Banking Antitrust: Are the Assumptions Still Valid? · 2019. 10. 9. · As such, credit unions do...

Banking Antitrust: Are the Assumptions Still Valid?R. Alton Gilbert and Adam M. Zaretsky

NOVEMBER/DECEMBER 2003 29

T he federal bank regulatory agencies and theU.S. Department of Justice (DOJ) scrutinizebank mergers and acquisitions for potential

antitrust violations. To perform this antitrustanalysis, the federal regulators make assumptionsabout the geographic scope of banking markets,the types of competitors that banks face in thesemarket areas, and the nature of banking services.The authorities assume that the relevant geographicmarket is a local area where banks compete to offerfinancial services to households and small busi-nesses. That market area is often approximated bya metropolitan area for mergers involving banksin urban areas and by a county for those involv-ing banks in rural areas. The antitrust authoritiesassume that the relevant competitors are bankswith offices in the same market area. They furtherassume that the relevant product for antitrust analy-sis is a cluster of financial services that is uniqueto banking. In some analyses, however, the focus ison competition among banks to provide individualcategories of deposit and loan services. Antitrustagencies typically use a bank’s deposits as themeasure of output of financial services each bankprovides.

The assumptions that underlie banking antitrusthave been subject to criticism in recent years (Austinand Bernard, 2001; Jackson and Eisenbeis, 1997;Moore, 1998; Petersen and Rajan, 2002; Radecki,1998, 2000; and Santomero, 1999). Some criticsfocus on assumptions about the relevance of localmarkets for antitrust analysis. They argue that finan-cial innovation and changes in banking regulations,including nationwide branch banking since 1997,have undermined the relevance of using local areasfor competitive analysis. Innovative financial firmsare now able to offer services, such as loans andinvestment options, to customers in areas wherethe firms do not have offices. In addition, the threat

of entry by out-of-market financial firms constrainsthe terms under which banks with offices in a givengeographic area can offer services to local customers.Finally, studies indicate that banks with offices inmany communities tend to offer financial servicesto all communities on the same terms. The resultsof these studies appear to undermine the assump-tion that the terms on which banks make their ser-vices available to customers depend to some extenton the structure of local market areas. Shull andHanweck (2001) also criticize the focus on localmarkets in banking antitrust analysis, arguing thatit is not constraining consolidation of the bankingindustry at the national level.

Critics also focus on the assumption that acluster of banking services is the relevant productin antitrust analysis. They argue that the success ofnonbank financial firms in providing services tohouseholds and small businesses has underminedthe premises that commercial banks are the relevantcompetitors in antitrust analysis and that a bank’srelevant product is a “cluster of banking services.”1

The large literature on the topic of bankingantitrust dates from the 1960s, when bank mergersin the United States became subject to the federalantitrust statutes. This article, by summarizing theresults of empirical studies written or publishedsince the early 1990s, assesses whether these morerecent studies provide empirical support for thecurrent assumptions that underlie banking antitrustanalysis.2

CURRENT METHOD OF ANTITRUSTANALYSIS IN THE BANKING INDUSTRY

Antitrust analysis of bank mergers and acquisi-tions dates back to 1963, when the U.S. SupremeCourt held that commercial banking, like other

1 See DeYoung, Hunter, and Udell (2003) for a description of thesechanges in the market for financial services to households and smallbusinesses.

2 For surveys of the earlier studies, see Rhoades (1982), Gilbert (1984),and Weiss (1989). See Rhoades (1996) for an examination of theimplications of empirical studies for the methods of banking antitrust.

R. Alton Gilbert is a vice president and banking advisor and AdamZaretsky is an economist at the Federal Reserve Bank of St. Louis.Neil R. Wiggins provided research assistance.

© 2003, The Federal Reserve Bank of St. Louis.

industries in the United States, is subject to theSherman Antitrust Act of 1890 and the Clayton Actof 1914.3 In its opinion, the Court noted that thetest for anticompetitive behavior is whether theeffect of a bank merger “may be substantially tolessen competition…in any line of commerce in anysection of the country.” To apply this test, the Courtdefined the “line of commerce” for the bankingindustry as the cluster of products and services—demand deposits, trust administration, and exten-sion of various types of credit, for example—thatbanks uniquely provide to their customers. In otherwords, the Court determined that the products andservices denoted by the term “commercial banking”compose a distinct line of commerce.

To define “section of the country”—that is, therelevant geographical market—the Court looked towhere the effect of a merger on competition wouldbe “direct and immediate.” For banking, this effectoccurs in the customers’ local communities becauseindividuals and firms typically conduct the bulk oftheir banking transactions at banks with local offices.

These two definitions—the relevant line ofcommerce is a cluster of products and servicesuniquely supplied by commercial banks, and therelevant geographical market is local—have guidedbanking antitrust analysis since the 1963 ruling. Thefederal banking regulators (Office of the Comptrollerof Currency, Federal Deposit Insurance Corporation,Office of Thrift Supervision, and Board of Governorsof the Federal Reserve System) have since adoptedthese definitions for their antitrust analyses.

Once a bank regulatory agency has identifiedthe cluster of products and services and the localmarket, its final step is to determine whether theeffect of the merger “may be substantially to lessencompetition.” In its ruling, the Supreme Court recog-nized that the answer to this question involved notonly the immediate effects of a merger on competi-tion, but also its anticipated future effects.4 Such aprediction relies on the structure of the relevantmarket—that is, market concentration, the marketshares of individual banks, and the number ofmarket competitors. Banking antitrust is based onthe assumption that the structure of a market influ-ences how firms in that market will act, which, in

turn, affects the firms’ overall performance.5 Inother words, the merger’s effect on these measuresof “structure,” particularly market concentration,is thought to be a reliable gauge of whether themerger will substantially lessen competition. There-fore, a proposed merger that increases market con-centration considerably would likely fail this test,and the federal regulator would not approve it.The federal regulator might approve it, however, ifother evidence exists to mitigate the proposal’s anti-competitive effects on market structure. That said,the DOJ could challenge the decision and possiblysue to prevent the merger.

To minimize the chances that a decision will bechallenged and to align the antitrust analyses of thefederal regulators, the DOJ has periodically issuedguidelines that define the circumstances underwhich an application is likely to exceed its antitruststandards and, therefore, warrant closer scrutiny.The federal banking regulators use these guidelinesto help them identify the proposals that are likelyto raise concerns about adverse effects of mergerson competition.

The DOJ’s antitrust standards identify potentiallyanticompetitive mergers in terms of prescribedlevels, and changes in levels, of a commonly usedmeasure of market concentration, the Herfindahl-Hirschman Index (HHI). HHI is calculated by squaringeach bank’s share of deposits in a market and thensumming these squared shares. The index numbercan range from zero (a perfectly competitive market)to 10,000 (a pure monopoly).6 HHI is the preferredconcentration index because it accounts for themarket share of each bank in the market and givesgreater weight to the firms with larger market shares.Other indices, such as the three-firm or four-firmconcentration ratio, do not have both of these features.

According to the guidelines, a market can bebroadly characterized as unconcentrated if HHI isless than 1000 points, as moderately concentratedif HHI is between 1000 and 1800, and as highly

30 NOVEMBER/DECEMBER 2003

5 The assumption of a link between market structure and the perfor-mance of firms in the market is commonly referred to as the structure-conduct-performance hypothesis.

6 The perfectly competitive market would consist of many firms, eachwith about the same market share. As the number of firms in thismarket increases, each firm’s share decreases until it approaches thelimit of zero. The square of zero is zero, so the sum of those squaresis still zero. The pure monopoly market would have only one firmthat controls 100 percent of the market. The square of 100 is 10,000.

Gilbert and Zaretsky R E V I E W

3 United States v. Philadelphia National Bank, 374 U.S. 321 (1963). SeeShull (1996) for an historical perspective on banking antitrust.

4 This is how the Court interpreted Congress’s directive to “arrest anti-competitive tendencies in their ‘incipiency’.”

concentrated if HHI is above 1800.7 These thresh-olds apply not only to banking, but to all industriesin the United States. The DOJ distinguishes bankingfrom other industries, however, by allowing it morelatitude for increases in HHI. That is, the DOJ normallywill not challenge a bank merger or acquisition unlessthe resulting increase in HHI is at least 200 pointsand the post-merger market HHI is at least 1800(highly concentrated).8 For other industries, anincrease in HHI of at least 50 points in a highly con-centrated market will trigger closer scrutiny by theDOJ. The additional cushion afforded the bankingindustry accounts for the competition banks nowface from thrifts, credit unions, and other providersof financial services. In fact, thrifts—that is, savingsand loan associations and savings banks—so resem-ble banks today in their financial service offeringsthat federal regulators routinely account for thriftdeposits when calculating a banking market’s HHI.9

Deposits at thrifts are commonly included at 50percent weight, however, because thrifts target theirfinancial services primarily to consumers and notbusinesses, which commercial banks typically ser-vice.10 Thrifts’ focus on consumers dates back tobefore deregulation (1980), when they were restrictedto accepting only savings deposits and to makingpredominantly residential real estate loans. Post-1980, changes in the law gradually allowed thriftsto expand their menu of offerings and services,enabling them to resemble and compete moredirectly with banks. These changes ultimately ledthe Federal Reserve Banks, in 1987, to incorporatethrift deposits (at half weight) into their market anti-trust analyses.11

Deposits at credit unions are rarely included inbanking antirust analyses. Being membership orga-nizations, credit unions offer their financial servicesonly to their members, and these services are usuallyquite limited when compared with those offered bybanks and thrifts. As such, credit unions do not nec-essarily compete in the same product market asbanks and thrifts. In certain cases, however, creditunion deposits may be included in the analysis of aspecific market (at fractional weighting) if substantialevidence supports their inclusion. One piece of suchevidence would be that the share of deposits at creditunions in the market area greatly exceeded thenational average. In addition, a particular credit unionshould have liberal membership rules (typically, atleast 70 percent of market residents must be eligiblefor membership) and offices that are easily accessibleto local residents.

Determining the change in HHI and its post-merger level is not the end of the story. If thesenumbers were to fall outside of the DOJ guidelinethresholds, it would not automatically mean thatthe merger or acquisition would be denied. Suchan outcome would indicate only that regulatorswould consider the concentration of the market tobe high enough to permit the firms in the market tokeep prices above the competitive level for a signifi-cant period. Such a case would require that a more-detailed economic analysis be conducted before adecision could be made. This analysis would seekto determine whether other factors, such as potentialcompetition and economic conditions of the market,could mitigate the anticompetitive structural effectof the merger and, thereby, suggest that the HHI doesnot tell the whole story. An applicant might avoidthe more-detailed analysis, however, if it were tochoose or agree to divest to a third party some of itsoffices in the affected markets to get those markets’competitive structures to fall within guidelines.12

Having a post-merger HHI and an increase inHHI that exceed DOJ thresholds is not the only rea-son an application might receive closer scrutiny. Abank that would end up controlling more than 35percent of the deposits in a particular market aftera merger or acquisition would also trigger a morein-depth examination by the Federal Reserve, evenif the HHI measures indicate no significant change


12 Most often, large banking organizations use divestiture of bankingoffices as a means of securing approval for a proposed merger oracquisition. See Pilloff (2002) and Webb (2001) for discussion ofdivestiture in antitrust analysis.

FEDERAL RESERVE BANK OF ST. LOUIS Gilbert and Zaretsky

7 DOJ, Antitrust Division, and the Federal Trade Commission, 1992Horizontal Merger Guidelines (revised April 8, 1997).

8 See Cyrnak (1998, p. 704).

9 Savings and loan associations, which may also be known as savingsassociations, S&Ls, building and loan associations, cooperative banks,or homestead societies, include both mutual and stock associations.Both mutual and stock savings banks are included.

10 In some cases, deposits at thrifts may be weighted more or less than50 percent, depending on the level of activity a particular thrift hasin a region’s commercial lending market. In addition, deposits of thriftsubsidiaries of commercial banking organizations are included in theHHI calculation at 100 percent.

11 The DOJ does not include any thrift deposits in its banking antitrustanalyses, unless the proposal fails its competitive screen and individualcircumstances warrant a particular thrift’s inclusion at 100 percentweight. One example would be a ratio of commercial and industrialloans to total assets at thrifts in a market area that exceeds 2 percent.

in market concentration.13 Indeed, a bank subject tothe antitrust authority of the Federal Reserve can cer-tainly control more than 35 percent of total deposits inany given market, but, in most of these cases, the highmarket share would not have resulted from a merger.For instance, a bank could have achieved a large

market share through internally generated growthrather than through acquisitions; alternatively, achange in market definition may have increasedthe share of total deposits the bank controls in thatmarket. If a merger or acquisition were to result ina bank controlling more than 35 percent of marketdeposits, the antitrust analysis would focus on whetherany factors might mitigate the anti-competitiveeffects of that merger. An example of such a mitigat-



MEASURES OF MARKET STRUCTURE

Summary of Deposits

The Summary of Deposits contains depositdata for more than 85,000 branches/offices ofinstitutions insured by the Federal Deposit Insur-ance Corporation, which collects deposit balancesfor commercial and savings banks as of June 30of each year. The Office of Thrift Supervisioncollects the same data for savings institutions.These data are used for measuring concentrationof deposits among depository institutions withoffices located in each market area.

Loans Reported under the CRA

Beginning in 1996, all banks and savingsinstitutions (savings banks and savings and loanassociations) with total assets in excess of $250million, or banks of any size that are subsidiariesof holding companies with assets of $1 billion ormore, must report the number and amount ofsmall business loans by location. Small businessloans are defined as commercial and industrialloans of less than $1 million. Coding of these CRAdata by geographic area makes it possible to mea-sure the influence of lending by large depositoryinstitutions on banking concentration in marketareas where these institutions do not have offices.

MEASURES OF BANK PERFORMANCE

Call Reports

Each quarter, each bank files a balance sheetand an income statement with its supervisor. Thesedata are used for calculating profit rates, commonlymeasured as net income after tax as a percentageof total assets (called return on assets, or ROA).

Some studies use the call report data to calcu-late the average interest rates that banks paid onvarious categories of deposits. The interest ratepaid on each category of deposits is derived bycalculating interest paid as a percentage of averagedeposits. A major challenge involves calculatinginterest rates on time deposits, since a bank’s out-standing time deposit liabilities may have beenissued at various points in time in the past. Averageinterest rates paid calculated with data from thecall report are likely to reflect more accurately theinterest rates that banks were offering to depositorsat the time of the call report for very short-termdeposits, such as savings accounts, than for timedeposits.

Monthly Survey of Selected Depositsand Other Accounts

Between April 1982 and September 1997, theFederal Reserve asked a sample of several hundredbanks to report monthly on the interest rates theypaid on specific categories of deposit accounts.In addition, the Fed asked these banks about thefee structures on these deposit accounts in annualsupplements that were collected between 1989and 1997. Since each bank in the sample filed onereport, the authors who use these data for testingthe SCP hypothesis have to make various adjust-ments for the fact that many of the banks in thesample had offices in more than one market area.1

Continued on p. 33

Sources of Data Used to Evaluate the Assumptions of Banking Antitrust

1 Simons and Stavins (1998) use data on interest rates from the“Monthly Survey of Selected Deposits and Other Accounts” to testthe SCP hypothesis for banks. They assume that each bank in theirsample paid the same interest rate on deposits at each of its offices,although they did not have information to support this assumption.

13 See 12 CFR 265.11 and DOJ, “Bank Merger Competitive Review—Introduction and Overview” (1995), current as of September 2000.

ing factor could be that recent economic growth inthe market has been strong enough to indicate thatit is attractive for entry by other banks or thrifts.14

This process of antitrust analysis appears quitecut and dried. It relies, however, on several assump-tions: Market concentration is the relevant dimen-sion of market structure; market concentration isaccurately and adequately measured; and the effectsof a merger on market concentration translatesconsistently into anticipated effects on the pricingbehaviors of the players in the market. In otherwords, the process assumes that the “structure-conduct-performance” (SCP) hypothesis correctlymodels the true market mechanism and that HHI,used to measure market concentration, reflects the

relevant facets of market structure. Suppose, how-ever, the hypothesis does not accurately model thetrue market mechanism. In that case, regulatorscould be using the hypothesis appropriately anddrawing the correct conclusions from it, but stillend up with unintended policy outcomes becausethe hypothesis is the wrong analytical model ofbanking competition in the first place. Several ofthe following studies directly test whether the SCPhypothesis holds for the banking industry.

In addition, the process of banking antitrustdepends on HHI being a reliable measure of marketstructure. If HHI is not the relevant measure of marketstructure, the regulators and the DOJ may be missingimportant structural information that is not capturedby HHI. This issue is also addressed in a number ofthe studies described in the following section.



Continued from p. 32Bank Rate Monitor

Bank Rate Monitor is the name of a publicationthat conducts surveys of the interest rates thateach office of large banks located in urban areascharges on various categories of loans and payson various categories of deposits. Data from thissource have important advantages over othersources of data for purposes of measuring bankperformance. As a survey of the interest rates thatbanks offer on categories of deposits at variouspoints in time, this source avoids problems inher-ent in the use of call report data to measure theinterest rates that banks offer to pay on timedeposits. Data from Bank Rate Monitor avoid theproblem inherent in the data from the MonthlySurvey of Selected Deposits and Other Accountsof interpreting reports by banks that have officesin more than one market area. In addition, theMonthly Survey of Selected Deposits and OtherAccounts ceased in 1997, whereas Bank RateMonitor continues to survey the interest rates thatbank offices charge on loans and pay on deposits.Bank Rate Monitor data, however, are availablefor only relatively large depository institutionswith offices in relatively large metropolitan areas.

Survey of the Terms of Lending toBusiness

This survey is conducted by the FederalReserve. Once each quarter, the banks in the survey

sample provide detailed information on eachbusiness loan made during a period of one week.This report is used to derive the average interestrate that banks in the sample charged on smallloans to businesses. The survey does not includeinformation that can be used to measure the sizeof the borrowers. These small, business loans areassumed to be made to small businesses.

National Survey of Small BusinessFinances

The survey, conducted by the National OpinionResearch Center for the Board of Governors ofthe Federal Reserve System during 1987, 1993,and 1998, includes characteristics of firms andowners, the firms’ use of financial services andfinancial service suppliers, and income and bal-ance sheet items. The survey was renamed theSurvey of Small Business Finances in 1998.

Survey of Consumer Finances

This Federal Reserve survey, which is con-ducted at intervals of three years, includes infor-mation on selected demographic characteristicsof U.S. families, including their income, balancesheets, and use of financial services. Data fromthis survey are used to examine trends over timein the degree to which households obtained theirfinancial services from depository institutionslocated in their communities.

14 For other examples of mitigating factors, see Holder (1993).

EMPIRICAL RESEARCH

The literature has pursued various approachesto analyzing the assumptions that underlie bankingantitrust. Some studies examine empirical evidenceon the assumptions themselves. Other studies testhypotheses about the association between localmarket concentration and measures of bank per-formance, such as the profit rates of banks and theinterest rates they charge on loans and pay ondeposits. In still other articles, which investigateissues other than the validity of the assumptions ofbanking antitrust, the empirical results end up havingimplications for the relevance of local market areasfor banking antitrust. The boxed insert of this paperdescribes the various surveys and other sources ofdata authors have used in such studies.

Evidence on the Validity of theAssumptions

Survey Data: Location of Banks and TheirCustomers. Several studies use survey data on thelocation of banks and their customers to assess thevalidity of the assumption that customers tend toobtain their financial services from firms withoffices located in their communities. Kwast, Starr-McCluer, and Wolken (1997) use data from the 1992Survey of Consumer Finances and the 1993 NationalSurvey of Small Business Finances to examine theextent to which households and small businessesobtain financial services from local bank offices.The authors conclude that the data presented intheir study support such an assumption.

Amel and Starr-McCluer (2002) use data fromthe Survey of Consumer Finances for the years 1989through 1998 to examine trends over time in thedegree to which households obtained their financialservices from depository institutions located in theircommunities. They interpret their results as indicat-ing that households with at least one bank-typeaccount or loan continue, to a substantial degree, toobtain certain key financial services at local deposi-tory institutions.15 The tendency to obtain servicesfrom local institutions is especially pronounced fortransactions accounts. The data for 1998, however,tend to undermine the concept of a cluster of finan-



cial services that households demand from commer-cial banks. Except for checking accounts, Amel andStarr-McCluer’s results demonstrate that the percent-age of households obtaining their financial servicesfrom local banks has fallen substantially over time.For instance, the share of households that obtainedmoney market accounts from local depository insti-tutions declined from 78.4 percent in 1989 to 63.6percent in 1998. The share of households that bor-rowed from local depository institutions declinedfrom 73.3 percent in 1989 to 44.8 percent in 1998.

Petersen and Rajan (2002) use data from the1993 National Survey of Small Business Financesto draw inferences about changes over time in thedistance between small businesses and the firmsthat provide their lending and transactions services.Knowing the date the lending relationship beganand the distance between the lender and the firm,the authors find that these distances have increasedover time, from an average of 51 miles for lendingrelationships that began in the 1970s to an averageof 161 miles for relationships that began in the1990s. To obtain transactions services, on the otherhand, small businesses continue to favor banks withoffices in their communities. Petersen and Rajanargue that their results support wider geographicareas for markets in banking antitrust than havebeen used in the past. The authors acknowledge,however, that their results are subject to severalpossible biases. One such bias involves the survivalof relationships over time. Results for the 1970s, forinstance, are based on relationships between smallbusinesses and lenders that began in the 1970s andremained in existence in 1993, the year of the survey.

Approaching the distance question from a differ-ent angle, Wolken and Rohde (2002) use data fromthe National Survey of Small Business Finances for1993 and 1998. Distances between small businessesand their financial service providers varied substan-tially by category of service. In both 1993 and 1998,about 96 percent of small businesses obtainedchecking account services from financial institutionswith offices located within 30 miles of the smallbusinesses’ headquarters. For those two years, thepercentage of small businesses with lines of creditfrom financial institutions located within 30 milesof their headquarters was about the same: 85.1percent in 1993 and 83.6 percent in 1998. Thus,small businesses continue to obtain these basic finan-cial services from financial institutions located intheir communities.

Although these four studies provide mildly

15 Bank-type accounts or loans include checking, savings, money market(both money market deposit and money market mutual fund), broker-age, individual retirement, and Keogh accounts; certificates of deposit;trusts and other managed asset accounts; first and second mortgages;motor vehicle loans; home equity and other lines of credit; and otherconsumer loans. It does not include credit cards.

conflicting empirical support for the assumptionthat customers continue to receive many financialservices from local depository institutions, theirresults are limited because they reflect only demandfor financial services under existing prices. Surveydata cannot help us understand how householdsand small businesses would respond to changes inthese prices. The results also do not describe or helpdetermine what share of financial services fromlocal depository institutions is sufficiently high tosupport the current methods of banking antitrust.

Evidence of National Versus Local Marketsfor Banking Services. Jackson (1992) tests thehypothesis that the geographic scope of bankingmarkets is national rather than local. In his regres-sion analysis, the dependent variables are themonthly changes in interest rates on three cate-gories of deposits from a sample of banks in 29metropolitan statistical areas (MSAs). The indepen-dent variables include the current and laggedchanges in the average monthly rate on the 6-month Treasury bill and a set of dummy variablesfor each MSA. The null hypothesis is that the sumof the coefficients on the dummy variables for anMSA, which represents how interest rates in theMSA adjust relative to a change in the national rate,is zero. If so, then the rate movements in the MSAmatch the movement in the national market. If thesum is significantly different from zero, then theinterest rate adjustment on a particular type ofdeposit at banks in the selected MSA is differentfrom that rate’s adjustment in the national market.Jackson finds that in a significant number of MSAsthe interest rate adjustments on money marketdeposit accounts (MMDAs) and super NOW accountsare statistically different from adjustments in thenational market. In other words, banks do not com-pete in a national market for MMDAs and superNOW accounts. For 6-month certificates of deposit(CDs), however, the null hypothesis cannot berejected, which implies that banks do compete forthese CDs in a national market. Depositors withMMDAs and super NOW accounts have frequentcontact with their banks, whereas customers whoinvest in 6-month CDs may limit contact with theirbanks to once every six months. Jackson’s resultsare consistent with the use of local market areasrather than one national market in banking antitrustanalysis for certain types of transactions accounts.

Jackson and Eisenbeis (1997), using the samedata as Jackson (1992), employ cointegration analy-sis to determine whether the interest rates on the

various deposit accounts are determined in local ornational markets. Using the interest rate on 6-monthTreasury bills to represent the national market, theauthors test whether the deposit interest rates arecointegrated with the 6-month Treasury bill rate—that is, whether all of the series follow a commonlong-run trend. If they do, then the authors can con-clude that the deposit interest rates are determinedin a national market. Jackson and Eisenbeis findthat MMDAs, super NOW accounts, and 6-monthCDs are all cointegrated with the 6-month Treasurybill rate and, therefore, are all determined in thesame national market.

Cointegration analysis, however, is not well-suited to test the hypothesis that banks competefor deposits in a national market because it candetect only the common long-run trend of the series.In the short run, the deposit interest rates bankspay in a local market could deviate substantiallyfrom the 6-month Treasury bill rate, thus suggestingthey do not compete in the same market. The twoseries could still be cointegrated (follow a commonlong-run trend), though.

Are Banks with Offices in the Local MarketArea the Relevant Competitors? Would the mea-sure of concentration in a local market area besubstantially different if it were adjusted for thefinancial services provided by firms that do nothave offices in the market area? Cyrnak (1998)investigates this issue using data from the reportsrequired of large banking organizations under theCommunity Reinvestment Act (CRA). He finds thatthe market concentration of loans to small busi-nesses tends to be substantially lower if the relevantcompetitors include all banks that lend to smallbusinesses in the local market area, whether thebanks have offices in that local market area or not.The effect on concentration ratios of including theout-of-market lenders is especially pronounced forrural banking markets. Woosley, King, and Padhi(2000) extend Cyrnak’s work by identifying thoserural banking markets where including out-of-market CRA data would cause market concentra-tion to fall below the DOJ’s HHI guidelines. Thesestudies raise questions about whether it is appro-priate to limit the relevant competitors in antitrustanalysis to the banks with offices located in localmarket areas.

Cyrnak and Hannan (1999) investigate whetherthe concept of a cluster of banking services is rele-vant for the pricing of bank loans for small busi-nesses. In addition, they investigate whether the



relevant competitors for antitrust analysis are bankswith offices located in the local market area. Theirmeasure of bank performance is the interest rateon loans to small businesses, derived from the Surveyof the Terms of Lending to Business. About 300 banksreport information on each of their business loansoriginated during one week each quarter.

The authors develop three measures of marketconcentration:

1. HHI based on the deposits of banks withoffices in the market area.

2. HHI based on small business loans by bankswith offices in the market area.

3. HHI based on CRA data and an estimate ofthe amount of small business loans made bysmall banks with offices in each marketarea.

The authors find that the HHI measure basedon the deposits of banks with offices located in themarket area has more power to explain the interestrates charged on small business loans than do theother concentration measures. They conclude thattheir results support the current approach to bank-ing antitrust, which is based on the HHI calculatedfor the deposits of banks with offices in each localmarket area.

Hannan (2003) uses CRA reports to examinethe magnitude of loans that relatively large bankingorganizations made to small businesses located inmarket areas where the lenders do not have offices.Including lenders with large credit-card business(because small businesses might view credit cardsas a substitute for bank loans), there was a largeincrease in the number of small business loansmade by out-of-market lenders between 1996 and2001. The effect of these lenders on the supply ofloans to small businesses is much smaller whenmeasured in terms of dollars of lending; it is evensmaller if the known credit-card lenders are elimi-nated as out-of-market lenders. These observationsindicate that considerable numbers of small loansfrom a few large banking organizations with sub-stantial credit-card operations dominate the mea-sures of out-of-market lending that we can derivefrom CRA data; that is, much of the out-of-marketlending is credit-card related.

Hannan also finds that the share of small busi-ness loans from out-of-market banks tends to behigher in markets with a relatively high concentra-tion of deposits at in-market banks. He concludesthat his results are consistent with an erosion in the

validity of the assumption that banking markets arelocal geographic areas; he argues, however, that itis not clear at this time whether the erosion wouldjustify a substantial broadening of defined geographicmarkets for antitrust analysis.

Relevance of Money Market Mutual Fundsfor Banking Antitrust Analysis. Pilloff (1999c)investigates the degree to which shares of retailmoney market mutual funds (MMMFs) are substi-tutes for federally insured accounts at depositoryinstitutions. This issue has implications for bankingantitrust because, if shares of MMMFs are closesubstitutes for deposit accounts, then depositoryinstitutions with offices in highly concentratedmarket areas would be less able to extract monopolyprofits by paying relatively low interest rates ondeposit accounts. Pilloff emphasizes three pointsin his argument why, for most households, MMMFsare not close substitutes for accounts at depositoryinstitutions. First, MMMFs require minimum initialinvestments that tend to be higher than the mini-mum initial deposit balances depository institutionsrequire. Second, although MMMFs permit customersto write checks against their shares, the minimumcheck amounts usually exceed the amount of manyroutine household payments. Third, MMMFs arenot federally insured like accounts at depositoryinstitutions, which adds a certain degree of risk.

Pilloff also uses survey data to support hisargument that MMMFs are not close substitutes fordeposit accounts. According to the survey, only 5.7percent of households in 1995 owned shares ofMMMFs, and almost all of these households alsohad accounts at depository institutions. Thus, eventhe small minority of households that held liquidassets with MMMFs did not find it in their interest toclose all of their accounts at depository institutions.

Uniform Pricing by Banks with Offices inMany Communities. Another assumption of bank-ing antitrust is that banks set each office’s interestrates—those charged on loans and those paid ondeposits—according to the concentration in themarket area where the office is located. Radecki(1998, 2000), however, finds that banks with officesin several communities within a state offer the sameinterest rate at each office for a particular loan ordeposit category. His data are from the Bank RateMonitor, which reports the interest rates posted bythe individual offices of a large number of bankslocated in many urban areas. Radecki thereforeconcludes that the geographic area for bankingmarkets in antitrust analysis should be no smallerthan a state.



Several staff of the Board of Governors haveconducted a number of studies that focus on theimplications of Radecki’s findings for banking anti-trust. For example, Heitfield (1999), using data fromthe Bank Rate Monitor, also finds that banks withoffices located in several local market areas (multi-market banks) tend to post the same interest rateon the same type of deposit at each office. Heitfeldextends this analysis by examining the interest ratesposted by banks that have all of their offices in onelocal market area (single-market banks). He findssignificant variation across local markets in the inter-est rates that single-market banks offer on variouscategories of deposits; he reasons that this variationreflects local market conditions. He does not presentresults for interest rates on the loan categories inthe Bank Rate Monitor survey that Radecki includesin his studies. Heitfeld concludes that the results ofhis study support the current practice of focusingon local market areas in banking antitrust.

Heitfield and Prager (2002) investigate whetherthe relevant geographic market areas for bankingantitrust have expanded beyond the traditional localmarket areas of MSAs for urban banks and countiesfor rural banks. They use call report data to estimatethe average interest rate each bank paid on NOWaccounts, MMDAs, and savings accounts in 1988,1992, 1996, and 1999. For each category of deposits,the authors regress the bank’s interest rate on, amongother independent variables, measures of concen-tration at the local and state levels. They find thatthe coefficients on the measures of local marketconcentration are negative and statistically signifi-cant in most equations and that the magnitude ofthe coefficients has not declined over time. In someof the equations, they also find that the coefficientson concentration at the state level are negativeand significant. Heitfield and Prager conclude that,although measures of local market concentrationremain useful indicators of the market power ofbanks, measures of market structure for broadergeographic areas may be relevant for banking anti-trust, too.

Hannan and Prager (2003) investigate whetherconcentration in local market areas affects thedeposit interest rates that single-market banks offer.They find that it does. The authors also find thatthe interest rates that single-market banks pay ondeposits tend to be lower in local areas where multi-market banks account for a greater share of marketdeposits. Furthermore, the relationship between

local market concentration and deposit interest ratesoffered by single-market banks becomes weaker asmulti-market banks account for larger shares ofdeposits in those market areas where both haveoffices.

Tests of the Structure-Conduct-Performance Hypothesis

According to the SCP hypothesis, the ability ofbanks in a local market area to set relatively highinterest rates on loans or low interest rates ondeposits depends on the structure of the market.Such behavior is assumed to be more effective inmarket areas where concentration is relatively high.Hannan (1991b) examines the theoretical founda-tion for this SCP hypothesis in banking. Table 1summarizes several features of studies that testhypotheses about the effects of local market struc-ture on various measures of bank performance.

Effects of Local Market Concentration onBank Profits. Banks that are more effective inaffecting the interest rates they charge on loansand pay on deposits will tend to have higher profits.Some studies test the hypothesis that there is apositive association between the profit rates ofbanks and local market concentration.

Rhoades (1995) tests the hypothesis that mea-sures of market structure in addition to HHI influencethe profit rates of banks. In regressions with theaverage profit rate of banks in market areas as thedependent variable, the coefficient on HHI is positiveand statistically significant. He finds, however, thatother indicators of market structure, including thenumber of banking organizations with offices in themarket area and measures of the inequality of banks’market shares in local areas, are also significant.

Moore (1998) investigates whether, in responseto financial innovations and changes in regulations,the influence of local market structure on bankprofits has tended to decline over time. In his analysis,the measure of performance is net income after taxesdivided by total assets (return on assets, or ROA) ofall banks in the market area. The measure of marketstructure is HHI. Moore finds that the statisticalsignificance of HHI has declined over time for ruralmarket areas. With the ROA of each market in thesample as the dependent variable, the coefficienton HHI was positive and statistically significant forrural areas in 1986 and 1987, but not in 1996 and1997. In similar regressions for urban markets, thecoefficient on HHI was not statistically significantfor any of these four years. Moore concludes that,





Tests of the Structure-Conduct-Performance Hypothesis for the Banking Industry

Consistent with Measures of Measures of current practice

Authors bank performance markets structure Sample of bank antitrust?

Effects of market concentration on the profit rates of banks

Rhoades (1995) Mean ROA of the HHI and other 1,684 urban and Yesbanks in each measures of rural markets, market market structure 1990-92

Moore (1998) ROA of individual HHI Urban areas and rural Nobanks counties, 1986, 1987,

1996, and 1997

Pilloff (1999c) ROA of individual HHI and measures 6,233 banks, 1992-95 Yesa

banks of multi-market contact

Pilloff (1999b) ROA of individual HHI and the presence 1,728 institutions with Yesbanks with all offices of large banking offices in 762 rural in the rural banking organizations in banking markets,markets included rural market areas 1995-96in the study

Pilloff and Mean ROA of banks in HHI National sample of YesRhoades (2002) each market urban and rural

markets, 1975-98

Akhigbe and Profit efficiency HHI 35,807 observations, YesMcNulty (2003) relative to the banks 1990, 1992, 1994,

on the efficient and 1996frontier

Berger (1995) ROA and ROE HHI 4,800 banks, 1980-89 Nob

Frame and ROA Market share of each Quarterly data for Qualified yesc

Kamerschen (1997) bank in the sample 208 banks with all offices in one rural county in Georgia, 1990:Q4–1994:Q4

Effects of market concentration on the interest rates that banks pay on deposits

Berger and The interest rates CR3 Quarterly data for 470 Yesd

Hannan (1989) that banks paid on banks in 195 local MMDAs, NOW banking markets, accounts, and time September 1983–deposits of various December 1985maturities, from the MSSDOA

Calem and Interest rates paid by CR3 466 banks in 105 urban YesCarlino (1991) banks on MMDAs banking markets,

and 3- and 6-month October 1983–CDs, from the November 1987MSSDOA

Table 1





Sharpe (1997) Interest rates paid by HHI Monthly data on Yesbanks on MMDAs interest rates for and 6-month CDs, 222 banks located from the MSSDOA in 105 markets,

October 1983–November 1987

Hannan (1997) Interest rates paid by HHI, measures of About 300 urban No: coefficients on banks on NOW market share banks, November HHI not accounts, MMDAs, inequality, and 1993 statistically and 3-month CDs, number of banks in different from from the MSSDOA the market zero

Prager and Interest rates paid by HHI 468 banks: 26 in Yes: declines in Hannan (1998) banks on NOW markets with interest rates over

accounts, MMDAs, substantial mergers, the sample period and 3-month CDs, 30 in markets with were larger at from the MSSDOA less-substantial the banks in

mergers, 412 in market areas with markets not affected substantial by mergers; interest horizontal rate data for mergers, as October 1991– defined in the August 1994 DOJ guidelines

Heitfield and Interest rates paid by HHI and CR3 Most banks in the Qualified yes: Prager (2002) banks on NOW United States, 1988, measures of

accounts, MMDAs, 1992, 1996, and 1999 concentration and savings statistically accounts, derived significant at both from call reports the local and

state level

Hannan and Interest rates paid by HHI 7,700 single-market Yes, except the Prager (2003) single-market banks banks in 1,925 urban results are

(those with most of and rural areas (1996) consistent with a their deposits from and 6,502 single- declining offices in one market) market banks in influence over on NOW accounts, 1,806 banking time in the MMDAs, and savings markets (1999) effects of local accounts, derived market structure from call reports on deposit

interest rates

Effects of market concentration on the responsiveness of deposit interest rates to changes in market rates

Hannan and Interest rates paid HHI Monthly data on Yes Berger (1991) by banks on deposit interest rates

MMDAs, from paid by 398 banks in the MSSDOA 132 banking areas,

September 1983–December 1986

Table 1, cont’d





Neumark and Interest rates paid HHI 255 banks in 105 YesSharpe (1992) by banks on MMDAs urban markets,

and 6-month CDs, interest rates on from the MSSDOA deposits, October

1983–November 1987

Hannan and Responsiveness to CR3 About 300 banks, YesLiang (1993) the yields on 1983-89

Treasury securities of the interest rates paid by banks on MMDAs and 2- and 3-year CDs, from the MSSDOA

Khan, Pennacchi, Interest rates paid by HHI Over 600 banks, Yes: banks located and Sopranzetti banks on MMDAs November 1983– in market areas (1999) and CDs with May 1994 with higher HHI

maturities of 3, 6, and are more likely 12 months, from the to set their MSSDOA deposit interest

rates as integers or quarter integerse

Effects of market concentration on the interest rates that banks charge on loans

Cyrnak and Interest rates charged Three measures of 228 banks located in YesHannan (1999) by banks on small HHIf 98 urban areas that

business loans participated in the ($100,000 or less) in STLB in May 1996the STLB

Hannan (1991a) Interest rates charged HHI 8,250 business loans Yesby banks on by 260 urban banks, business loans, from reports in from the STLB different interest rate

environments, 1984, 1985, and 1986

Hannan and Interest rates charged HHI Over 300 banks’ small Qualified yes: Liang (1995) by banks on business loans, equations with

business loans less August 1989, May the best fit use than $100,000, from 1990, and May 1991 HHI calculated the STLB with zero weight

for thrift institutions

Table 1, cont’d





Hannan (1997) Interest rates HHI measures of Loans reported by Qualified yes: HHI charged by banks market share sample of banks in does not reflect on small business inequality and November 1983: the only relevant loans (less than number of banks 511 unsecured loans measure of $100,000), from the in the market and 2,059 secured market structure; STLB loans the number of

banks in the market also affects loan rates

Berger, Rosen, Interest rates charged HHI and the asset 520 small businesses Qualified yesh

and Udell (2001) by banks on loans size of organizations that obtained credit to small businesses with offices in the from their banks under lines of credit market areas under lines of credit in 1993g in 1993

Kahn, Pennacchi, Interest rates charged HHI Weekly surveys of Yes: interest rates and Sopranzetti by banks on interest rates quoted on consumer (forthcoming) consumer and auto by large banks in 10 loans are higher

loans: data from the urban areas, 1989-97 in market areas Bank Rate Monitor with higher HHIi

NOTE: ROA, annual net income after taxes divided by average annual assets; ROE, annual net income after taxes divided by the bookvalue of equity; HHI, Herfindahl-Hirschman Index (see section “Current Method of Antitrust in the Banking Industry”); CR3, percentageof deposits at banking offices in a market area at the institutions ranked first through third in terms of deposits at offices in the marketarea; MMDA, money market deposit account: a short-term deposit account at a depository institution on which customers may writea limited number of checks; NOW account, a transactions account available to individuals and non-profit organizations on whichdepository institutions may pay interest; CD, abbreviation for certificate of deposit, which is an interest-earning deposit account witha fixed maturity date; MSSDOA, Monthly Survey of Selected Deposits and Other Accounts (see boxed insert); STLB, Survey of the Termsof Lending to Businesses (see boxed insert).aPilloff (1999c) also finds that the coefficient on the measure of multi-market contact is consistent with the linked oligopoly theory.bA positive association between profit rates and market concentration disappeared when measures of X-efficiency were added asindependent variables.cThe empirical results are consistent with the hypothesis that the banks in the sample exercised market power. The measure of marketstructure, however, is the market share of the sample bank, not a measure of market concentration. The only aspect of market sharethat is relevant under the current banking antitrust procedures for banking involves closer scrutiny for cases in which a merger createsa bank with a market share of 35 percent or higher.dAlthough the published results use CR3 as the measure of market concentration, the results are qualitatively similar using HHI.eThe model developed by Kahn, Pennacchi, and Sopranzetti (1999) implies that banks in less competitive markets are more likely touse integers to set the interest rates they offer to pay on deposits.fThe three measures of HHI in Cyrnak and Hannan (1999) are based on (i) the deposits of banks with offices located in each marketarea, (ii) the small business loans of banks with offices in the market areas, and (iii) the second measure of HHI adjusted for businessloans to residents of each market area by large banks that do not have offices in the market area.gBerger, Rosen, and Udell (2001) derive interest rates on loans from the National Survey of Small Business Finances, 1993.hBerger, Rosen, and Udell (2001) find a dimension of local market structure not considered in banking antitrust: the total assets ofbanks with offices in the market area.iIn Kahn, Pennacchi, and Sopranzetti (forthcoming), results of analysis of the effects of mergers on consumer interest rates and thedynamics of consumer interest rates in response to changes in market areas are also consistent with the assumptions of banking antitrust.

Table 1, cont’d

because market concentration no longer seems tohave a significant effect on bank profits, local areasare no longer the relevant market areas for antitrust.

Pilloff (1999a) estimates a regression in whichthe dependent variable is the ROA of individualbanks and the independent variables include marketHHI and measures of multi-market contact amongthe banks with offices in the market. His findingthat the coefficients on HHI are positive and statis-tically significant supports the current approach tobanking antitrust. Pilloff also finds that the coeffi-cients on measures of multi-market contact amongthe banks in each market area are positive and sta-tistically significant, providing empirical supportfor the linked oligopoly theory—that is, when thesame banks compete with each other in severaldifferent markets, they will tend to limit their rivalryfor customers in each of the markets.

Pilloff (1999b) examines the determinants ofROA among small, rural banks that have all of theiroffices in local markets, which are identified as coun-ties. Pilloff estimates the effects of market HHI andthe presence of the offices of large banking organi-zations on the ROA of small banks. While Pilloff findsthat the presence of offices of large banks in ruralmarkets tends to increase the ROA of small banks(a sign of reduced competition), he also finds thatthe coefficient on HHI is insignificant. The insignifi-cant coefficient on HHI may reflect the inclusionof another independent variable, a measure ofmarket size that tends to be correlated with marketconcentration.

Pilloff and Rhoades (2002), using data for alarge number of rural and urban banking marketsfor the years 1975 through 1998, test the hypothesisthat banks located in market areas with higherconcentration tend to have higher profit rates. Theauthors regress the mean ratio of net income tototal assets for banks in market areas (the dependentvariable) on HHI and other measures of market struc-ture, including the number of banking firms withoffices in the market areas (the independent vari-ables). The authors find that the coefficients on HHIare consistently positive and statistically significantat the 1 percent level. Other measures of marketstructure, including number of firms, have significantcoefficients in some years, but not in others.

Akhigbe and McNulty (2003) examine the associ-ation between bank profits and local market concen-tration using a technique that estimates the profitefficiency frontier for given levels of loans, deposits,and other determinants of bank profits. They esti-mate each bank’s deviation from the profit-efficiency



frontier as a function of various explanatory vari-ables, including market concentration (HHI). Theyfind that the banks located in market areas withhigher HHI (more concentrated) tend to be closerto the profit frontier.

These six studies examine whether the datasupport the hypothesis of a static, contemporaneousassociation between bank profits and market con-centration. Amel and Liang (1997), in contrast, exam-ine the evidence of a dynamic relationship betweenbank profits and market structure. They report evi-dence that lagged profit rates of local banks tend toinduce entry by additional banks, which, in turn,reduces that market’s concentration. De novo entry,however, would likely not affect a market’s concen-tration too much because de novo banks tend tocontrol only a small share of a market’s total depositsfor many years. Amel and Liang’s results are espe-cially interesting because their data are for the years1977 through 1988, a period when barriers to entryinto banking markets were higher than they are now.

SCP Versus Efficiency. A problem with using therelationship between bank profit rates and marketconcentration as a way to test the validity of the SCPhypothesis is that an alternative hypothesis, theefficient-structure hypothesis, leads to the samerelationship. Because of this similarity, a more-detailed description of the two hypotheses willhelp place the related empirical studies in theirproper context. The SCP hypothesis takes localmarket concentration as given and considers, forexample, the implications of that concentration forthe ability of banks to effectively collude on theterms of the services they offer to their customers.According to the SCP hypothesis, banks located inmore concentrated market areas should be able todetect local banks cheating on collusive agreements,and enforce penalties for such cheating, more effec-tively than banks located in less concentrated marketareas. The observed outcome would be that marketswith higher concentration have banks that earnhigher profits.

The efficient-structure hypothesis, rather thantaking market concentration as given, considersthe economic factors that help explain variation inconcentration across markets. To illustrate howeconomic factors could influence market concen-tration, assume initially that all market areas haveequal demand for banking services and that eachbank has the same cost structure. In addition, assumethat each bank is small relative to the demand forbanking services in each market area. Under these

to entry: banks with all of their offices located inrural Georgia counties during a period of the 1990swhen these banks were protected from intrastatebranching. In their profit equation, the measure ofcost efficiency is not statistically significant, whilethe measure of market structure is statistically sig-nificant with and without the cost efficiency mea-sure as an independent variable. The sign of thecoefficient on the market structure variable is con-sistent with the SCP hypothesis. The results in Berger(1995) and Frame and Kamerschen (1997) implythat the interpretation of a positive associationbetween profit rates and market concentration maydepend on the level of barriers to entry.

The studies examining the association betweenbank profit rates and market structure do not provideconsistent support for the current approach to bank-ing antitrust. Analysis of the association betweenthe prices of bank services and market structuremay yield more conclusive tests of the SCP hypoth-esis for the banking industry.

Effects of Local Market Concentration onInterest Rates Paid on Deposits. The SCP hypoth-esis implies that banks located in market areaswith relatively high concentration will tend to payrelatively low interest rates on deposits. Most of thestudies that examine the effect of banking marketconcentration on deposit interest rates use datafrom the “Monthly Survey of Selected Depositsand Other Account.” The boxed insert describesthis survey.

In a widely cited article, Berger and Hannan(1989) examine the effects of local market concen-tration on the interest rates banks paid on MMDAsbetween September 1983 and December 1985. Theyfind that the coefficient on local market concentra-tion is negative and statistically significant at the 1percent level. The size of this coefficient indicatesthat if the concentration of the least concentratedmarket were increased to that of the most concen-trated market, the interest rate banks in that marketpaid on MMDAs would decline by about 50 basispoints. The results of most of the studies listed inthe section of Table 1 on deposit interest rates arequalitatively similar to those of Berger and Hannan(1989).

Prager and Hannan (1998) investigate the effectsof bank mergers on deposit interest rates. Table 1describes their sample of banks and measures ofdeposit interest rates. Market interest rates declinedover their sample period. Declines in deposit interestrates were larger at banks located in markets where

assumptions, each market has the same number ofidentically sized banks and the economic profitsat all banks are zero. That is, each bank’s profitequals the return it would have earned had itinvested its capital in a firm in an industry otherthan banking.

Now suppose that some of the banks in somemarkets discover ways to change their cost structuressuch that at each level of output their total costs arelower than those of the other banks. The low-costbanks can then reduce their prices slightly (pre-sumably by not as much as costs fell) to increasetheir market shares and, consequently, earn higherprofits than the high-cost banks. And since the low-cost banks will be larger than other banks, marketconcentration will be higher in the market areaswhere the low-cost (high-profit) banks are located.The observed outcome again would be that marketswith higher concentration have banks that earnhigher profits.16

This analysis illustrates why evidence of a posi-tive association between bank profits and marketconcentration is consistent with both the SCP andefficient-structure hypotheses. The analysis alsodemonstrates why evidence of an associationbetween the prices of bank services and marketconcentration would tend to provide more relevanttests of the SCP hypothesis than evidence of anassociation between bank profit rates and marketconcentration.

Berger (1995) investigated whether empiricalevidence of an association between local marketconcentration and bank profits would provide sup-port for the SCP hypothesis or the efficient-structurehypothesis. He found that the positive associationbetween bank profit rates and market concentra-tion disappeared when a measure of the cost effi-ciency of banks was added to the regressions as anindependent variable, which is consistent with theefficient-structure hypothesis. Thus, his empiricalresults do not support the current approach to bank-ing antitrust. Berger emphasized that each of theequations in his study explains little of the variationin bank profit rates, with median R2 across equationsbelow 0.1, a finding similar to that in Pilloff andRhoades (2002) and Pilloff (1999a).

Frame and Kamerschen (1997), following Berger(1995), included a sample of banks that operatedin an environment with relatively high legal barriers



16 See Carlton and Perloff (2000, Chap. 8) for a discussion of the literatureon the SCP and efficient-structure hypotheses.

mergers created substantial increases in HHI, asidentified in the DOJ’s guidelines, than at bankslocated in other market areas. The results in Pragerand Hannan (1998) are consistent with the assump-tions that underlie the current approach to bankingantitrust analysis.

Two studies, Heitfield and Prager (2002) andHannan and Prager (2003), use recent call reportdata to derive estimates of the interest rates thatbanks paid on various short-term deposits. To derivethese estimates, the authors divided each depositcategory’s quarterly interest expense by its averagequarterly deposit balance, a method that is moreappropriate for short-term deposits than for long-term deposits because data for long-term depositsreflect deposits made over various dates in the past.Both studies find evidence to support the assump-tion that deposit interest rates tend to be lower inmarket areas with higher market concentration.Heitfield and Prager (2002) also find evidence thatbanks in states with higher state-level banking con-centration rates tend to pay lower deposit interestrates. Hannan and Prager (2003) find evidence thatindustry consolidation through nationwide branchbanking is weakening the influence of local marketconcentration on deposit interest rates.

Effects of Local Market Concentration on theResponsiveness of Deposit Interest Rates toChanges in Market Rates. The studies discussedin this section examine the influence of local marketconcentration on the dynamics of the interest ratesthat banks pay on deposits. This section discussessome of the details of each study because theyhave unique features. Table 1 presents additionalinformation about these studies.

Hannan and Berger (1991) develop a theoreticalframework to illustrate how local market concen-tration affects banks’ deposit pricing in the face ofchanging market interest rates. In this model, banksin more concentrated markets exhibit more pricerigidity than banks in less concentrated markets.The authors test the model empirically by estimatingthe probability that a bank, given its market’s con-centration, will adjust its deposit interest rate up,down, or not at all in response to a change in themarket rate—in this case, the 3-month Treasury billrate. They find that banks in more concentratedmarkets have a lower probability of increasingdeposit rates when market rates rise. Market concen-tration has no effect on the probability of decreasinginterest rates when market rates fall, however. Inother words, banks in less competitive markets are



less responsive to upward changes in Treasury billrates than those in more competitive markets; butthey are just as responsive to downward changesin Treasury bill rates. These results support thehypothesis that banks in a more concentratedmarket behave less competitively.

Neumark and Sharpe (1992) estimate the effectsof market concentration on deposit interest rateswith an asymmetric partial equilibrium model.The authors find that banks in markets with higherconcentration paid lower equilibrium interest rateson MMDAs and 6-month CDs. Moreover, Neumarkand Sharpe find that, when market interest rateschanged, MMDA rates at banks in more concentratedmarkets tended to go up slower and down fasterthan at banks in less concentrated markets. Marketconcentration affected rates on 6-month CDs simi-larly, but not as strongly. The difference betweenthe MMDA and 6-month CD results may reflect atendency for bank customers to shop in wider geo-graphic areas for longer-term investments. It mayalso reflect greater reluctance for banks to raiseinterest rates on MMDAs than on CDs, since anincrease in the interest rate a bank pays on MMDAsaffects its interest expense on all MMDAs, whereasan increase in the interest rate it pays on CDs affectsonly its marginal CD accounts—newly contractedCDs and rollovers—because the interest rates paidon CDs issued in the past remain unchanged.

Hannan and Liang (1993) test the hypothesisthat banks are price takers in the markets for MMDAs,2-year CDs, and 3-year CDs. For each of the morethan 300 banks in the sample, they estimate a time-series equation for each deposit category, usingmonthly data between October 1983 and May 1989.In each equation, the authors regress the interestrate paid by the bank on the yield on Treasurysecurities with comparable maturity. If the coeffi-cients on the Treasury security yields are less than 1,Hannan and Liang can reject the hypothesis thatbanks are price takers. For MMDAs, the coefficientis significantly less than 1 for almost all of the banksin the sample. For 2- and 3-year CDs, the coefficientis significantly less than 1 for most of the banks.The mean coefficient for MMDAs is less than, andstatistically different from, the mean coefficientsfor 2- and 3-year CDs. These results are consistentwith the view that banks exercise market power intheir local market areas. It also implies that bankershave greater ability to exercise market power whenpricing MMDAs than when pricing time depositswith 2- or 3-year maturities.

in estimating the effect of market concentrationon loan rates involves holding constant lendingrisk and the effects of other loan terms. Althoughthe studies discussed in this section attempt to dothis, their estimates may still be biased because ofimperfect adjustments for risk and loan terms.

One study on this issue, Cyrnak and Hannan(1999), is discussed above. Hannan (1991a) derivesempirical tests for two of the assumptions of bankingantitrust: first, markets are local; second, perfor-mance is influenced by market concentration. Hetests the hypothesis that markets are local by esti-mating two equations. In one equation, the interestrates on business loans are determined in a nationalmarket; in a second equation, the interest rates onbusiness loans are assumed to vary among metro-politan areas. The equation that incorporates localeffects has more explanatory power for businessloans than the equation that is based on a nationalmarket. Hannan also finds a positive associationbetween the interest rates that banks charge on rel-atively small business loans and the concentrationof deposits in the local banking market areas. Thus,he finds empirical evidence to support the twobanking antitrust assumptions.

The objective of Hannan and Liang (1995) is toinvestigate the weight that deposits at thrift institu-tions should receive when calculating measures ofmarket concentration. The dependent variable isthe interest rate on business loans. Independentvariables include market HHI, calculated undervarious assumptions about the weights assigned todeposits at thrifts; other measures of market struc-ture; and other variables that reflect characteristicsof the loans, the lending banks, and the market areaswhere the banks have their offices. In the regressionswith zero weights for thrift institutions, the coeffi-cients on HHI are positive and statistically significant.Assigning positive weights to thrift deposits whencalculating market HHI makes the fit of the regres-sion equations worse. These results do not supportinclusion of thrift institutions in the HHI calculationfor purposes of banking antitrust. Because routineantitrust analyses focus on deposits and not loansto small businesses, however, this empirical testdoes not necessarily rule out a positive weight forthrift deposits in those routine analyses.

Hannan (1997) uses interest rate data on indi-vidual business loans as the measure of bank performance. As independent variables, he usesHHI and other measures of market structure, alongwith other variables that reflect the characteristics

In the next step in their analysis, Hannan andLiang investigate whether the degree of marketconcentration affects the size of the coefficientsdescribed above. They find that when MMDA interestrates are the dependent variable, the coefficientson market interest rates tend to be lower amongbanks located in market areas with higher concen-tration. For interest rates on 2- or 3-year CDs, thereis not a statistically significant correlation betweenthe coefficients on market interest rates and bankingmarket concentration. These results imply that thebanks operating in more concentrated local marketsexercise greater market power when pricing MMDAs,but, because competition for CDs occurs on abroader geographic scale, local market concentra-tion does not influence their rates. The results inHannan and Liang (1993) are consistent with theassumptions that underlie banking antitrust.17

Kahn, Pennacchi, and Sopranzetti (1999), takinga different approach, develop a theory of how banksset the interest rates they pay on retail deposits (indenominations of less than $100,000) that is basedon customers’ limited ability to recall numbers whencomparison shopping. Because customers areassumed to recall only a limited number of digits,banks tend to quote deposit interest rates either inwhole integers, such as 3 percent or 4 percent, orat a limited number of points between whole inte-gers. The authors find empirical support for theirtheory. Deposit interest rates were more likely toremain unchanged after changes in wholesalemarket interest rates if banks had initially set thedeposit rates at whole integers. In addition, banksin local markets with higher concentration were lesslikely to respond to changes in wholesale marketinterest rates and more likely to set deposit interestrates at integers or quarter-integers. The results ofthe articles cited in this section are consistent withthe assumptions that underlie banking antitrust.

Effects of Local Market Concentration onInterest Rates Charged on Bank Loans. The SCPhypothesis implies that banks in market areas withhigher concentration will tend to charge higherinterest rates on loans.18 An important challenge



17 Jackson (1997) discusses reasons why the relationship between marketconcentration and price rigidity may be nonlinear and presents evi-dence of such a nonlinear relationship for banking markets.

18 In a theoretical model that incorporates asymmetric information aboutborrowers’ likelihood of repaying a loan, Shaffer (2002) examinesthe relationship between the interest rates on bank loans and marketconcentration. He finds that there are conditions under which the SCPhypothesis is valid even if the empirical results indicate no associationbetween measures of interest rates on loans and measures of marketstructure.

of the loans, lending banks, and market areas wherethe banks have offices. The regression coefficientson HHI are positive and statistically significant. Thecoefficients on the number of banking organizationsin the market are negative and statistically significant.These results are consistent with giving greaterweight to the number of banking organizations inmarket areas in banking antitrust analyses.

Berger, Rosen, and Udell (2001) expand on theconventional methods of examining how interestrates on loans to small businesses are determinedby including a measure of the size structure of themarket in their analysis. The size structure of themarket represents the distribution of the banks withoffices in the local area by the size of their totalassets in all market areas. The authors posit thatlarge regional or national banking organizationsmay compete differently from small, local institu-tions and that banks’ competitive strategies maydepend on the sizes of banks in a local market.Berger, Rosen, and Udell find several results to sup-port their suppositions: First, small businesses paylower interest rates when large banks dominate amarket; second, size structure primarily affects theprices at large banks; and, third, size structure isstatistically significant only in the markets in whichsmall banks control relatively large shares of marketdeposits. These findings show that market size struc-ture is important to bank pricing behavior and thatbanks compete less aggressively in markets domi-nated by small banks. One implication of this study,which supports the current approach to bankingantitrust, is that local market areas are still relevantfor bank behavior. Another implication, however, isthat market size structure may also be relevant forantitrust analysis, a facet of markets not consideredin current banking antitrust procedures.

Kahn, Pennacchi, and Sopranzetti (forthcoming)examine the association between banking marketconcentration and interest rates on consumer andauto loans. They find that local banking marketstructure does not affect auto loan rates. One expla-nation for this result is that interest rates on car loansare now predominantly determined by financingavailable through the auto manufacturers. Theresults are different for the interest rates on con-sumer loans. These interest rates tend to be higherat banks in more concentrated market areas. More-over, interest rates on consumer loans tend torespond asymmetrically to changes in marketinterest rates. Banks increase consumer loan ratesin tandem with market rates as they rise, but they



reduce consumer loan rates more slowly than mar-ket rates as they fall. This asymmetric response ismore pronounced for banks in more highly concen-trated markets.

The results from studies cited in this sectionare consistent with some of the assumptions thatunderlie banking antitrust: Markets for bankingservices are local and local banks are the relevantcompetitors. Results in Hannan (1997), however,indicate that the number of banking organizationswith offices in a market area should also receivesome weight in antitrust analysis. Berger, Rosen,and Udell (2001) indicate that, in addition to localmarket concentration, the size structure of localmarket areas may affect the degree of competitionamong banks in those markets.

Monopoly Profits and the “Quiet Life.” Ifrelatively high market concentration facilitatescollusion among banks, then banks located inmore highly concentrated markets should faceless pressure to minimize their operating costs.Such an inverse relationship between cost efficiencyand market concentration is a version of the “quietlife” hypothesis, which is based on a statement byJohn Hicks that “the best of all monopoly profitsis a quiet life” (Hicks, 1935, p. 8). Analysis of thequiet life hypothesis has a long history in the litera-ture on banking antitrust (Rhoades and Rutz, 1982;and Edwards and Heggestad, 1973). Berger andHannan (1998) test the hypothesis using data fromthe 1980s and find evidence that banks in morehighly concentrated markets exhibit poorer costefficiency than do other banks, all else equal. Infact, the authors show that if the concentrationlevel in the most concentrated market were reducedto the minimum level observed in their sample,operating costs at banks in that market would beexpected to decline between 8 percent and 32 per-cent. In addition, they conclude that the size of theadditional operating cost due to market concentra-tion (efficiency loss) is several times larger than thesize of the additional revenues due to noncom-petitive pricing of banking services (welfare loss).

Bergstresser (2001a) tests another version ofthe quiet life hypothesis: Banks with greater marketpower tend to assume less risk than other banks.He tests this version of the hypothesis by examiningthe association between the percentage of bankloans in a high-risk category (construction andland development loans) and local market concen-tration. The data are derived from bank call reportsfor the years 1980 through 1994. Bergstresser finds

that banks located in market areas with higher con-centration tend to have lower shares of constructionand land development loans in their loan portfo-lios. These results are consistent with the currentapproach to banking antitrust.

Other Evidence Relevant for BankingAntitrust

Effects of Local Banking Concentration onthe Decision To Join a Credit Union. Emmonsand Schmid (2000) examine how local bankingmarket concentration influences the decisions ofresidents to join credit unions. They measure creditunion participation as the percentage of potentialmembers who actually choose to become members.They find a positive association between this mea-sure of credit union participation in the currentperiod and the lagged value of concentration inthe banking markets where the credit unions arelocated. Their results can be interpreted as evidencethat banks in areas of relatively high concentrationoffer their banking services at relatively unattractiveterms, such as high interest rates on loans, lowinterest rates on deposits, or poor service. Inresponse, a relatively high percentage of eligiblehouseholds in these areas join credit unions.

Effects of Competition from Credit Unionsfor the Pricing of Bank Services. Three recentstudies find that credit unions influence the interestrates that banks charge on loans and pay ondeposits. Tokle and Tokle (2000), using interest ratedata on small-denomination deposit accountsgathered from telephone surveys of banks in Idahoand Montana, estimate the effect local credit unionshave on deposit interest rates at local banks. Theirresults indicate that the share of market depositsat credit unions has positive effects on depositinterest rates at local banks. Feinberg (2001), usingBank Rate Monitor data, finds that the share ofmarket deposits at credit unions has a negative andstatistically significant effect on new car loan ratesat local banks. Hannan (2002), also using BankRate Monitor data, finds that various measures ofcredit union market penetration have positive andstatistically significant effects on the interest ratesthat banks pay for deposits. The results of thesethree studies provide empirical support for thecurrent practice of defining markets as local. Theresults raise questions, however, about the currentpractice of rarely including credit union depositsin local market concentration calculations.

Effects of Market Concentration on EconomicGrowth. Collender and Shaffer (2003) examinehow various measures of local banking marketstructure, including market HHI, affect local econ-omic growth. They measure economic growth asthe change in per capita personal income at thecounty level for rural counties and at the MSA levelfor urban counties. They find that economic growthis slower in urban markets with higher concentra-tions of deposits. In contrast, they find that localmarket concentration has no effect on economicgrowth in rural areas.

Demonstrating that a bank merger would leadto slower economic growth in the local market isnot necessary to establish a violation of antitruststandards. That said, the results in Collender andShaffer have implications for the delineation ofbanking market areas in antitrust analysis: Hadfinancial innovations and regulatory changes madelocal markets irrelevant for antitrust analysis, theauthors would not have found any associationbetween local economic growth and measures oflocal banking structure.

Local Market Concentration and CreditAvailability to Borrowers without CreditHistories. Petersen and Rajan (1995), starting fromthe assumption that lenders have less informationabout new firms than about more established firms,develop a theory of the supply of bank credit tonew firms. In this theory, Petersen and Rajan positthat credit availability to new firms depends on thedegree of bank competition in local market areas.Banks located in areas with relatively limited com-petition know that, if they lend to new firms andthe firms are successful, they are likely to keepthese firms as customers in the future. In otherwords, banks that face limited competition viewloans to new firms as risky investments that mayyield long-term profits if the new firms survive.Banks located in areas with more intense compe-tition, on the other hand, know that competitorseventually will bid away any long-term profits thatbanks might expect to gain from relationships withnew firms that are successful. These banks, then,do not have reason to view risky loans to new firmsas potentially yielding long-term profits.

Petersen and Rajan use data from the 1987National Survey of Small Business Finances todevelop an empirical test for their theory of creditavailability to new firms. The following quotationpresents the conclusions of their study:



Young firms in concentrated markets receivemore institutional finance than do similarfirms in competitive markets. As firms getold, the difference in the relative firms bor-rowing from institutions disappears. Youngfirms who get institutional loans are moreindebted in concentrated markets than incompetitive markets, but this patternreverses for older firms. Creditors seem tosmooth interest rates over the life cycle ofthe firm in a concentrated market, charginga lower-than-competitive one when the firmis young and a higher-than-competitive ratewhen the firm is old. (p. 439)

Petersen and Rajan’s results are consistent with theview that local market areas are relevant for bankingantitrust.19

Bergstresser (2001b) applies the Petersen andRajan (1995) framework to consumer lending. Heposits that consumers without credit historieslocated in more concentrated banking marketsshould face fewer constraints when obtaining creditthan similar consumers located in less concentratedmarkets. This theory, similar to that for new firms,is based on the idea that, in less competitive (moreconcentrated) areas, banks view loans to consumerswith no credit histories as risky investments thatare likely to yield long-run profits if these consumerspay their debts. In the more competitive (less concen-trated) markets, borrowing constraints are higherfor consumers without credit histories because, oncethese customers have established credit histories,competing banks will bid interest rates on theirloans down to the competitive level.

To test this theory, Bergstresser estimates twoequations using data from the 1983 Survey ofConsumer Finances, which is the last of these sur-veys for which the MSA of each respondent is pub-licly available. In each equation, the observationsare for individual households. In the first equation,the dependent variable is a dummy variable with avalue of 1 if the household reports that it receivedless credit from banks than it requested, zero other-wise. One of the independent variables is the HHI



of the MSA where the household resides. In variousspecifications, the coefficients on HHI are negativeand statistically significant. That is, living in an areawith relatively high banking concentration reducedthe odds that consumers reported receiving lesscredit than requested.

In Bergstresser’s second equation, the depen-dent variable is the interest rate that the householdpaid on a bank loan. Independent variables includethe age of the head of the household and othercontrol variables. This equation is estimated sepa-rately for households in areas with HHI above andbelow a threshold for relatively high concentration.In areas of high and low concentration, interest ratestend to decline with age, but the decline is smallerin the high concentration markets than in the lowconcentration markets. The difference in the ratesof decline because of age between these two equa-tions is statistically significant. The empirical resultsfor both equations are consistent with the theorydeveloped by Petersen and Rajan (1995). The resultsin Bergstresser (2001b) are consistent with the viewthat local market areas are relevant for bankingantitrust.

The relevance of the Petersen and Rajan (1995)and Bergstresser (2001b) studies for banking antitrustanalysis is limited, however, because both studiesuse data from the 1980s. Financial innovation andthe relaxation of branching restrictions may havereduced the ability of banks in markets with rela-tively high concentration to derive long-term econ-omic profits from relationships with new firms orwith households without credit histories. Evanoffand Fortier (1988) emphasize that branching restric-tions alter the nature of the relationship betweenlocal market concentration and bank performance.

Zarutskie (2003) extends the analysis of bankmarket structure and the availability of credit tonew businesses into the 1990s by using income tax data for small U.S. corporations for the years1987 through 1998. She finds that local bankingmarket structure affects the likelihood a new firmwill borrow from banks in a manner consistentwith Petersen and Rajan’s (1995) theory: In moreconcentrated markets, young firms are more likelyto borrow from banks than from owners’ savings.She also finds, though, that the effects of localmarket structure are weaker after 1995, the periodwhen nationwide banking was permitted underfederal legislation.

19 The discussion of the following studies focuses on those that test thePetersen and Rajan (1995) theory empirically by using local marketareas in the United States as the relevant market areas. Several otherstudies, some using international rather than just U.S. data, have alsotested this theory, including Cetorelli and Gambera (2001) and Beck,Demirgüç-Kunt, and Maksimovic (forthcoming). Cetorelli (2003) usesdata for regions in the United States, and Bonaccorsi and Dell’Aricca(forthcoming) use data for local market areas in Italy.

published in recent years have found evidence sup-porting a number of the assumptions that underliebanking antitrust. The studies also show, however,that the effects of financial innovations and changesin bank regulation are starting to call into questionsome of the current practices in banking antitrustanalysis.

REFERENCES

Akhigbe, Aigbe and McNulty, James E. “The Profit Efficiencyof Small U.S. Commercial Banks.” Journal of Banking andFinance, February 2003, 27(2), pp. 307-25.

Amel, Dean F. and Liang, J. Nellie. “Determinants of Entryand Profits in Local Banking Markets.” Review of IndustrialOrganization, February 1997, 12(1), pp. 59-78.

___________ and Starr-McCluer, Martha. “Market Definitionin Banking: Recent Evidence.” Antitrust Bulletin, Spring2002, 47(1), pp. 63-89.

Austin, Douglas V. and Bernard, Craig D. “The CompetitiveAnalysis in the Banking Industry No Longer ReflectsReality.” Proceedings of the 37th Annual Conference onBank Structure and Competition, The Financial SafetyNet: Costs, Benefits, and Implications for Regulation,Federal Reserve Bank of Chicago, May 2001, pp. 624-47.

Beck, Thorsten; Demirgüç-Kunt, Asli and Maksimovic,Vojislav. “Bank Competition and Access to Finance.”Journal of Money, Credit, and Banking (forthcoming).

Berger, Allen N. “The Profit-Structure Relationship inBanking—Tests of Market-Power and Efficient-StructureHypotheses.” Journal of Money, Credit, and Banking, May1995, 27(2), pp. 404-31.

___________ and Hannan, Timothy H. “The Price-Concentration Relationship in Banking.”Review ofEconomics and Statistics, May 1989, 71(2), pp. 291-99.

___________ and ___________. “The Efficiency Cost ofMarket Power in the Banking Industry: A Test of the‘Quite Life’ and Related Hypotheses.” Review of Economicsand Statistics, August 1998, 80(3), pp. 454-65.

___________; Rosen, Richard J. and Udell, Gregory F. “The Effect of Market Size Structure on Competition:The Case of Small Business Lending.” Working Paper2001-10, Federal Reserve Bank of Chicago, October 2001.

Bergstresser, Daniel. “Market Concentration and Loan

CONCLUSIONS

Antitrust analysis of bank mergers and acquisi-tions is based on assumptions about the geographicscope of banking markets, the nature of the productor products sold in banking markets, and the relevantcompetitors within the market areas. The marketareas for banking services are assumed to be local:generally, metropolitan areas for cases involvingbanks in urban areas and counties for cases involvingbanks in rural areas. The relevant product is a clusterof financial services demanded from commercialbanks, rather than individual financial services, suchas various types of deposit accounts or types of loans.The relevant competitors are banks and thrifts withoffices located in the same market area.

The studies surveyed in this article provide evi-dence that is consistent with some of the assump-tions that underlie banking antitrust. The findingsof these studies are consistent with the view thatthe relevant market areas for banking antitrust arelocal communities. In addition, the studies con-tinue to yield evidence that banks located in marketareas with relatively low concentration tend to offertheir financial services at terms that reflect greatercompetition.

Several of the surveyed studies also provideevidence that is not consistent with the current prac-tice of banking antitrust. One such study presentsevidence that concentration at both the local andstate levels affects the pricing of banking services.A study based on recent data indicates that bankingconsolidation is diminishing the effect of localmarket concentration on deposit interest rates.Other studies find evidence that local market con-centration is not the only measure of market struc-ture that affects the pricing of banking services. Thesizes of the organizations with offices in local areasaffect the pricing of banking services, and the degreeto which these institutions compete with eachother in other markets affects bank profits. In addi-tion, evidence that large banks are lending to smallbusinesses located in areas where the banks do nothave offices raises questions about the assumptionthat local banks are the relevant competitors inbanking antitrust analysis. Finally, several studiesfind evidence that the presence of credit unions inmarket areas affects how banks price financial ser-vices. This evidence brings into question the currentpractice of rarely including credit union deposits inmarket concentration calculations.

In conclusion, many of the studies written or



Portfolios in Commercial Banking.” Unpublishedmanuscript, MIT Department of Economics, November2001a.

___________. “Banking Market Concentration and ConsumerCredit Constraints: Evidence from the Survey of ConsumerFinances.” Unpublished manuscript, MIT Department ofEconomics, November 2001b.

Bonaccorsi Di Patti, Emila and Dell’Aricca, Giovanni. “BankCompetition and Firm Creation.” Journal of Money,Credit, and Banking (forthcoming).

Calem, Paul S. and Carlino, Gerald A.“The Concentration/Conduct Relationship in Bank Deposit Markets.” Reviewof Economics and Statistics, May 1991, 73(2), pp. 268-76.

Carlton, Dennis W. and Perloff, Jeffrey M. Modern IndustrialOrganization. Third Edition. Reading, MA: Addison WesleyLongman, 2000.

Cetorelli, Nicola. “Life-Cycle Dynamics in Industrial Sectors:The Role of Banking Market Structure.” Federal ReserveBank of St. Louis Review, July/August 2003, 85(4), pp.135-48.

___________ and Gambera, Michele. “Banking MarketStructure, Financial Dependence and Growth:International Evidence from Industry Data.” Journal ofFinance, April 2001, 56(2), pp. 617-48.

Collender, Robert N. and Shaffer, Sherrill. “Local BankOffice Ownership, Deposit Control, Market Structure,and Economic Growth.” Journal of Banking and Finance,January 2003, 27(1), pp. 27-57.

Cyrnak, Anthony W. “Bank Merger Policy and the NewCRA Data.” Federal Reserve Bulletin, September 1998,84(9), pp. 703-15.

___________ and Hannan, Timothy H. “Is the Cluster StillValid in Defining Banking Markets? Evidence from a NewData Source.” Antitrust Bulletin, Summer 1999, 44(2),pp. 313-31.

DeYoung, Robert; Hunter, William C. and Udell, Gregory F.“The Past, Present and Probable Future for CommunityBanks.” Presented at the conference Whither theCommunity Bank? Federal Reserve Bank of Chicago, 13-14 March 2003.

Edwards, Franklin R. and Heggestad, Arnold A. “Uncertainty,



Market Structure, and Performance in Banking: theGalbraith-Caves Hypothesis and Managerial Motives inBanking.”Quarterly Journal of Economics, August 1973,87(3), pp. 455-73.

Emmons, William R. and Schmid, Frank A. “BankCompetition and Concentration: Do Credit Unions Matter?”Federal Reserve Bank of St. Louis Review, May/June 2000,82(3), pp. 29-42.

Evanoff, Douglas D. and Fortier, Diana L. “Reevaluation ofthe Structure-Conduct-Performance Paradigm in Banking.”Journal of Financial Services Research, June 1988, 1(3),pp. 277-94.

Feinberg, Robert M. “The Competitive Role of Credit Unionsin Small Financial Services Markets.” Review of Economicsand Statistics, August 2001, 83(3), pp. 560-63.

Frame, W. Scott and Kamerschen, David R. “The Profit-Structure Relationship in Legally Protected BankingMarkets Using Efficiency Measures.” Review of IndustrialOrganization, February 1997, 12(1), pp. 9-22.

Gilbert, R. Alton. “Bank Market Structure and Competition:A Survey.” Journal of Money, Credit, and Banking,November 1984, 16(4, part 2), pp. 617-45.

Hannan, Timothy H. “Bank Commercial Loan Markets andthe Role of Market Structure: Evidence from Surveys ofCommercial Lending.” Journal of Banking and Finance,February 1991a, 15(1), pp. 133-49.

___________. “Foundations of the Structure-Conduct-Performance Paradigm in Banking.” Journal of Money,Credit, and Banking, February 1991b, 23(1), pp. 68-84.

___________. “Market Share Inequality, the Number ofCompetitors, and the HHI: An Examination of BankPricing.” Review of Industrial Organization, February1997, 12(1), pp. 23-35.

___________. “The Impact of Credit Unions on the RatesOffered for Retail Deposits by Banks and ThriftInstitutions.” Finance and Economics Discussion Series2003-6, Board of Governors of the Federal ReserveSystem, 10 September 2002.

___________. “Changes in Non-Local Lending to SmallBusinesses.” Unpublished manuscript, Board of Governorsof the Federal Reserve System, 10 January 2003.

___________; ___________ and ___________. “BankConsolidation and the Dynamics of Consumer LoanInterest Rates.” Journal of Business (forthcoming).

Kwast, Myron L.; Starr-McCluer, Martha and Wolken, John D.“Market Definition and the Analysis of Antitrust inBanking.” Antitrust Bulletin, Winter 1997, 42(4), pp.973-95.

Moore, Robert R. “Concentration, Technology, and MarketPower in Banking: Is Distance Dead?” Federal ReserveBank of Dallas Financial Industry Studies, December1998, pp. 1-10.

Neumark, David and Sharpe, Steven A. “Market Structureand the Nature of Price Rigidity: Evidence from theMarket for Consumer Deposits.” Quarterly Journal ofEconomics, May 1992, 107(2), pp. 657-80.

Petersen, Mitchell A. and Rajan, Raghuram G. “The Effectof Credit Market Competition on Lending Relationships.”Quarterly Journal of Economics, May 1995, 110(2), pp. 407-43.

___________ and ___________. “Does Distance Still Matter?The Information Revolution in Small Business Lending.”Journal of Finance, December 2002, 57(6), pp. 2533-70.

Pilloff, Steven J. “Multimarket Contact in Banking.” Reviewof Industrial Organization, March 1999a, 14(2), pp. 163-82.

___________. “Does the Presence of Big Banks InfluenceCompetition in Local Markets?” Journal of FinancialServices Research, May 1999b, 15(3), pp. 159-77.

___________. “Money Market Mutual Funds: Are They aClose Substitute for Accounts in Insured DepositoryInstitutions?” Antitrust Bulletin, Summer 1999c, 44(2),pp. 365-85.

___________. “What’s Happened at Divested Bank Offices?An Empirical Analysis of Divestitures in Bank Mergers?”Finance and Economics Discussion Series 2002-60, Boardof Governors of the Federal Reserve System, December2002.

___________ and Rhoades, Stephen A. “Structure andProfitability in Banking Markets.” Review of IndustrialOrganization, February 2002, 20(1), pp. 81-98.

Prager, Robin A. and Hannan, Timothy H. “Do SubstantialHorizontal Mergers Generate Significant Price Effects?

___________ and Berger, Allen N. “The Rigidity of Prices:Evidence from the Banking Industry.” American EconomicReview, September 1991, 81(4) pp. 938-45.

___________ and Liang, J. Nellie. “Inferring Market Powerfrom Time-Series Data: The Case of the Banking Firm.”International Journal of Industrial Organization, June 1993,11(2), pp. 205-18.

___________ and ___________. “The Influence of ThriftCompetition on Bank Business Loan Rates.” Journal ofFinancial Services Research, June 1995, 9(2), pp. 107-22.

___________ and Prager, Robin A. “The CompetitiveImplications of Multimarket Bank Branching.” Unpublishedmanuscript, Board of Governors of the Federal ReserveSystem, 27 January 2003.

Heitfield, Erik A. “What Do Interest Rate Data Say aboutthe Geography of Retail Banking Markets?” AntitrustBulletin, Summer 1999, 44(2), pp. 333-47.

___________ and Prager, Robin A. “The Geographic Scopeof Retail Deposit Markets.” Finance and EconomicsDiscussion Series 2002-49, Board of Governors of theFederal Reserve System, 13 August 2002.

Hicks, John R. “Annual Survey of Economic Theory: TheTheory of Monopoly.” Econometrica, January 1935, 3(1),pp. 1-20.

Holder, Christopher L. “The Use of Mitigating Factors inBank Mergers and Acquisitions: A Decade of Antitrust atthe Fed.” Federal Reserve Bank of Atlanta Economic Review,March/April 1993, 78(2), pp. 32-44.

Jackson, William E. III. “Is the Market Well Defined in BankMerger and Acquisition Analysis?” Review of Economicsand Statistics, November 1992, 74(4), pp. 655-61.

___________. “Market Structure and the Speed of PriceAdjustments: Evidence of Non-Monotonicity.” Review ofIndustrial Organization, February 1997, 12(1), pp. 37-57.

___________ and Eisenbeis, Robert A. “Geographic Integrationof Bank Deposit Markets and Restrictions on InterstateBanking: A Cointegration Approach.” Journal of Economicsand Business, July/August 1997, 49(4), pp. 335-46.

Kahn, Charles M.; Pennacchi, George G. and Sopranzetti,Ben J. “Bank Deposit Rate Clustering: Theory andEmpirical Evidence.” Journal of Finance, December 1999,54(6), pp. 2185-214.



Evidence from the Banking Industry.” Journal of IndustrialEconomics, December 1998, 46(4), pp. 433-52.

Radecki, Lawrence J. “The Expanding Geographic Reach ofRetail Banking Markets.” Federal Reserve Bank of NewYork Economic Policy Review, June 1998, 4(2), pp. 15-34.

___________. “Competition in Shifting Product andGeographic Markets.” Antitrust Bulletin, Fall 2000, 45(3),pp. 571-613.

Rhoades, Stephen A. “Structure-Performance Studies inBanking: An Updated Summary and Evaluation.” FederalReserve Bulletin, August 1982, 68(8), pp. 477-78.

___________. “Market Share Inequality, the HHI, and OtherMeasures of the Firm-Composition of a Market.” Reviewof Industrial Organization, December 1995, 10(6), pp.657-74.

___________. “Competition and Bank Mergers: Directionsfor Analysis from Available Evidence.” Antitrust Bulletin,Summer 1996, 41(2), pp. 339-64.

___________ and Rutz, Roger D. “Market Power and FirmRisk: A Test of the ‘Quiet Life’ Hypothesis.” Journal ofMonetary Economics, January 1982, 9(1), pp. 73-85.

Santomero, Anthony M. “Bank Mergers: What’s aPolicymaker To Do?” Journal of Banking and Finance,February 1999, 23(2-4), pp. 637-43.

Shaffer, Sherrill. “Competitive Bank Pricing and AdverseSelection, with Implications for Testing the SCPHypothesis.” Quarterly Review of Economics and Finance,Summer 2002, 42(3), pp. 633-47.

Sharpe, Steven A. “The Effect of Consumer SwitchingCosts on Prices: A Theory and Its Application to theBank Deposit Market.” Review of Industrial Organization,February 1997, 12(1), pp. 79-94.



Shull, Bernard. “The Origins of Antitrust in Banking: AnHistorical Perspective.” Antitrust Bulletin, Summer 1996,41(2), pp. 255-88.

____________ and Hanweck, Gerald A. Bank Mergers in aDeregulated Environment: Promise and Peril. Westport,CT: Quorum Books, 2001.

Simons, Katerina and Stavins, Joanna. “Has Antitrust Policyin Banking Become Obsolete?” Federal Reserve Bank ofBoston New England Economic Review, March/April 1998,pp. 13-26.

Tokle, Robert J. and Tokle, Joanne G. “The Influence ofCredit Union and Savings and Loan Competition onBank Deposit Rates in Idaho and Montana.” Review ofIndustrial Organization, December 2000, 17(4), pp. 427-39.

Webb, Robert L. “Divestiture: A Prescription for HealthyCompetition.” Federal Reserve Bank of St. Louis RegionalEconomist, January 2001, pp. 10-11.

Weiss, Leonard W. “A Review of Concentration-Price Studiesin Banking,” in Leonard W. Weiss, ed., Concentration andPrice. Cambridge, MA: MIT Press, 1989, pp. 219-54.

Wolken, John D. and Rohde, Douglas. “Changes in theLocation of Small Businesses’ Financial Service Suppliersbetween 1993 and 1998.” Unpublished manuscript,Board of Governors of the Federal Reserve System, 28January 2002.

Woosley, Lynn W.; King, B. Frank and Padhi, Michael S. “IsCommercial Banking a Distinct Line of Commerce?”Federal Reserve Bank of Atlanta Economic Review, FourthQuarter 2000, 85(4), pp. 39-58.

Zarutskie, Rebecca. “Does Bank Competition Affect HowMuch Firms Can Borrow? New Evidence from the U.S.”Presented at the 39th Annual Conference on BankStructure and Competition: Corporate Governance:Implications for Financial Services Firms, Federal ReserveBank of Chicago, May 2003.

Date post:	22-Sep-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Banking Antitrust: Are the Assumptions Still Valid? · 2019. 10. 9. · As such, credit unions do...

Documents