicallygathering and evaluating informationrelated to market responses to creditreports and scores. Public policy ispremature; research is not.
As a model, federal policymakersshould consider the results from theIllinois pilot predatory lending data-base, created in July 2005 (Public ActNo. 94-280). This law authorizes theIllinois Department of Financial andProfessional Regulation to develop andmaintain a database on mortgageproducts sold in Cook County over thenext four years. Consumer creditscores, types of mortgage products,and the price of those products areamong the many variables that will beincluded in this database. This will bethe first public database in the countryto systematically collect the full of rangeof information analyzed by lenderswhen making access and pricing deci-sions.51 Among its many attributes,these data will provide policymakerswith the capacity to assess the extentto which mortgage prices respond tochanges in credit scores. Given the farreach of credit reports and scores, itseems quite reasonable to collect simi-lar data to study other applications ofcredit reports and scores.
Endnotes
1. Robert Hunt, “A Century of Credit Reporting in
America.” (Federal Reserve Bank of Philadelphia,
Working Paper No 05-13, 2005); and U.S. Census
Bureau, County Business Patterns.
2. Credit reports preceded credit scores. Reports were
originally gathered by local or regional bureaus to
track the performance of loans made by individual
creditors, from banks to retail establishments. Then
in 1958, Bill Fair and Earl Isaac developed a model
for generating credit scorecards, which, by the end
of the 1970s, were used by most banks. The draw-
back of these early scores, however, was that they
were often not comparable across either industries
or businesses within industries, because scorecards
were custom developed by bureaus for different
clients. In fact, it was really not until a 1995 rec-
ommendation by Fannie Mae and Freddie Mac that
mortgage lenders use FICO scores in their mort-
gage lending decisions that the use of credit scores
began to explode in this country, now influencing
billions of decisions every year. For more informa-
tion about the evolution of this market product, see
Lyn C. Thomas, “A Survey of Credit and Behavioral
Scoring: Forecasting Financial Risk of Lending to
Consumers.” International Journal of Forecasting,
16 (2000):149–72; and Hollis Fishelson-Holstine,
“Credit Scoring’s Role in Increasing Homeowner-
ship for Underserved Populations” in Nicolas P.
Retsinas and Eric S. Belsky, Building Assets, Build-
ing Credit (Cambridge: Joint Center for Housing
Studies and Washington: Brookings Institution
Press, 2005).
3. A) The permissible uses of credit reports were iden-
tified in the Fair Credit Reporting Act (1970) to
protect the privacy of individuals. However, the
range of legal uses is quite broad. Among the many
legal applications, credit reports can be used for
any purpose related to a credit transaction (includ-
ing marketing related products), underwriting
insurance, evaluating credit risks, and for employ-
ment purposes. New state laws/regulations
somewhat curb these applicable uses, although
these laws are uneven across the country. Also, the
1996 amendments to FCRA somewhat curb the use
of reports, by allowing consumers to sue when their
report is used for a purpose not permitted by
FCRA. At the same time, the Fair and Accurate
Credit Transactions Act (FACT), passed in Decem-
ber 2003, preempted the states from passing
certain laws that curb the content and permissible
uses of credit reports. B) There are numerous sur-
veys that suggest employers are more frequently
using credit reports to screen job applicants. A
2004 survey of companies by the Society for
Human Resource Management, for instance, found
that 35 percent of companies used credit reports
for this purpose, up from 19 percent in 1996. This
closely paralleled findings in the 2003, University
of Florida, National Retail Security Survey. C)
There are private companies which provide credit-
checking services for landlords. For instance, see
http://www.amerusa-tenant-screening.com/.
There is no evidence that we have been able to find
that attests to how many renter applications are
accepted or rejected because of their credit scores,
although the existence of nationwide companies to
provide this information suggests there is a large
market for this information. D) On the basis of
background conversations with state regulators, we
know that it is commonly believed that a growing
number of utility companies are using credit report
information. But, the market impact of these appli-
cations is essentially unknown.
4. An increasing number of insurance companies use
“insurance scores”, which are constructed from
information in credit reports. Insurance scores are
used to predict the likelihood of a future adverse
event, like an insurance claim. State policymakers
are increasingly acting to curb the use of credit
reports by insurance companies. Hawaii bans the
use of credit scores, and all states regulate its use. A
growing number of statehouses are also considering
legislation to limit the legal uses of credit scores in
insurance pricing, such as justifying an increase in
an insurance premium. In this case, state authority
to regulate insurance is in tension with provisions
in FACT that preempt state limits on credit report
uses.
5. In particular, the “high risk” market was mostly
ignored prior to the invention of credit scores,
which are based on information contained in credit
reports. The prediction power that these scores
have given companies substantially expanded their
ability to extend credit. For more information,
please refer to Nicolas P. Retsinas and Eric S. Bel-
sky,. Building Assets, Building Credit: Creating
Wealth in Low-Income Communities (Cambridge
and Washington: The Joint Center on Housing and
The Brookings Institution, 2005); and Thomas A.
Durkin and Michael E. Staten,. The Impact of Pub-
lic Policy on Consumer Credit (Boston: Kluwer
Academic Publishers, 2004).
6. Mark Furletti, “An Overview and History of Credit
Reporting,” Federal Reserve Bank of Philadelphia,
TransUnion Workshop (held April 5, 2002).
7. The Bankruptcy Abuse Prevention and Consumer
Protection Act of 2005 (signed into law in April
2005), may very well continue to reduce the num-
ber of personal bankruptcies (at the time of
publication, bankruptcy filings have substantially
dropped). Among its several major provisions, the
law requires credit counseling for filers and makes
17MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
it more difficult for consumers to file for Chapter 7
bankruptcies, which accounted for about 74 per-
cent of all personal filings in 2004. Chapter 7
bankruptcy filings are more popular than Chapter
13 filings because outstanding debt is cancelled
after non-exempt assets are sold-off. On the other
hand, Chapter 13 filings require filers to establish a
five-year pre-payment plan.
8. Teresa Sullivan, Elizabeth Warren, and Jay
Lawrence Westbrook, The Fragile Middle Class:
Americans in Debt (New Haven: Yale University
Press, 2000).
9. See for instance Timothy Egan, “Newly Bankrupt
Raking in Piles of Credit Offers,” New York Times,
December 11, 2005, p. 1.
10. “It is estimated that more than 70 million Ameri-
cans make rent, mortgage, and other recurring bill
payments that are not reported to traditional credit
bureaus. As a result these consumers often have
lower credit scores than they should, and pay more
for housing, credit, and insurance than they
deserve.” Terry Clemans, Executive Director of
National Credit Reporting Association, Inc., quoted
in an October 3, 2005 Press Release by NCRA, Inc.
11. Information Policy Institute, “Giving Underserved
Consumers Better Access to the Credit System: The
Promise of Non-Traditional Data” (2005).
(http://www.infopolicy.org).
12. For some examples see: Consumer Federation of
America and National Credit Reporting Associa-
tion, “Credit Score Accuracy and Implications for
Consumers.” (Washington: Consumer Federation of
America, 2002); Michael E. Staton and Fred H.
Cate, “Accuracy in Credit Reporting,” in Nicolas P.
Retsinas and Eric S. Belsky (2005); Robert B.
Avery, Paul S. Calem, and Glenn B. Canner (2004);
Alison Cassady and Edmund Mierzwinski, “Mis-
takes Do Happen: A Look at Errors in Consumer
Credit Reports.” (Washington: National Association
of State PIRGs, 2004).
13. For evidence of consumers being under-informed,
see the annual survey administered by Providian
Financial and the Consumer Federation of America
or GAO, “Credit Reporting Literacy: Consumers
Understood the Basics but Could Benefit from Tar-
geted Educational Efforts.” (GAO-05-223, 2005).
Systematic overcharging would likely lead to
adverse selection among clients if it were applied
across the board. More likely, evidence of informa-
tion asymmetries among consumers may lead to
systematic effects that substantively vary across
consumer segments.
14. For more information about these data, please refer
to John M. Barron, Gregory Elliehausen, and
Michael E. Staten, “Monitoring the Household
Sector with Aggregate Credit Bureau Data.” Busi-
ness Economics (35) (2000): 63–76.
15. Consumer credit scores do vary across bureaus,
although the extent to which they vary is hardly cer-
tain. There is no evidence that aggregate level
data—the type of data we analyze here—signifi-
cantly varies from place to place. For this reason,
we don’t have a reason to suspect that reliance on
the evaluations of one bureau for county level sta-
tistics should systematically bias our analysis.
16. Barron, Elliehausen, and Staten. 2000. .
17. The Fair and Accurate Credit Transaction Act of
2003 mandated that the Federal Reserve and the
Federal Trade Commission explore these relation-
ships and report back to Congress. The first of
those reports was due in December 2005, but, at
the time of publication, that report had not yet
been released.
18. The Brookings Institution’s Urban Market Initiative
hosted a roundtable on alternative data sources in
credit scoring, and has made available most of the
presentations on their webpage: http://www.brook-
ings.edu/metro/umi/20051215_paidroundtable.h
tm.
19. The importance of these different categories in the
calculation of a credit score varies, but myFICO
reports a weight distribution for the general popula-
tion: 35 percent payment history, 30 percent
amount owed, 15 percent length of credit history,
10 percent types of credit used, and 10 percent new
credit [www.myfico.com, accessed September
2005]. Debt-to-equity ratios, for instance, are not
used in credit card lending, and are not often
included in a score for mortgage lending (but as a
separate consideration).
20. For instance, myFICO reports the general weights
assigned to classes of variables on its webpage. The
same information is not available for other credit
scores, such as those calculated by the three major
bureaus or by insurance companies.
21. For more information, please refer to the web page
of the Population Division of the U.S. Census,
http://www.census.gov/popest/counties.
22. When fully implemented, the American Commu-
nity Survey (ACS) will become a superior source of
information.
23. The same distribution is reported by myFICO. See:
MyFICO. “Understanding your Credit Score,”
available at www.myfico.com/Offers/
myFICO_UYCS%20booklet.pdf.
24. Current rates are August 2005. The range of con-
sumer credit scores in the private bureau data used
in this report is nearly identical to the Fair Isaac
credit score, or the FICO score. Fair Isaac updates
a daily, national estimate of the relationship
between its credit scores and the average interest
rate charged by mortgage lenders for several major
mortgage products. All additional analyses in this
paper related to this relationship rely on these data
(www.myfico.com).
25. MyFICO homepage [www.myfico.com]
26. Although there is no hard set rule, it is frequently
reported that scores above 650 will qualify a bor-
rower for a prime mortgage rate. Consumers with
scores above 840 may not be offered different terms
than consumers with scores of 800. We select this
cutoff point only because TransUnion’s trend data-
base has this value as a cutoff point.
27. To determine whether the selected time period
affected this analysis, we also looked at the percent-
age change between 1999 and 2003, 1999 and
2002, and 1999 and 2001. All of these analyses
confirm the trend illustrated in Figure 5.
28. Upcoming analyses by the Federal Reserve will exam-
ine panel data, which will yield insight into how this
trend plays out among individual borrowers.
29. Dean S. Karlan and Jonathan Zinman, “Observing
Unobservables: Identifying Information Asymme-
tries with a Consumer Credit Field Experiment.”
(Dartmouth University, 2005).
30. Concerns about unequal access to credit across
racial and income groups led to the creation of the
Home Mortgage Disclosure Act (HMDA) and, then,
the Community Reinvestment Act (CRA). For a
recent summary of HMDA see Robert B. Avery,
Glenn B. Canner and Robert E. Cook,. “New Infor-
mation Reported under HMDA and Its Application
in Fair Lending Enforcement.” Federal Reserve Bul-
letin, Summer (2005): 354–394. For a recent
assessment of the CRA see Michael Barr, “Credit
Where It Counts: Maintaining a Strong Commu-
nity Reinvestment Act” (Washington: Brookings
Institution, 2005).
31. We did run a regression analysis that predicted
credit scores as a function of a variety of informa-
tion found in credit reports along with county
socioeconomic information. We found that there
was a significant, direct relationship between credit
scores and median income, proportion black, pro-
portion Latino, and educational attainment. But, an
analysis of the robustness of these results suggested
that the variables included in a credit report have a
much more powerful effect on credit scores than
socioeconomic characteristics. Still, we do not
report these results, and only note them here,
because the information available from the reports
was not sufficient for the model to fully control for
information accounted for in a credit score. Future
analyses, based on all of this information, are still
needed to sort out the direct, indirect, and total
relationship between credit scores and socioeco-
nomic variables. It is for this reason that we only
report associational relationships and are very
explicit that our research does not point to causal
claims in this case.
18 MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
32. See annual survey administered by Providian
Financial and the Consumer Federation of Amer-
ica; or GAO, “Credit Reporting Literacy:
Consumers Understood the Basics but Could Bene-
fit from Targeted Educational Efforts” GAO-05-223
(2005).
33. This premise that more information leads to better
scores was one of the motivations for passing the
Fair and Accurate Credit Transactions Act of 2003
(FACT Act). This required the bureaus to make one
free credit report available to consumers every year,
which gives consumers the ability to more easily
assess the accuracy of the information contained in
the report.
34. Robert B. Avery, Paul S. Calem, and Glenn B. Can-
ner, “Credit Information Reporting and the
Practical Implications of Inaccurate or Missing
Information in Underwriting Decisions,” BABC 04-
11 (Harvard University: Joint Center for Housing
Studies, 2004).
35. The average credit score of a county is not used to
assess the costs of credit. Rather, the higher propor-
tion of consumers with high or low credit adds up
to systematic differences across counties.
36. Credit scores are generally used by businesses to
assess risk within markets, and the range within a
market likely differs from a national range. Still,
consumer credit scores are generated from a com-
mon set of criteria, which means that the depth of
risk posed in a particular market by a score should
only marginally vary across markets. At the same, it
is important to note that there are many different
types of credit scores in use in the market today.
The three major national bureaus, for instance,
each generate its own credit score and a FICO
score, which is directly comparable across the three
major bureaus. Although these bureaus account for
over half of the market, there are over 1,000 differ-
ent credit bureaus, many of whom generated some
type of predictive score.
37. The overall delinquency rate in a county is the pro-
portion of borrowers in a county who have at least
one loan that is 60 days or more past due. Mortgage
delinquency rates are the proportion of borrowers
with at least one mortgage 60 days or more past
due; revolving loans are the proportion of borrowers
with at least one revolving loan 60 days or more
past due. Delinquency is measured by bureaus as
any account that is past due, but the data are only
available for accounts 60 days or more past due.
For this reason, our estimates of delinquency
should be interpreted as very conservative esti-
mates.
38. Delinquencies can also be brought on by unex-
pected events, such as an unexpected health costs
or loss of a job.
39. These statistics come from the Current Population
Survey, available at
http://www.census.gov/hhes/www/housing/hvs/his-
toric/index.html (August, 2005).
40. The trend data is a sample of borrowers, which
includes both solely owned and joint accounts. For
this reason, it may be the case that some estimates
are biased in types of lines where joint accounts are
particularly prevalent. The extent of this bias
remains to be seen.
41. For instance, see Stan Sienkiewicz,. “Credit Cards
and Payment Efficiency” (Federal Reserve Bank of
Philadelphia, Payment Cards Center Discussion
Paper, 2001); and Tamara Draut and Javier Silva,.
“Borrowing to Make Ends Meet: The Growth of
Credit Card Debt in the ‘90s” (Washington::
Demos, 2003).
42. Do keep in mind that this is the average among all
consumers with revolving debt. This means that the
effect of home equity loans on this average is
weighted down by the much larger number of con-
sumers with credit cards and retail cards, which
tend to have lower balances than home equity
loans.
43. But, see: Gregg Easterbrook,. The Progress Paradox:
How Life Gets Better While People Feel Worse (New
York: Random House, 2003)..
44. See annual survey administered by Providian
Financial and the Consumer Federation of Amer-
ica; or GAO,”Credit Reporting Literacy: Consumers
Understood the Basics but Could Benefit from Tar-
geted Educational Efforts.” (GAO-05-223, 2005)..
45. For a summary of this survey please refer to:
www.consumerfed.org/pdfs/Providian_Press_Rel
ease_9_05.pdf (October 2005). Also, see the GAO
report mentioned in notes 44 and 46.
46. This finding was also reported in a recent analysis
of credit report literacy by the Government
Accountability Office, GAO-05-411SP, which was
mandated in the 2003 FACT Act.
47. Robert M. Hunt,”Whither Consumer Credit Coun-
seling?” Business Review, Quarter 4 (2005).
48. According to the Michigan Department of Insur-
ance, some companies price insurance by using
multiple “insurance score” categories, whereas
other companies used just two categories and oth-
ers do not use credit information. Drivers may be
able to find a better price for insurance if they had
this information.
49. To access your free credit report, go to
https://www.annualcreditreport.com/cra/
index.jsp
50. Peter L. McCorkell, “The Impact of Credit Scoring
and Automated Underwriting on Credit Availabil-
ity,” in Thomas A. Durkin and Michael E. Staten,
eds, The Impact of Public Policy on Consumer
Credit( Boston: Kluwer Academic
Publishers, 2004).
51. The 2004 version of the Home Mortgage Disclo-
sure Act data took a big step forward with the
inclusion of price information. But, it still does not
have information about credit scores. To be sure,
this is a difficult undertaking, since so many differ-
ent credit scores exist. The Illinois effort here may
very well provide a model for federal policymakers
as they determine how to include these data.
19MAY 2006 • THE BROOKINGS INSTITUTION • SURVEY SERIES
The Brookings Institution
1775 Massachusetts Avenue, NW • Washington D.C. 20036-2188Tel: 202-797-6000 • Fax: 202-797-6004
www.brookings.edu
Direct: 202-797-6139 • Fax/direct: 202-797-2965
www.brookings.edu/metro
For More Information:Matt Fellowes
Senior Research AssociateBrookings Institution Metropolitan Policy Program
For General InformationThe Brookings Institution Metropolitan Policy Program
(202)797-6139www.brookings.edu/metro
To receive weekly updates of Metro Program products, please subscribe athttp://www.brookings.edu/metro/metro.htm
AcknowledgmentsThe Brookings Institution Metropolitan Policy Program thanks the Annie E.Casey Foundation for its generous support of our research on programs andpolicies that support lower-income working families, and the Fannie MaeFoundation, the John D. and Catherine T. MacArthur Foundation, theGeorge Gund Foundation, and the Heinz Endowments for their general sup-port of the program.
The author is grateful to Evan DeCorte and Naotaka Sugawara for theirresearch assistance. He would also like to thank a number of people whoprovided comments on earlier versions of this paper, including Ezra Beckerof TransUnion, Alan Berube of Brookings, Glenn B. Canner of the FederalReserve Board, Robert M. Hunt of the Federal Reserve Bank of Philadelphia,Bruce Katz of Brookings, Amy Liu of Brookings, Clifton O’Neal of Tran-sUnion, Sally Park of the National Foundation for Credit Counseling, PariSabety of Brookings, and Michael Turner of the Information Policy Institute.All errors remain the sole responsibility of the author.