cra and fair lending regulations: resulting trends in .../media/publications/... · additional...

FEDERAL RESERVE BANK OF CHICAGO 19

CRA and fair lending regulations:Resulting trends in mortgage lending

In response to concerns thatbanks were not adequatelyserving the credit needs oftheir local communities andnot treating all applicants fair-

ly, during the 1960s and 1970s Congress passedthe fair lending laws and the Community Rein-vestment Act (CRA).1 These laws, aimed ateliminating discriminatory lending practices andencouraging lending to low-income individualsand in low-income areas, have been controver-sial since their inception. Community advocatesargued that the acts were either inadequate orinadequately enforced and that banks continuedto channel deposits away from local communi-ties, resulting in inadequate financing for theareas most in need. Bankers argued that theytreated applicants fairly and the acts smackedof credit allocation that could adversely affectbank safety and soundness.

Although there continues to be significantdisagreement regarding these regulations, re-cently there has been a wave of positive reviewsof their effectiveness.2 The regulations havebeen given credit for encouraging banks toimplement special loan programs aimed at lower-income communities and for effectively chan-neling funds toward previously underservedareas and minority groups. Community advo-cates argue that significant progress has beenmade and continued enforcement will reapadditional benefits. Some bankers state that inresponding to the CRA they have discoverednew, profitable, previously untapped lendingopportunities. These opportunities have come

Douglas D. Evanoff and Lewis M. Segal

at a most convenient time as the demand intraditional lending markets has slowed.

While the arguments for the fair lendinglaws and the CRA are essentially ones of equi-ty, there may also be economic arguments forconstraining private market behavior and chan-neling funds to underserved areas. It may bethat these credit flows produce positive exter-nalities which, from a societal perspective,generate a total return greater than that receivedby the providers of the credit.3 That is, althoughsociety reaps the full benefits of providing thiscredit, the service provider (a bank in this case)may not. While this provides economic justifi-cation for channeling credit to particular markets,it does not necessarily warrant doing so throughthe banking system.

In this article, we examine the evolution ofthe fair lending regulations and the CRA. Wethen summarize the economic literature thatpertains to these regulations. Finally, we evaluatethe effectiveness of the fair lending laws and theCRA by analyzing recent trends in mortgagelending activity and discussing whether thesetrends are in line with the intent of the regulations.We ask whether the trends can be attributed to theregulations and whether the data suggest that theregulations have been successful.

Douglas D. Evanoff and Lewis M. Segal are econ-omists at the Federal Reserve Bank of Chicago.They would like to thank John Bergstrom, RaphaelBostic, Paul Calem, Glenn Canner, and LorrieWoos for helpful comments on earlier drafts, andPat Dykes for her generous data support. Theyalso acknowledge the technical assistance ofJonathan Siegel.

ECONOMIC PERSPECTIVES20

Evolution of the CRA and the fairlending laws

Although it is common to group togetherthe Fair Housing Act, the Equal Credit Oppor-tunity Act (ECOA), and the CRA, they aremore accurately classified into two groups: thefair lending laws and the CRA. The fair lend-ing laws are aimed at eliminating lending dis-crimination based on the inherent attributes ofthe borrower, such as race or gender. The CRAprimarily addresses geographic discrimina-tion, that is, failing to serve the credit needsof the local community in which the bank waschartered. The Home Mortgage DisclosureAct (HMDA) provides information on lend-ing to individuals and locations, supportingthe enforcement of both the fair lending lawsand the CRA.

Fair lending lawsThe Fair Housing Act was approved by

Congress in 1968 as part of the Civil RightsAct of that year. It prohibits discrimination inresidential real estate transactions based onrace or color, religion, national origin, gender,handicap, or family status.4 The ECOA encom-passes a broader array of transactions. Passedin 1974, it prohibits discrimination with regardto any aspect of a credit transaction (consumer,commercial, or real estate loan) based on raceor color, religion, ethnic origin, gender, maritalstatus, age, and receipt of public assistance.5

It has been argued that fair lending enforce-ment prior to the 1990s was generally unaggres-sive.6 The techniques employed to detect discrimi-nation (reviewing whether internal policies werefollowed and performed uniformly across theprotected factors) typically detected only themost blatant cases of discrimination. Since thattime, in response to growing public concernabout lending discrimination and well-publicizedresearch that reported evidence of discrimination,regulatory agencies and the U.S. Department ofJustice have stepped up their enforcement efforts.For example, a 1988 study of mortgage discrim-ination in Atlanta led the Justice Department toinitiate an investigation into fair lending practic-es by depositories in that market.7 The investi-gation resulted in the first major lawsuit filed bythe department against an institution for vio-lating fair lending laws.8 This is in sharpcontrast to the number of suits filed for civilrights violations in other areas, for example,

housing and employment. Congress alsoresponded to repeated claims of lending dis-crimination by amending the Fair Housing Actin 1988 to allow private parties to originatemortgage discrimination lawsuits more easily.The ECOA was amended in 1991 to requirebank regulators to refer cases to the Departmentof Justice instead of handling them independently,sending a signal that the department was goingto be more aggressive in the prosecution of suchcases. Perhaps most significantly, the 1975HMDA was amended in 1988, 1989, and 1991to develop a database that would provide regula-tors and the public with data to analyze deposi-tory institution lending patterns.

As originally enacted, HMDA requireddepository institutions and their subsidiaries toprovide the total number and dollar value ofmortgages originated and purchased in thelocal market, typically segmented by censustract. The 1989 amendment required lenders toreport information at the loan application levelregarding race, gender, and income, along withdetails on the disposition of the application(deny/accept/withdraw, reason for denial,etc.).9 Banks were required to make the datapublicly available. These expanded data haveenabled regulators to complement their manualreviews of loan files with systematic statisticalanalysis.10 The additional data also allow thepublic to more closely scrutinize lending patternsof depository institutions.

There have also been recent efforts bybank regulators to help depository institutionscomply with fair lending regulation by clarify-ing the compliance requirements. While thepurpose of fair lending laws and regulations isrelatively straightforward, there have beenproblems in implementation, and disagreementshave arisen between regulatory agencies andlenders as to interpretations of the law. Toprovide guidance, a 1994 interagency taskforce representing the federal depository regu-lators released guidelines as to what couldconstitute discriminatory lending practices.11

Under these fair lending guidelines, a lendermay not, because of a prohibited factor:

■ Fail to provide information or services orprovide different information or servicesregarding any aspect of the lending process,including credit availability, applicationprocedures, or lending standards;


■ Discourage or selectively encourage applicantswith respect to inquiries about or applicationsfor credit;

■ Refuse to extend credit or use different stan-dards in determining whether to extend credit;

■ Vary the terms of credit offered, including theamount, interest rate, duration, or type of loan;

■ Use different standards to evaluate collateral;

■ Treat a borrower differently in servicing aloan or invoking default remedies;

■ Use different standards for pooling or pack-aging a loan in the secondary market;

■ Express, orally or in writing, a preferencebased on these prohibited factors, or indicatethat it will treat applicants differently basedon these factors; or

■ Discriminate because of the characteristicsof a person associated with a credit applicantor the prospective occupants of the areawhere property to be financed is located.12

While blatant discrimination may beobvious to most parties, there are times whensound business practices may result in anunintended discriminatory practice against aprotected group. To emphasize to lenders theneed to avoid unintended effects in settingunderwriting criteria, the interagency taskforce also listed the forms of discriminationthat the courts had previously recognized asillegal. These include: overt discrimination—the lender openly discriminates; disparatetreatment—the lender treats applicants differ-ently based on one of the prohibited factors(whether or not it is motivated by prejudiceor intent to discriminate); and disparateimpact—the lender applies a practice uni-formly to all applicants, but the practice has adiscriminatory effect and cannot be justifiedby business necessity.

As a result of the increased scrutiny oflending practices by regulators, there has beena significant increase in the number of ECOAviolations referred to the Department of Justiceby the regulatory agencies and in the numberof suits filed by the department for violationof the fair lending laws. Most of the suits havebeen settled through well-publicized consentagreements, which relayed the message ofstringent enforcement of the fair lending laws.

In evaluating the effect of the fair lendinglaws on mortgage activity, therefore, onewould expect to see more of an impact onlending patterns in the 1990s, as institutionsrespond to increased regulatory pressure.13

The CRAThe major impetus for the 1977 passage of

the CRA was concern by community groupsthat banks and thrifts were not respondingadequately to the credit needs of local commu-nities. Depository institutions were accused ofdiscriminating against individuals based on thecharacteristics of their neighborhood, that is,redlining. This was seen as having a particu-larly adverse impact on minority groups andcontributing to the deterioration of inner-cityneighborhoods. However, the emphasis of theact was on adequately preserving communitiesand not on channeling credit based on race.Community groups argued that it was commonfor banks to reinvest a relatively small portionof deposits generated from local communitiesback into those markets.14

The initial community reinvestment billwas much more intrusive to banks than thefinal act. The initial proposal argued thatbanks were chartered institutions with accessto a government safety net and, as such, had aformal responsibility to perform social func-tions in addition to pursuing the objectives of aprivate enterprise. The proposal defined thebank’s relevant local market from which itreceived deposits and required it to focus onsatisfying credit demands in this market priorto exporting funds to other areas. Banks ar-gued that such behavior would run counter toexisting safety and soundness regulation andconstituted overt credit allocation withoutregard to the credit quality of applicants indifferent geographic areas.

The final act omitted the explicit creditallocation criteria. It required financial institu-tions to serve the convenience and needs of thecommunities in which they were chartered with-out mandating how this was to be accomplished.Additionally, it emphasized the need for bankmanagement to be conscious of communitycredit needs and stressed that this was to bedone without sacrificing safety and soundness.

The mandate of the CRA, to have institu-tions serve the needs of the community in whichthey are chartered, was actually already in place.


The 1935 Banking Act required banks to meetthe convenience and needs of their communities,as did the 1956 Bank Holding Company Act andthe bank charter itself. The fair lending laws,while not explicitly outlawing redlining, addressedsimilar concerns. Finally, while HMDA provid-ed no mechanism for imposing sanctions ondepository institutions, the data were beingcollected precisely for the purpose of monitoringlending patterns and detecting neighborhoodredlining. The real thrust of the CRA was toreemphasize the need for good lending practic-es, to shift the emphasis on reinvestment awayfrom the liability side of the balance sheet (depositgathering) to the asset side (credit generation), andto put the onus squarely upon regulators to monitorthe lending patterns of financial institutions andencourage investment in local communities.

In the early years of the CRA, regulatoryagencies required banks to specify their localcommunity; develop a public statement, includ-ing the local community definition and listingthe type of credit instruments the bank intend-ed to provide; post a list of consumer rightsunder the CRA; and maintain a file of publiccomments for public inspection. These proce-dural requirements were relatively straightfor-ward. In addition, regulators performed anevaluation to “assess the institution’s record ofmeeting the credit needs of the entire communi-ty, including low- and moderate-income neigh-borhoods, consistent with the safe and soundoperation of each institution” (Regulation BB).

To assess the institution’s performance insatisfying this requirement, the regulators devel-oped 12 assessment factors grouped into fiveperformance categories:15

Category A: Ascertainment of communitycredit needs

1. Communication with members of thecommunity to ascertain credit needs; and

2. Extent of involvement by the board ofdirectors in the CRA activities.

Category B: Marketing and types of creditoffered and extended

3. Marketing efforts to make the types ofcredit offered known in the community;

4. The extent of loans originated in the com-munity; and

5. The extent of participation in governmentloan programs.

Category C: Geographic distribution andrecord of opening and closing offices

6. The geographic distribution of credit appli-cations, approvals, and denials; and

7. The record of office openings and closingsand extent of service provided at the offices.

Category D: Discrimination and other illegalcredit practices

8. Practices to discourage credit applica-tions; and

9. Discriminatory or other illegal practices.

Category E: Community development

10. Participation in community developmentprojects or programs;

11. The institution’s ability to meet communitycredit needs; and

12. Other relevant factors which could bearupon the extent to which the institution ishelping to meet the credit needs of thecommunity.

For each of the assessment factors, theexaminer was to assign a grade of 1 (excep-tional) to 5 (significantly inferior), similar tothe CAMEL rating given for safety and sound-ness evaluation. Later, to avoid confusion withsafety and soundness ratings, the CRA ratingwas changed to a four scale grading system:outstanding, satisfactory, needs to improve, orsubstantial noncompliance.

The regulations did not impose explicitsanctions on institutions found not to haveadequately served the needs of their commu-nities. Instead, the regulator was to considerthe CRA rating along with other factors, suchas safety and soundness, when ruling on anapplication for a geographic expansion offacilities through a merger or acquisition,the introduction of new branches, an officechange, etc. However, there are additionalcosts from having a poor CRA rating or beingaccused of poor CRA performance, even ifthe application is ultimately approved. Forexample, the application process can be sig-nificantly lengthened and complicated ifcommunity groups protest the application.In a period in which banks were aggressivelyexpanding geographically, the potential forlost deals, delays in expansion, and negativepublic relations could be quite burdensome.


Following passage of the act, bankersfrequently complained about the vagueness ofthe requirements, including the lack of a spe-cific ranking or weighting scheme for the as-sessment factors to guide the allocation ofresources. Regulatory agencies would periodi-cally issue policy statements providing guidanceto institutions as to how the assessment criteriawere scored and discussing elements of effec-tive CRA programs. Most of these statementsemphasized effort, and the documentation ofsuch effort, instead of performance. In the late1980s, Congress amended the act to have theassessments made public and increased publicscrutiny of banks and regulators.

As with the fair lending laws, enforcementof the CRA intensified in the early 1990s.Denials of merger or acquisition applicationsbased on poor CRA performance became morecommon. Although the ratings were not madepublic prior to 1990, evidence suggests thatregulators have tightened enforcement and havebeen more strict in assigning CRA ratings.16

To stress the commitment to low-income financ-ing, Congress passed the Federal HousingEnterprises Financial Safety and SoundnessAct in 1991. This act put the Federal National

Mortgage Association and the Federal HomeLoan Mortgage Corporation under an affirma-tive obligation to facilitate financing of low-and moderate-income housing. It also estab-lished mortgage purchasing goals for theseagencies relating to low- and moderate-in-come families for affordable housing and forthe central city. Bankers continued to com-plain about the vagueness of CRA require-ments and the resulting regulatory burden.Community groups continued to complainthat banks were inadequately serving thecredit needs of their local communities andthat regulators were inadequately enforcingthe act.

After much public and congressional debate,new CRA regulations were issued in 1995 forimplementation over the following two years.The new regulations stressed performance overeffort in meeting CRA requirements and intro-duced a new evaluation system, replacing theprevious 12 assessment factors with three newtests: lending, investment, and service. For eachtest a bank is assigned one of five grades fromoutstanding to substantial noncompliance.There is also an overall composite rating forCRA compliance.17

The lending test evaluateswhether a bank has a record ofmeeting the credit needs of itslocal community. The regulatorevaluates the number, amount,and distribution across incomegroups and geographic areas ofmortgage, small business, smallfarm, and consumer loans in theassessment area(s) or communi-ties.18 The regulator also consid-ers the innovativeness of the bankin addressing the credit needs oflow- or moderate-income individ-uals or areas and in generatingcommunity development loans.As illustrated in table 1, the lend-ing test carries a disproportionalweight in determining the com-posite rating. A bank cannotreceive a composite rating ofsatisfactory or better unless itreceives a minimum of low satis-factory on the lending test.19

The investment test evaluateshow well a bank satisfies the credit

CRA test ratingsTABLE 1

Component test ratings are assigned to reflectthe bank’s lending, investment, and services.

Componenttest ratings Lending Investment Service

Outstanding 12 6 6

High satisfactory 9 4 4

Low satisfactory 6 3 3

Needs to improve 3 1 1

Substantialnoncompliance 0 0 0

Preliminary composite rating is assigned by summing the threecomponent test ratings and referring to the chart below.

Points Composite assigned rating

20 + Outstanding

11–19 Satisfactory

5–10 Needs to improve

0–4 Substantial noncompliance

Note: Adjustments to preliminary composite rating—no bank may

receive a composite assigned rating of satisfactory or higher unless

it receives at least low satisfactory on the lending test.


needs of its local neighborhoods through quali-fied community investments that benefit theassessment area. Again, the bank’s innovative-ness in responding to community developmentneeds is also taken into account.

Finally, the service test evaluates how wellthe credit needs of the community are met bythe bank’s retail service delivery systems. Thisincludes the distribution of branches across areasserving low- to moderate-income individualsand geographies, as well as alternative deliverysystems, such as ATM, telephone, computer,and mail. The delivery systems and servicesshould be directed at meeting the needs of thelocal community, for example, low-balancechecking accounts and extended lobby hours.Again, the innovativeness of the bank in usingthese alternative delivery systems to serve thelow- and moderate-income individuals andneighborhoods within the community is takeninto consideration.

Although it is too early to determine theeffectiveness of the revisions to the CRA, arecent Government Accounting Office reviewof the new guidelines argued that regulatorsmay face significant challenges in imple-menting the reforms.20 The potential prob-lems are similar to those which existed beforethe reforms, namely:

■ A continued need for excessive documenta-tion of effort and process;

■ Inconsistency in ratings and uncertaintyabout the performance criteria;

■ Incomplete consideration of all relevantmaterial in determining the performance ofthe institution; and

■ Dissatisfaction with regulatory enforcement(depending on one’s perspective, too strin-gent or too lax).

To minimize potential problems, the reportrecommended that significant efforts be madeto provide improved examiner training, improvethe quality of the data used in evaluating perfor-mance, and increase the use of public disclo-sure of the ratings.

Evidence of discrimination inmortgage lending

The CRA was introduced because redliningwas believed to be a common practice by banks.The fair lending laws were passed because there

was a perception that certain borrowing groupswere not being treated equitably. However,there continues to be significant disagreementas to the extent of these problems.

Housing and mortgage discrimination hasbeen a topical issue since the 1960s, whencommunity groups argued that neighborhoodswere deteriorating as a result of practices bymortgage originators. The originators wereaccused of using noneconomic criteria to limitfunding to non-white applicants and/or non-white neighborhoods. Research in this area hasintensified in recent years as amendments toHMDA reporting requirements have increasedthe availability of data used to compare lendingpatterns across race and ethnic groups, incomegroups, and geographic areas. However thedata exclude many of the more relevant vari-ables used in the credit evaluation process.21

The most meaningful studies of the role of raceand neighborhood effects in mortgage lendingincorporate information beyond HMDA dataand evaluate discrimination based either on theneighborhood of the applicant or the character-istics of the individual applicant. These studiesare divided into four classes: neighborhoodredlining studies, application accept/rejectstudies, studies of default rates, and perfor-mance of institutions specializing in loans tolow-income individuals or in low-incomeneighborhoods. Below, we summarize thestudies to emphasize the ongoing controversiesin this area of research.

Redlining studiesRedlining is the practice of having the loan

decision based on, or significantly influencedby, the location of the property without appro-priate regard for the qualifications of the appli-cant or the value of the property. As a result,the neighborhood’s financial needs are notadequately served and the region is unable todevelop economically. Redlining studies typi-cally take the neighborhood as the unit of obser-vation, evaluating whether the aggregate sup-ply of funds made available is related to theracial composition of the area.

Early analysis of differences in loan origi-nations across markets found significant differ-ences based on the racial composition of theneighborhood. However, these studies attribut-ed all market differences to the race variable.22

The findings from a number of recent studies,


which either directly or indirectly addressedthe redlining issue and attempted to explicitlyaccount for market differences, are summa-rized in box A.

Although improvements have been madein redlining studies, inherent methodologicalproblems remain. First, in a number of redlin-ing studies the unit of observation may be toolarge. To the extent redlining occurs, it couldbe for a relatively small area, such as two orthree city blocks. In larger areas, such as met-ropolitan statistical areas (MSA), redliningmay not be detectable in the aggregate data.Additionally, assuming some lenders redlineand others do not, if borrowers eventually findthe non-redlining lender, data at the broaderlevel will imply that no redlining has occurred.The unit of observation should, therefore, berelatively small. There may also be a signifi-cant omitted variable bias. Exclusion of vari-ables correlated with race may produce a sig-nificant coefficient for race even in the absenceof discrimination. A standard criticism ofredlining studies is that they inadequately ac-count for demand factors. Thus, it is impossi-ble to attribute differences in mortgage activityacross markets to an inadequate supply offunding (redlining) or to a lower demand frompotential borrowers. The creditworthiness ofthe applicant pool is also important since theriskiness of the loan will obviously be a deter-mining factor in the underwriting decision.Additional variables to account for differencesin borrower credit demand and creditworthi-ness that have been included in the recent studiesare neighborhood average income, percent ofowner-occupied houses, changes in propertyvalues, poverty and welfare rates, percentof housing units vacant, crime rates, wealthmeasures, mobility rates, average age of pop-ulation and housing stock, total housing units,duration of residency, and the stock of conven-tional mortgages.

Typically, studies that have accounted forthese market characteristics more comprehen-sively have reported a less significant impactof racial composition than that found in earlierstudies. For example, when Holmes and Hor-vitz (1994) excluded measures of risk in theiranalysis of the Houston market, they found thatthe flow of mortgage credit was negativelyassociated with minority status, consistent withredlining. When the risk measure was includ-

ed, minority status was not found to influencethe flow of credit.23 Studies which employ asingle-equation model to explain the amount ofcredit made available in a neighborhood willbe mixing elements of both supply and demandfor credit. Redlining will affect the supplyloans. However, with the single-equationapproach the supply and demand effects cannotbe separated (Yezer, Phillips, and Trost, 1994).Arguing that the race variable represents dis-crimination requires that there be no demand-side effects. As mentioned above, a number ofstudies have shown this to be incorrect. Final-ly, model specification has been shown to drivesome results (Horne, 1997). Concern withmodel specification argues that one should usea relatively flexible financial form which hasthe more commonly used alternative formsnested within it.

Some researchers have argued that theproblems associated with the above credit flowtype of redlining studies are too large to over-come and, as a result, these studies cannotadequately identify the role of racial composi-tion of the neighborhood in loan decisions.An alternative approach, which addresses theproblem of individuals eventually finding thenon-redlining lender, is to directly survey indi-viduals who were active in the mortgage mar-ket. Benston and Horsky (1992, 1979) surveyedhome sellers and buyers to gather informationon credit difficulties encountered in attemptingto sell or purchase homes in several U.S. cities.Instead of viewing only the mortgages approved,the survey gathered information on individualswho requested credit but were unable to obtainit (for reasons such as redlining), in areas inwhich charges of redlining had been made andin control areas. If obtaining credit was a prob-lem, additional information as to the reason forthe problem was obtained—for example, unem-ployment, inadequate down payment, or loca-tion of the house. The survey explicitly askedhome buyers if either a lending institution orreal estate agent had stated or implied thatobtaining a mortgage might be difficult becauseof the neighborhood in which the home waslocated. In both studies, the authors were unableto detect evidence of discrimination or unmetdemand. The bottom line appears to be thatthere is little convincing evidence to suggestthat redlining explains lending patterns in low-income neighborhoods.


Accept/Reject studiesGiven the above criticisms of credit flow

studies, the availability of more detailedHMDA data since 1990, and a desire to moredirectly address the discrimination issue, recentresearch has taken a more microeconomicapproach. Accept/reject studies take individualapplication data and evaluate the determinantsof the lender’s decision. They estimate a prob-ability of rejection function based on variousrisk factors and include a race variable to ac-count for discrimination.24 While these studiescan also be used to test for redlining, theirfocus is on discrimination with respect to indi-vidual applicants. (For a sample of these stud-ies, see box B.)

Prior to the availability of HMDA data,Black et al. (1978) used special survey data todetermine the economic variables important tothe lending decision and whether personal vari-ables such as race played a role. After accountingfor economic variables and terms of the loans,they found that, although the personal charac-teristics did not significantly add to the powerof their model in explaining the accept/denialdecision, race was significant. Black applicantshad a higher probability of denial at the 90 per-cent significance level.

In a well-publicized accept/reject study,Munnell et al. (1992) used HMDA data aug-mented with survey information about thecreditworthiness of borrowers to analyze lend-ing behavior in Boston. A variable to accountfor the racial composition of the market wasnot found to affect the lender’s accept/rejectdecision, but applicant race was found to bestatistically related to the decision. Minoritieswere rejected 56 percent more often thanequally qualified whites.

The Boston study has been criticized for anumber of reasons.25 First, as with the creditflow studies, there is the potential for omittedvariable bias. If omitted variables are associat-ed with the race variable, the coefficient onrace will account for the true effect of race plusthat of the omitted variable(s). The Bostonstudy included several variables to account forborrower risk. However, not all risk factorscould be captured, and some researchers arguethat the race coefficient is actually capturingthe riskiness of the applicant. Race wouldappear significant in an analysis which fails toaccount for wealth if, as has been shown else-where, minorities have lower levels of wealth.

There was also little consideration of the char-acteristics of the property and credit history ofthe applicant. Second, the study has beencriticized for data errors. These potential dataerrors include monthly incomes that are incon-sistent with annual levels, negative interest rates,loan to value ratios exceeding one, loan toincome ratios outside reasonable ranges, theinclusion of black applicant denials because ofover-qualification for special lending programs,and a number of extreme outliers. Brown andTootell (1995) and Munnell et al. (1996) contendthat even after accounting for the data concerns,the fundamental result remains—minorities aremore likely to be denied mortgages than similarlyqualified whites.

Other follow-up studies have shown mixedresults. Using data from Munnell et al. (1992),Zandi (1993) found no race effect, while Carrand Megbolugbe (1993) found the effect remainsafter “cleaning the data,” as did Glennon andStengel (1994). Using a model similar to thatin Munnell et al. to evaluate the Boston andPhiladelphia markets, Schill and Wachter(1993) found evidence consistent with redlin-ing and discrimination. When variables areincluded to proxy for neighborhood risk, theneighborhood racial composition became insig-nificant, although racial status still significant-ly decreased the probability of acceptance.Stengel and Glennon (1995) also found that itis important to use bank-specific guidelines inthe analysis to capture unique, but economicallybased, underwriting criteria. Using a moregeneric market model, for example, secondarymarket criteria, can lead to misleading results.26

Using cleansed data from Munnell et al.(1992), Hunter and Walker (1996) did not findevidence of discrimination via higher underwrit-ing standards for all minorities. They contendthat race matters only in the case of marginallyqualified applications. Needless to say, there islittle uniformity of view.

Yezer (1995) and Rosenblatt (1997) arguethat fundamental problems in the use of accept/reject models to evaluate discrimination resultfrom the informal prescreening of applicants.Both applicants and lenders only want to pro-ceed with applications that appear likely toqualify for a loan because denials are costly forboth parties. Thus, during the initial lender–borrower contact, the lender and borrowerdecide whether the application warrants pursuing.Then, the formal application takes place, and


denials occur only in those cases in whichinformation not available in the initial contactaffects the decision (for example, bad credithistory). Therefore, denial may be as closelyrelated to communication skills and culturalbackground as to economic variables. Sophis-ticated underqualified potential applicants willnot reach the formal application process becausethey realize they will not be accepted, whileunsophisticated candidates will follow throughonly to be denied. Thus, there is a significantselection bias problem in the formal applicationstage which may explain the race differentials.To support this view, Rosenblatt (1997) citesevidence that education levels are strongly pre-dictive of credit approvals. The argument, there-fore, is that the information in denial rate datamay not be what researchers perceive it to be.

Default rate studiesAn alternative means of evaluating lender

discrimination is to examine the default ratesof borrowers thought to be discriminatedagainst relative to other borrowers. Research-ers have compared default rates across groupsbased on the theory that if minorities are overt-ly discriminated against, the average minorityborrower should be of higher credit qualitythan the average nonminority borrower.27 Thisshould be reflected in mortgage default ratesand resulting loss rates; for minority loans,both should be lower. However, studies havenot found evidence of lower default rates forminority holders of mortgages. In critiquingthe Boston study, Becker (1993) cited dataindicating the default rates were equal forwhite and minority sections of the Bostonmarket, which was not consistent with overtdiscrimination.28 A more recent study byBerkovec et al. (1996) also tests for discrimi-nation using default rates. Controlling forvarious loan, borrower, and property relatedcharacteristics, the authors evaluated the defaultrates and resulting losses for FHA-insuredloans and found a higher likelihood of defaulton the part of black borrowers and higher lossrates. These results suggest that lenders, per-haps as a result of regulatory pressure, may haveover-extended credit to minorities.

However, this line of research has alsobeen criticized. First, if discrimination occurs,while the marginal minority borrower may bebetter qualified than the marginal white appli-cant, inferences about the average borrower

cannot be made without making assumptionsabout the distribution of creditworthinessacross the two groups of potential borrowers,for example, Ferguson and Peters (1995).The distributions could be significantly differ-ent. Additionally, minorities may also be treat-ed differently once they are in default. Defaultstudies typically use data on foreclosures.Bank forbearance in defaults favoring one ofthe two groups could bias the results.29

Performance studiesThere are two general areas of research

relating bank performance to the CRA and fairlending regulations. The first deals with theprofitability associated with lending in low-income markets. If such lending is not profit-able, regulation requiring it should adverselyaffect performance. The second area of re-search addresses the implications of mortgagediscrimination on bank performance. If somebanks are choosing to discriminate and foregoprofitable lending opportunities, other banksthat do not discriminate should be able to exploitthese opportunities.

During the debate prior to the enactmentof the CRA, critics argued that economics wasdriving lending patterns and the CRA mighteither have no impact, but be costly to imple-ment, or actually generate bad loans. From thebanks’ perspective it would be a tax and, iflending patterns did not change, it would bewithout benefits. If increased lending in thelow-income market did occur, but was not asprofitable as that in alternative markets, thenthe CRA would act as a tax and a credit redis-tribution mechanism. The argument in favorof the CRA was that banks were foregoingprofitable opportunities because of discriminato-ry behavior or market failure, and performancecould be enhanced if they became more activelyinvolved in this market (although performancecould be adversely affected in the short run asstart-up costs were incurred).

There have been a limited number of studiesevaluating the effect on performance of lendingin the low-income market. Canner and Pass-more (1996) offered a number of testablehypotheses concerning the potential impact onprofitability and the relationship between theextent of the bank’s activity in this market andperformance. They found no evidence of lowerprofitability at banks specializing in the low-income market, consistent with the view that,


once start-up cost are incurred, lending in thismarket can be just as profitable as in othermarkets.30 Beshouri and Glennon (1996) eval-uated the relative performance of credit unionsthat specialize in the low-income market andfound that while these specialized firms havegreater return volatility, higher delinquencyrates, charge-off rates, and operating costs,they are compensated for these differences andgenerate similar rates of return. Similarly, inanalyzing the performance of low-income andminority lending, Malmquist, Phillips-Patrick,and Rossi (1997) found that while low-incomelending was more costly, lenders were com-pensated with higher revenues, making profitssimilar for both low- and high-income lending.Finally, Esty (1995) evaluated the performanceof Chicago’s South Shore Bank, which has beenheld up as the model community developmentbank with the dual objectives of making aprofit and aiding in the development of thelocal community. Esty’s analysis found theeconomic return of the bank to be substandard.Shareholders, however, appeared to be willingto trade off the lower return for the social returnreceived from community improvement. Thatis, the shareholders’ objectives were apparentlyaligned with the dual objectives of the bank.31

In interviews with shareholders and employees,Lash and Mote (1994) found similar evidenceof a willingness to trade off economic profit toemphasize the development objective. Whilethe behavior of South Shore’s management andshareholders may be admirable, if Esty’s anal-ysis is correct, it is not obvious that this modelcan be implemented across the entire industry.

The second performance-related area ofresearch deals with the profit implications ofdiscrimination. If an institution overtly discrim-inates, it will deliberately forego profitablelending opportunities. This implies that lendersthat do not discriminate will be the beneficiariesof this behavior. Assuming that minority-ownedbanks do not discriminate against minorities,one might expect them to outperform the dis-criminating-banks. Calomiris, Kahn, andLonghofer (1994) developed a model of culturalaffinity to explain differences in minority deni-al rates. Their basic argument is that becauseof a general lack of familiarity with the cultureof minority applicants, the typical white loanofficer may not be as accommodating withthese applicants as he would with a white

applicant. For the minority applicant, the loanofficer will rely more heavily on low-cost,objective information instead of making theextra effort, as with the white applicant, toobtain additional information to improve thechances of approval. There is some empiricalsupport for this argument (see Hunter andWalker, 1996). Again, this implies that minority-owned banks should benefit, since they will notlack a cultural affinity with minority applicants.32

If discriminatory banks forego profitableopportunities, ceterus paribus, minority-ownedbanks should have superior profitability, lowerminority denial rates, and lower bad loans.However, the empirical evidence does notsupport this. A number of studies have foundthat minority-owned banks have lower profits(Bates and Bradford, 1980, Boorman and Kwast,1974, and Brimmer, 1971). There is also evi-dence of higher loan losses at minority-ownedbanks (Kwast, 1981). Additionally, there isevidence that bank ownership shifts from whiteto black control result in fewer loans beinggenerated (Dahl, 1996). Generally, there isevidence that minority-owned banks do nothave particularly good performance or lendingrecords and have relatively poor CRA ratings(Kwast and Black, 1983, Clair, 1988, andBlack, Collins, and Cyree, 1997). This evidenceis not consistent with overt discrimination.

In summary, the findings for the variousforms of discrimination are quite mixed.While some studies have found race to be afactor in loan decisions, the evidence is farfrom conclusive. Additionally, methodologicalproblems bring into question the validity ofmany studies. Parties on either side of theissue frequently draw uncritically on the stud-ies that align with their own position. Addi-tional research is needed before we can drawmeaningful conclusions about race and thecredit decision.

Recent trends in mortgage activity:The effect of regulation

How successful has the recent enforcementof the CRA and the fair lending laws been?Headlines proclaiming surges in credit to minor-ity groups suggest that the stricter enforcementof the CRA and fair lending laws during the1990s has been successful. Most of theseclaims are based on recent trends in lending tolow-income individuals or in low-income neigh-borhoods, such as those presented in figure 1.33


FIGURE 1

Mortgage originations—1–4 family, owner-occupied

0

50

100

150

200

250

1990 ’91 ’92 ’93 ’94 ’95

thousands

Low

By tract income

ModerateLowModerate

0

100

200

300

400

500

1990 ’91 ’92 ’93 ’94 ’95

thousandsBy personal income

0

70

140

210

280

350

1990 ’91 ’92 ’93 ’94 ’95

thousandsBy race

Other minorityBlack

0

30

60

90

120

150

1990 ’91 ’92 ’93 ’94 ’95

thousandsBy tract minority (percent)

>75% minority50-75% minority

Note: See footnote 33 for the definition of income groups and footnote 40 for an explanation of inclusion of data in the sample.

Source: Home Mortgage Disclosure Act data, various years.

Between 1990 and 1995 the annual number ofmortgage originations to low- and moderate-income households, in low- and moderate-income census tracts, and to minorities almostdoubled. New loans in census tracts whereminorities accounted for at least half the popu-lation also grew significantly. As figure 1shows, there was a considerable increase in thenumber of loans to individuals targeted by fairlending and CRA regulations. Some bankersargue that the regulatory mandate to increaselending in low-income neighborhoods and tolow-income individuals has actually been ablessing in disguise as it has opened up new,lucrative, previously untapped markets. Oth-ers continue to criticize the regulations.34

A full assessment of the success of theCRA and the fair lending programs would re-quire a comprehensive cost-benefit analysis.Accurately quantifying the cost is difficult.35 Itis also difficult to quantify the success of theseprograms because of the vagueness of the legis-lation and the regulations enforcing it. Themandate to banks was to use the proper criteria

in making loan decisions and to reach out tothe local community, including low- and mod-erate-income neighborhoods and individuals.Based on this mandate, success may not requireany change in lending patterns. Another prob-lem with associating recent lending patternswith regulation is the lack of a control group.The issue is not whether lending to the targetedgroups increased, but whether it increased as aresult of the regulations,

There are, however, a number of creditflow measures often associated with the CRAand fair lending laws. Concerning redlining,one would want to analyze changes in thevolume and dollar value of loans flowing intolow-income or minority neighborhoods. Thenumber of applications in these areas couldalso be considered if redlining resulted in ap-plications not being accepted from these areas.Some argue that the purpose of the regulationswas to increase the flow of credit to specificgroups of borrowers (based either on income orrace), therefore, credit flows to those particulargroups could be analyzed. It has also been


argued that the elimination of dif-ferences in denial rates may bedesirable.36 Concerning fair lend-ing, one would want to analyzechanges in lending to minorityindividuals and/or denial rates.

Although data limitationshamper the degree to which rigor-ous analysis of the regulatory impactcan be undertaken, we can evaluatelending patterns and check fortrends consistent with what aretypically perceived to be desiredchanges. We use three differentcontrol group specifications. First,we compare lending patterns pre-and post- the recent regulatorychanges. Second, we compare thedegree of lending to targetedgroups (minority and low-incomeindividuals and neighborhoods)with lending to nontargeted groups. Last, wecompare the lending behavior of more heavilyregulated depository institutions with that ofless regulated mortgage companies.

Below, we present evidence on these creditflows and discuss how they align with the goalsof the legislation. We would expect the CRAand fair lending reforms of the late 1980s andearly 1990s to have increased lending to mi-nority and low-income individuals and low-income neighborhoods. We would also expectthat most of the impact would be concentratedin recent years as regulatory, legal, and publicscrutiny intensified and the cost of failing tosatisfy the requirements increased. We analyzetwo potential effects. If the regulations weresuccessful in getting lenders to expand theirbusiness into new markets, this might influencethe overall level of mortgage activity. Alterna-tively, we may see distributional effects aslenders allocated a larger share of the creditpool toward the new markets.

To analyze the effect on the aggregatelevel of mortgage activity, we used a nominaldollar measure of all mortgage activity for1970 through 1995, combining originationsand refinances, from the Survey of MortgageLending Activity issued by the U.S. Departmentof Housing and Urban Development (HUD)(see figure 2). Figure 2 suggests considerablegrowth in mortgage originations over the 1990s,in particular 1993–94. Can the high rate of

growth in the 1990s be attributed to regulatory-induced changes in lending behavior or toother factors related to the aggregate demandfor housing credit? Three pieces of evidencesuggest that the latter hypothesis may be moreaccurate. First, there is considerable growththroughout the entire period, even in real termsas indicated by the colored line depicting thedollar value of mortgage originations deflatedby the Consumer Price Index for Urban Workers.The colored line also highlights the procyclicaland seasonal nature of mortgage originations.Second, there is a substantial decline in thenumber of mortgage originations after the 1993peak, which might be due to a curtailment ofrefinance activity.37 Clearly, there is no regula-tory explanation for this decline. There areprobably a number of factors beyond regula-tion that affect the number of originations andincrease the difficulty of graphically detectinga structural break. To address this, we use aregression model of the quarterly growth rateof originations, controlling for the growth rateof gross domestic product (GDP), the changein mortgage rates, and the growth rate of theconsumer price index.38 Quarterly indicatorsare included to absorb the seasonality in thedependent variable. Table 2 displays the re-gression results. In column 1, over 50 percentof the variation in the growth of originations isexplained by the right-hand-side variables.The coefficients suggest that an increase in

FIGURE 2

Historical mortgage origination activity

0

100

200

300

1970 ’75 ’80 ’85 ’90 ’95

millions of dollars, quarterly

Nominal

Real

Note: Shaded areas indicate recessions.

Source: U.S. Department of Housing and Urban Development,

Survey of Mortgage Lending Activity, various years.


mortgage rates corresponds to a decrease in thegrowth of mortgage originations, while anincrease in the growth rate of GDP correspondsto an increase in the level of originations. Thecontrols for seasonality, the quarterly indicatorvariables, suggest faster growth in mortgageoriginations in the second and third quartersthan in the first and fourth. Column 2 of thetable presents the regression model with theaddition of a binary variable to capture a struc-tural shift in the post-1990 period. The coeffi-cient of the post-1990 indicator variable isactually negative, but is significant at only the74 percent confidence level. Therefore, theregression model is unable to support the hy-pothesis that mortgage originations were stron-ger in the period following the recent regulato-ry changes. Tests for structural breaks forpost-1992 and post-1993, columns 3 and 4,produced similar results.

Although we find the recent growth inmortgage lending is consistent with earlier

patterns, the regulations may have resulted indistributional changes in lending patterns.39 Thatis, there may be a shift in lending emphasis awayfrom traditional markets toward low- or moder-ate-income groups and individuals. To evaluatethis, we assembled HMDA data for depositoryinstitutions and their affiliates over the 1982–95period and decomposed the data by incomegroups and, when possible, racial groups.40

To the extent that regulations influencelending patterns, we have argued above thatthe effect would be most evident in recentyears, because of increased scrutiny, and mostpronounced among low-income and minorityborrowers and neighborhoods. We thereforedivided total lending activity into four incomecategories and evaluated growth trends in thenumber of mortgage applicants, originations,and dollar value of originations for the 1990s.We also developed data for the number anddollar value of originations for the 1980s byneighborhood income levels.41 Table 3 shows

TABLE 2

Model of aggregate growth in U.S. mortgage activity

Model 1 Model 2 Model 3 Model 4

Intercept –10.01** –6.98 –7.88 –6.75

(4.66) (5.37) (5.35) (5.13)

Growth of real GDP 3.18** 2.71* 2.92* 2.90*

chain weighted (1.58) (1.64) (1.62) (1.58)

Change in mortgage rates –11.84** –11.24** –11.41** –11.04**

(2.91) (2.95) (2.96) (2.94)

Growth of CPI–U –2.15 –3.27 –3.02 –3.53

(2.11) (2.33) (2.37) (2.30)

Quarter 2 seasonal 35.66** 35.47** 35.52** 35.39**

(4.19) (4.18) (4.20) (4.16)

Quarter 3 seasonal 16.49** 16.31** 16.37** 16.29**

(4.08) (4.08) (4.09) (4.06)

Quarter 4 seasonal 1.66 1.47 1.53 1.46

(4.07) (4.07) (4.08) (4.05)

1990 and beyond –4.2

(3.72)


(4.02)


(4.21)

Error degrees of freedom 91.00 90.00 90.00 90.00

Adjusted R2 0.54 0.54 0.54 0.55

Durbin Watson 2.21 2.19 2.20 2.22

** Indicates a t–statistic greater than or equal to 1.96.

* Indicates a t–statistic greater than or equal to 1.64.

Note: Standard errors are in parentheses.


dollars lent. Low- and moderate-income individuals, as opposed tohome purchases in low- and mod-erate-income tracts, representnearly 20 percent of applicationsand originations, but still less than10 percent of the dollars lent.Roughly 80 percent of mortgageactivity (applicants, originations,and dollars) involved white appli-cants. These figures demonstratethat the targeted populations are arelatively small share of the aggre-gate, which may explain whyincreases may not be observablein the aggregate data.

The year-over-year growthrates for the number of loansoriginated by neighborhood in-come groups for the 1980s and1990s are presented in figures 3and 4, respectively. For the1980s, growth in the low- andmoderate-income groups lagged

that in other areas. For four of the years in the1980s, the low-income tracts showed the slow-est growth and the moderate-income tracts alsoshowed relatively slow growth. In these years,growth in the overall market was quite robust.Thus, for the 1980s overall, growth in loanvolume was not particularly concentrated in

the low- and moderate-incomegroups. Originations to the low-and moderate-income groupsgrew more than 40 percent be-tween 1985 and 1986, exceedinggrowth for these groups for anysingle year throughout the 1990s.However, the 1986 growth ofmortgage originations in middle-and upper-income tracts stillexceeded that of the low- andmoderate-income groups.

Things changed in the1990s. After 1991, growth wasrelatively fast in the two lowestincome groups, particularly inthe years when overall marketgrowth was greatest. This find-ing suggests that banks haveresponded to the CRA and havemade significantly more loansin the low- and moderate-in-come markets. The change is

Base mortgage lending data, 1990TABLE 3

Applications Acceptances Dollars

(000s) (000s) (billion)

Total Applications 1,491.35 1,276.16 127.72

Tract income/MSA income shares

Low income 0.01 0.01 0.01

Moderate income 0.10 0.09 0.08

Medium income 0.57 0.57 0.49

High income 0.31 0.33 0.43

Applicant income/MSAincome shares

Low income 0.05 0.04 0.01

Moderate income 0.17 0.16 0.08

Medium income 0.27 0.28 0.20

High income 0.50 0.53 0.71

Race shares

White 0.82 0.84 0.82

Black 0.06 0.05 0.04

Other minority 0.11 0.11 0.13

Source: HMDA data, see footnotes 33 and 40.

FIGURE 3

Annual growth in mortgage originations,by neighborhood income, 1980s

1982 ’83 ’84 ’85 ’86 ’87 ’88 ’89 ’90-20

0

20

120percent, year/year growth rates

HighMiddleModerateLow

67

39

88

104

117

47

67

74

Note: See footnote 33 for the definition of income groups and

footnote 40 for an explanation of inclusion of data in the sample.


the 1990 levels and relative shares of mortgageactivity on which our analysis is based. Thetargeted groups received a relatively smallportion of the applications, originations, anddollars. Low- and moderate-income tractsaccount for approximately 10 percent of theapplications and loans and slightly less of the


overwhelmingly statistically significant basedon a test of whether the share of loans in eachincome category is constant throughout the1990s. We conclude that the growth in mortgageoriginations has not been uniform throughoutthe 1990s, consistent with the 1992 through1994 growth spurt in lending to low- andmoderate-income groups.

Figure 5 presents data for recent mortgageapplications.42 After 1991, low- and moderate-

income neighborhoods saw signifi-cantly stronger growth than oc-curred in other areas. Growth inthe middle-income group, wherethe majority of mortgage activityis occurring (see table 3), saw theslowest increase over this period.These trends are consistent withthe view that banks have beenmaking a significant effort toencourage applications from lower-income neighborhoods and withstatements by community groupsthat progress is being made inless affluent neighborhoods. Thetest of differences across catego-ries is again highly statisticallysignificant.43

Figures 6 and 7 analyze mort-gage activity by the income levelof the borrower instead of theneighborhood in which the prop-

erty is located. While not quite as pronouncedas the neighborhood data, figure 6 showsgrowth in mortgage activity to low- and mod-erate-income individuals, particularly for theyears in which overall growth was greatest.The mortgage application data in figure 7 tella similar story. Overall, the data suggest anincrease in the growth of loan applications andloans approved for low- and moderate-incomeindividuals, with much of the growth coming

after 1992. However, the differ-ences are only significant at the 46percent and 54 percent confi-dence levels, respectively, fororiginations and applications.Thus, based on this test, lendingto low- and moderate-incomeindividuals was uniform through-out the 1990s. Data on applicantincome were not collected for the1980s so we cannot compare thetwo periods.

Figures 8 and 9 show loanactivity by applicant race. Whileneither the CRA nor the fair lend-ing laws explicitly require lendersto change underwriting criteriaand affirmatively pursue additionalminority mortgage business, lend-ers may believe that doing so willhelp them avoid charges of dis-crimination and be looked upon

FIGURE 4

Annual growth in mortgage originations, byneighborhood income, 1990s

-25

0

25

50

1991 ’92 ’93 ’94 ’95

percent, year/year growth rates





FIGURE 5

Annual growth in mortgage applicants byneighborhood income

-20

0

20

40

1991 ’92 ’93 ’94 ’95







more favorably during their regulatory assess-ment. The growth in minority applications andoriginations during the 1990s has been highrelative to that for nonminorities. The increasein applications and originations among blacks iseven more significant.44 Figures 10 and 11present a similar analysis based on the minorityproportion of the census tract as opposed to theminority status of the applicant. Since 1991,

growth appears to have been relative-ly similar across the groups.

Several of the results in fig-ures 3–11 are consistent withefforts by banks to target low-and moderate-income individualsand neighborhoods in their mort-gage business. This observationis based not on the level of loansmade but on the fact that thegrowth in lending to the targetedgroups exceeded that to the non-targeted groups. One could ar-gue, however, that the improve-ments are somewhat diminished,being from such a small base (seetable 3). Lending in low- andmoderate-income neighborhoodsconstitutes approximately 10percent of total originations andeven less of the dollar value ofloans originated. Based on in-

come alone, we would expect the demand formortgages in these neighborhoods to be rela-tively low. Mortgage activity among low-income individuals constitutes approximately20 percent of the market. However, the 31percent growth in mortgage originations inlow- and moderate-income tracts from 1993to 1994 corresponds to nearly 35,000 loansand approximately $2.7 billion. If all of this

change is attributed to the regula-tions, it translates to just over 100loans and $8 million per MSA.

In addition to the number ofloans and applications, we evalu-ated denial rates for ethnicgroups. Minorities have typicallybeen shown to have higher denialrates than other applicants. Oneof the common debates in theliterature and popular press iswhether the differences acrossracial groups can be explainedby economic characteristics.45

We present two measures of dif-ferences in denial rates. The firstis a standard odds ratio: Based onthe loan decision, we calculate theodds of a minority applicant beingdenied a loan relative to the oddsof a nonminority applicant being

FIGURE 6

Annual growth in mortgage approvalby applicant income

-20

0

20

40

60

80

1991 ’92 ’93 ’94 ’95






FIGURE 7

Annual growth in mortgage applicationsby applicant income

-20

0

20

40

60

80

1991 ’92 ’93 ’94 ’95







FIGURE 8

Annual growth in mortgage originationsby applicant race

-20

0

20

40

60

1991 ’92 ’93 ’94 ’95


MinorityBlackWhite




The odds ratios are presented intable 4.48

Interpretation of the oddsratios as evidence of discrimina-tion is difficult due to the smallnumber of variables collected inthe HMDA data. Instead, wefocus on the changes in the ratioover time. Under the assumptionof constant quality of the appli-cant pools, changes in the ratiosmay be attributed to changes inlender behavior. Cyclical econom-ic changes, however, are likely toaffect the creditworthiness of theapplicant pool. To account forthis, we also calculated the twoodds ratio measures for a sampleof independent mortgage compa-nies. These are typically thoughtto be less stringently regulatedwith respect to the CRA, but they

still report for HMDA purposes through HUD.Thus, we use these as a control group that wecan contrast with the regulated banks to distin-guish the effect of regulation.

As table 4 indicates, for depository institu-tions the unconditioned odds ratio is relativelystable during the 1990s.49 The odds of a minor-ity applicant being denied a mortgage requestare approximately twice those of a nonminority

applicant. The odds ratio andtrend estimates are statisticallydifferent from 0 at the 99 percentconfidence level. Differencesbetween years are typically statis-tically significant at conventionallevels. The conditioned ratio forbanks is somewhat similar to theunconditioned measure, but de-clines throughout the period. Theadditional information embeddedin the conditioned measure doesexplain part of the differencebetween the two borrowing groups;however, it leaves much unex-plained. Additional informationon the creditworthiness of theapplicant and any discriminatoryeffects would be needed to explainthe remainder. The decliningtrend in the odds ratio, after condi-tioning on a flexible specification

denied.46 An odds ratio of 1 corresponds toequality of the denial odds for white and mi-nority applicants; values above 1 correspond tomore minority denials. If minority status isassociated with lower creditworthiness, wewould expect the odds ratio to be higher becauseof the differences in qualifications. To partial-ly account for this, we also calculated an oddsratio conditioned on income and loan value.47

FIGURE 9

Annual growth in mortgage applicationsby applicant race

-25

0

25

50

1991 ’92 ’93 ’94 ’95


MinorityBlackWhite





of loan amount and applicant income, suggestsa change in the treatment of minority appli-cants relative to nonminority applicants overthe 1990–95 period. Essentially, lenders be-came more accommodating to minorities. Theeffect is more apparent in the last two columnsof the table, where we repeat the analysis usingblack/white odds ratios. These results areconsistent with more stringent regulations

producing a change in lenderbehavior, assuming that the un-observed characteristicsof the applicant pool remainconstant over time.

The odds ratios for the inde-pendent mortgage companies arealso presented in table 4. Theratios for these less regulatedcompanies are much more erratic,but display a similar downwardtrend. Disparities between minor-ity and nonminority applicantsand between black and whiteapplicants decline over time forboth sets of institutions, suggest-ing that the change may not bethe result of the regulations.50

Another way to assess theextent to which the supply of mort-gage credit to minorities increasedin the mid 1990s is to examine

home ownership rates over time. The relaxationof binding credit constraints should causeminority originations and home ownershiprates to increase. Figure 12 displays homeownership rates from 1970 to 1994 for white,black, and other minority households. Recenthome ownership rates are still well below therates observed in the early to mid-1980s. Recallthat blacks experienced the strongest growth in

mortgage originations in 1993and 1994, yet there was littleeffect on home ownership. Withinour sample, mortgage originationsto blacks increased by 23,000loans from 1992 to 1993 andanother 4,000 from 1993 to 1994.In 1994 there were roughly 11.3million black households in theU.S., implying that 113,000 newloans would be necessary to movethe home ownership rate a singlepercentage point. Viewed this way,the 1993–94 changes appear small.

Summary and conclusionsIn this article, we have pro-

vided background on the evolu-tion of the CRA and fair lendingregulations, summarized the eco-nomic literature which pertains tothis type of regulation, and present-ed evidence on the effectiveness

FIGURE 10

Annual growth in mortgage originationsby neighborhood racial composition

>75%50-75%25-50%<25%

-30

-20

-10

0

10

20

30

40

1991 ’92 ’93 ’94 ’95





FIGURE 11

Annual growth in mortgage applicationsby neighborhood racial composition

-30

-20

-10

0

10

20

30

40

1991 ’92 ’93 ’94 ’95


>75%50-75%25-50%<25%





TABLE 4

Non–HUD regulated

ConditionedDenial rates Odds ratio odds ratio

White Black Minority Black Minority Black Minority

1990 0.12 0.29 0.22 2.98 1.98 2.68 2.06

1991 0.14 0.31 0.25 2.78 2.05 2.52 2.09

1992 0.11 0.27 0.22 2.85 2.20 2.58 2.18

1993 0.11 0.25 0.22 2.72 2.25 2.39 2.15

1994 0.10 0.23 0.20 2.62 2.13 2.22 1.98

1995 0.11 0.23 0.20 2.42 2.00 2.02 1.84

Trend –0.10 0.01 –0.13 –0.05

HUD regulated

ConditionedDenial rates Odds ratio odds ratio

White Black Minority Black Minority Black Minority

1990 0.10 0.26 0.20 3.27 2.27 3.13 2.27

1991 0.26 0.36 0.28 1.62 1.11 1.43 1.37

1992 0.14 0.26 0.22 2.20 1.73 2.03 1.83

1993 0.13 0.24 0.20 2.19 1.70 1.98 1.75

1994 0.14 0.23 0.19 1.72 1.40 1.53 1.44

1995 0.20 0.26 0.23 1.40 1.19 1.33 1.27

Trend –0.13 0.03 –0.11 –0.09

Source: HMDA data, see footnotes 33 and 40.

Comparison of mortgage denial rates

Although early studies appear to find evidenceof redlining, more recent studies do not supportthis finding. Concerning disparate treatment

based on the race of the applicant,some studies have found differenc-es in the probability of a loanapplication being denied that arenot explained by economic charac-teristics. However, these studieshave been criticized as havingmethodological and data problems.

Our major purpose, however,is to provide the reader with areview of the literature and not totake a position on the merits ofthe various positions concerningwhether there is a need for theCRA and fair lending regulations.Regardless of one’s position, theregulations do exist and are beingenforced. We evaluated how theregulations may alter lending be-havior, using a variety of measures

FIGURE 12

Home ownership rates

35

45

55

65

75

1976 ’79 ’82 ’85 ’88 ’91 ’94

percent

White

Black

Other minorities

Source: U.S. Department of Commerce, Bureau of the Census, various years.

of these regulations by analyzing recenttrends in mortgage lending activity. Theliterature review indicated mixed results.


of changes in lending patterns. We found thatover the 1990–95 period (particularly 1993–94),loan applications and originations increasedsignificantly to groups targeted by the CRAand fair lending legislation. Additionally, itappears there may have been a compositionalshift toward blacks and a minor shift towardlow-income groups. These changes appearlarge in terms of growth rates, but they startedfrom a very small base. The changes appearsmaller when measured in dollars rather thanthe number of loans or applications.

It is difficult to attribute the increase solelyto the strengthening of the regulations. Weassessed the regulatory impact in three ways.First, we used historical trends as a controlgroup. After accounting for economic condi-tions, we found that aggregate lending in re-cent years has not been unusually strong. Sec-ond, we considered changes in the compositionof mortgage activity by examining year-to-yeargrowth rates of applications and mortgageoriginations. We used the nontargeted groupsas controls. The analyses presented some evi-dence of a change toward increased lending tominority and low-income individuals and neigh-borhoods. Last, we compared recent trends indenial disparity measures, black/white and

minority/nonminority odds ratios over timebetween depository institutions and mortgagecompanies. Typically thought to be less strin-gently regulated, mortgage companies mightdepict the market result in the absence of addi-tional regulation. Our analysis shows a declinein the odds ratio for both depository and non-depository institutions, suggesting that theeffect may not be the result of increased regu-latory scrutiny.

Overall, our results are mixed. There issome evidence of changes in lending patterns,and some of the evidence is consistent withchanges related to the new regulatory envi-ronment. However, the growth rates are notunprecedented and, if entirely attributed to theregulations, translate to approximately 100loans and $8 million per MSA. We wouldemphasize that we have not addressed thecost of implementing the regulations relativeto the benefits. In addition, we have notaddressed the question of whether these creditmarket regulations are the most effective wayof changing the fundamental economic statusof the targeted groups. As the regulation ofcredit markets continues to evolve, it remainsimportant to continually revisit these issues.

NOTES

1Throughout we use the term bank in a generic sense toencompass all depository institutions.

2See, for example, Coplan (1996), Lindsey (1996), Seiberg(1996a, 1996b), or Wilke (1996).

3For a discussion of potential information externalities, seeNakamura (1993) and Calem (1996).

4Gender was added as a protected category through 1974amendments, and handicap and family status were added byamendment in 1988.

5A number of these protected categories were added byamendments in 1976.

6See Hula (1991) and GAO (1996).

7See Dedman (1988).

8For a discussion and examples of increased scrutiny byregulators in more recent years, see Garwood and Smith(1993) and Macey and Miller (1993).

9The reporting requirement now extends beyond depositoryinstitutions to, generally, all lending institutions with assets ofmore than $10 million with an office in a metropolitan statis-tical area (MSA) (for depositories) or loan activity in MSAs(for nondepositories).

10For a description of the process by which the Federal Re-serve Banks use statistical models to test for disparate treat-ment of applicants, see Bauer and Cromwell (1994) andAvery, Beeson, and Calem (1997 forthcoming). For discus-sion of the use of statistics in detecting mortgage discrimina-tion, see Yezer (1995).

11See Interagency Regulatory Task Force (1994). Represent-ed in the interagency group were the Department of Housingand Urban Development, Department of Justice, Comptrollerof the Currency, Office of Thrift Supervision, Federal Re-serve System, Federal Deposit Insurance Corporation, FederalHousing Finance Board, Federal Trade Commission, andNational Credit Union Administration.

12Again, although the fair lending laws and the CRA aredifferent, there is significant overlap as evidenced in the lastof these items. This is very close to an anti-redlining state-ment—the very reason for the passage of the CRA.

13For examples of enforcement activities by the Department ofJustice and bank regulatory agencies, see Macey and Miller(1993) and GAO (1996). For a discussion of alternativestrategic responses by banks to the increased regulatorypressure from fair lending and CRA regulations, see Evanoffand Segal (1997 forthcoming).


14This argument was also being made at the time to contestthe liberalization of bank branching laws. The concern wasthat banks would export deposits through their branch net-work to other, more lucrative, markets.

15Regulation BB of the Code of Federal Regulations describesthe requirements of the community reinvestment regulation.

16See Macey and Miller (1993) and U.S. Congress (1989).

17There are alternative tests for wholesale or limited purposebanks, still another streamlined test for small banks, andanother for banks choosing to develop and be held account-able for a strategic plan which details how the banks intend tosatisfy CRA requirements.

18Consumer loans will be considered if the bank collects dataon this activity and requests that they be considered in theevaluation, or if regulators determine that this activity consti-tutes a substantial portion of the institution’s business.

19The explicit weight assigned for each test and grade addressesa criticism of the earlier rating system under which bankersfrequently complained about the uncertainty as to how theyshould allocate resources to improve their CRA rating.Emphasizing the lending test also addresses criticisms of theearlier grading scheme by being less process-oriented andmore results-oriented.

20See GAO (1995).

21For example, HMDA data does not contain information oncredit history, wealth, and employment stability of the appli-cant, or the value or purchase price of the property. Adversecredit history is the most common reason given in the HMDAdata for denying loans. For discussions of lending trends inHMDA data, see Canner and Passmore (1994, 1995a, 1995b).

22A review of much of the early literature can be found inBenston (1981) and Canner (1982).

23Canner, Gabriel, and Woolley (1991) reported similarfindings after controlling for market characteristics. Theinclusion of variables to capture risk factors, however, maynot resolve the endogeneity problem since their values may besupply induced.

24Some studies have addressed whether minority groups are“discriminated” against in that they are more likely to receivea particular type of loan which may have less favorable terms.For example, Canner, Gabriel, and Woolley (1991) foundminorities are more likely to obtain an FHA loan than arenonminorities.

25See Horne (1994, 1997 forthcoming), Liebowitz (1993), andDay and Liebowitz (1996).

26The authors find considerable differences in underwritingstandards across banks concerning threshold values for debtratios, loan-to-value ratios, etc. The authors realize thatunlike the more generic market model, regressions with anemphasis on bank-specific underwriting standards will notallow for a direct test of disparate impact.

27Becker (1971) is typically cited as the source of this argument.

28Although Becker actually used data from the Boston study,it is questionable if the data are appropriate for critiquing theBoston study based on default rates. The data were for loansoriginated prior to the period discussed in the Boston studyand the analysis was a comparison of default rates acrossgeographic areas based on racial composition. From the data,one cannot conclude how racial default rates compare.

29For a discussion of these concerns see Tootell (1993) andRoss (1996).

30The authors, however, emphasize the limitations of theiranalysis. For example, information on the profitability oflow-income lending is not available. Thus, the authors arecomparing overall profit levels across banks with differentlevels of low-income lending although this lending is typical-ly a relatively small portion of the overall portfolio. Therehave also been concerns expressed recently about the growingnumber of special lending programs to accommodate thetargeted groups and the resulting high default rates, seeSeiberg (1996b).

31Esty also evaluated the impact South Shore had on the localcommunity and argued that there was no evidence of anyunique positive relative impact on the local community.

32One could make the argument that banks with a significantnumber of minority loan officers could also benefit from acultural affinity. We do not have data to directly test for this.

33Low-income neighborhoods are defined as census tractswhere the median income is less than 50 percent of the MSAmedian income. The moderate-income category correspondsto greater than or equal to 50 and less than 80 percent, themiddle-income category corresponds to greater than or equalto 80 percent and less than 120 percent, and the upper-incomecategory corresponds to at least 120 percent of the MSAmedian income. Similar break points define the categories forapplicant income relative to MSA income. Tracts are alsoclassified by minority composition into low (less than 25percent minority), moderate (25 percent to 50 percent),middle (50 percent to 75 percent) and high (75 percent to 100percent).

34See, for example, Wilke (1996), Seiberg (1996), or Lind-sey (1996).

35Barefoot et al. (1993) attempt to quantify the compliancecost associated with consumer protection regulations includ-ing the CRA. They note that bankers find the CRA to be oneof the most onerous regulations faced by banks.

36A narrowing of denial rate differentials is frequently cited inthe popular press as a measure of progress in fair lending, forexample, Coplan (1996). Lawmakers have also argued thatalthough some differential may be warranted based on creditquality, the current differences are too great and imply somediscrimination is being undertaken; for example, in U.S.Congress (1989), see Illinois Senator Dixon’s opening com-ments during hearings on the CRA.

37Refinancing and new originations cannot be separated in theHUD data.

38Changes in mortgage rates measure the contract rate on 30-year fixed rate conventional mortgage commitments reportedby the Federal Home Loan Mortgage Corporation.

39For critiques of the CRA as a means of accomplishing thisredistribution, see Lacker (1995), Macey and Miller (1993),and White (1993).

40It is important to emphasize that our sample may differ fromHMDA data reported elsewhere because, unless noted, we areviewing lending activity only for depository institutions ortheir affiliates, and we are screening out observations thatcannot be classified into income and racial subgroups whichwe expect to be affected by the regulation. By analyzing thisgroup of institutions, we have a homogenous group through


time. Certain mortgage originators were only added to theHMDA in recent years. Our sample consists of loan applica-tions for one-to-four family, owner-occupied properties wherethe mortgage is valued between $1,000 and $1 million, theapplicant’s income is less than $1 million, and the loan-to-income ratio is less than five. For the 1980s these require-ments were imposed on market (census tract) averages sinceindividual loan data were not available on HMDA until 1990.We only consider mortgages for properties in MSAs, and werequire complete data on the location of the property—state,county, MSA, and census tract—to allow us to combineHMDA data with census information concerning neighbor-hood income and composition. Reporting institutions mustreport on applications for property located in an MSA wherethey have an office. If the property is outside of an MSA orin an MSA in which the institution does not have an office, ithas the option of reporting the MSA information. Thus, someloans made in MSAs will be omitted from our sample becausethe bank chose not to include the information. The MSAinformation is necessary to merge the data with censusinformation. Finally, the data must pass the validity checks(Board of Governors 1995).

41The data are not precisely comparable to that for the 1990sbecause mortgage originations for purchase and refinancingwere not separated during this earlier period. However, therole played by “refis” in the 1980s is not expected to benearly as erratic as in the 1990s. All analysis of the 1990sexcludes refinances.

42Application data are not available for the 1980s.

43We reject the hypothesis that shares by income categoriesare constant over the 1990–95 period based on a chi-squaretest statistic of 53 with 15 degrees of freedom.

44The increase in mortgage activity for blacks was also spreadacross all income levels. The 1990–95 growth rates forblacks in low-, moderate-, middle-, and upper-income neigh-borhoods were 157 percent, 129 percent, 99 percent, and 101percent, respectively. In their evaluation of redlining Holmes

and Horvitz (1994) found that, after accounting for neighbor-hood characteristics including risk, more credit was beingmade available in certain minority neighborhoods. Theyfound a systematic preference on the part of insured lenders(the FHA) toward lending in areas of high or growing minori-ty populations. They attribute this to the pressures created bythe CRA and community groups to increase lending in theseareas. Similarly, Malmquist, Phillips-Patrick, and Rossi(1997 forthcoming) found that while low-income lending wasmore expensive, lenders were also compensated with higherrevenues making profits similar for both low- and higher-income loans. However, the authors found that profits wereinversely related to the share of mortgages originated toblacks. They suggest this is a result of firms “bending overbackwards” to yield to regulatory pressures to make moreminority loans.

45Minorities are defined as non-whites. Asians are an excep-tion and typically do not have high denial rates.

46Odds are defined as the ratio of the probability of denial tothe probability of acceptance.

47We calculate this based on a logit regression which in-cludes, in a flexible functional form, applicant income, loansize, and race.

48The denial rates may be related to other factors, for exam-ple, geographic differences. In our analysis, we emphasizechanges through time and assume the effects from otherfactors are constant across time.

49The mortgage company data should be interpreted cautiously,since these institutions do not go through the same rigorousediting process and resubmission of revised data as thedepository institutions.

50Alternatively, these “nonregulated” firms can be prosecutedby the U.S. Department of Justice for fair lending violations,so they may not be immune to this regulation.

REFERENCES

Ahlbrandt, Jr., Robert S., “Exploratory researchon the redlining phenomenon,” AREUEA Journal,Vol. 5, 1977, pp. 473–481.

Art, Robert C., “Social responsibility in bankcredit decisions: The Community Reinvestment Actone decade later,” Pacific Law Journal, Vol. 18,July 1987, pp. 1071–1139.

Avery, Robert B., Patricia E. Beeson, and PaulS. Calem, “Using HMDA data as a regulatoryscreen for fair lending compliance,” Journal ofFinancial Services Research, Vol. 11, 1997 forth-coming.

Avery, Robert B., Patricia E. Beeson, and MarkS. Sniderman, “Underserved mortgage markets:Evidence from HMDA data,” Federal Reserve Bankof Cleveland, working paper, No. 94-16, 1994.

Avery, Robert B., and Thomas M. Buynak,“Mortgage redlining: Some new evidence,” Eco-nomic Review, Federal Reserve Bank of Cleveland,Summer 1981, pp. 18–32.

Barefoot, Marrianan & Associates, Inc., A.V.Thakor, and J. C. Beltz, “Common ground: In-creasing consumer benefits and reducing regulatorycosts in banking,” Herbert V. Prochnow Education-al Foundation, monograph, 1993.

Bates, T., and W. Bradford, “An analysis of theportfolio behavior of black-owned banks” Journalof Finance, Vol. 35, 1980, pp. 753–768.

Bauer, Paul W., and Brian A. Cromwell, “AMonte Carlo examination of bias tests in mortgagelending,” Economic Review, Federal Reserve Bankof Cleveland, Quarter 3 1994, pp. 27–44.

Becker, Gary S., The Economics of Discrimina-tion, Chicago: University of Chicago Press, 1971.

, “The evidence against banks doesn’tprove bias,” Business Week, April 19, 1993.

Benston, George J., “Mortgage redlining re-search,” Journal of Bank Research, Vol. 12, Spring1981, pp. 8–23.


Benston, George J., and Dan Horsky, “Redliningand the demand for mortgages in the central cityand suburbs,” Journal of Bank Research, Vol. 10,Summer 1979, pp. 72–87.

, “The relationship between the de-mand and supply of home financing and neighbor-hood characteristics: An empirical study of mort-gage redlining,” Journal of Financial ServicesResearch, Vol. 5, 1992, pp. 235–260.

Berkovec, James A., Glenn B. Canner, Stuart A.Gabriel, and Timothy H. Hannan, “Mortgagediscrimination and FHA loan performance,” Pro-ceedings of a Conference on Bank Structure andCompetition, Chicago: Federal Reserve Bank ofChicago, May 1995, pp. 218–237.

, “Mortgage discrimination and FHAloan performance,” Cityscape, Vol. 2, February1996, pp. 9–24.

Beshouri, Christopher P., and Dennis C. Glennon,“CRA as ‘market development’ or ‘tax’: An analysisof lending decisions and economic development,”Proceedings of a Conference on Bank Structure andCompetition, Chicago: Federal Reserve Bank ofChicago, 1996 forthcoming, pp. 556–585.

Black, Harold, M. Cary Collins, and Ken B.Cyree, “Do black-owned banks discriminateagainst black borrowers?” Journal of FinancialServices Research, Vol. 11, 1997 forthcoming.

Black, Harold, Robert L. Schweitzer, and LewisMandell, “Discrimination in mortgage lending,”American Economic Review, Vol. 68, May 1978,pp. 186–191.

Board of Governors of the Federal ReserveSystem, “Regulation BB, community reinvest-ment,” Federal Reserve Regulatory Services Hand-book, Vol 3, 1995, section 228.

, “Processing procedures for HomeMortgage Disclosure Act,” technical memo, No.96, 1995.

Boorman, J. T., and Myron L. Kwast, “The start-up experience of minority-owned commercialbanks: A comparative analysis,” Journal of Fi-nance, Vol. 29, 1974, pp. 1123–1142.

Bradbury, Katherine, Karl E. Case, and Con-stance R. Dunham, “Geographic patterns of mort-gage lending in Boston, 1982–87,” New EnglandEconomic Review, Federal Reserve Bank of Bos-ton, September/October 1989, pp. 3–30.

Brimmer, A.I., “The black banks: An assessmentof performance and prospects,” Journal of Finance,Vol. 26, 1971, pp. 379–405.

Browne, Lynn Elaine, and Geoffrey M. B. Too-tell, “Mortgage lending in Boston—A response to

the critics,” New England Economic Review, Feder-al Reserve Bank of Boston, September/October1995, pp. 53–78.

Calomiris, Charles W., Charles M. Kahn, andStanley D. Longhofer, “Housing-finance interven-tion and private incentives: Helping minorities andthe poor,” Journal of Money, Credit, and Banking,Vol. 26, August 1994, pp. 634–674.

Canner, Glenn B., “Redlining: Research and federallegislative response,” Board of Governors of theFederal Reserve System, staff study, No. 121, 1982.

Canner, Glenn B., Stuart A. Gabriel, and J. M.Woolley, “Default risk and mortgage redlining: Astudy of the FHA and conventional loan markets,”Southern Economic Journal, Vol. 58, July 1991,pp. 249–262.

Canner, Glenn B., and Wayne Passmore, “Res-idential lending to low-income and minorityfamilies: Evidence from the 1992 HMDA data,”Federal Reserve Bulletin, Vol. 80, February1994, pp. 79–108.

, “Home purchase lending in low-income neighborhoods and to low-income borrow-ers,” Federal Reserve Bulletin, Vol. 81, February1995a, pp. 71–103.

, “Credit risk and the provision ofmortgages to lower-income and minority homebuyers,” Federal Reserve Bulletin, Vol. 81, No-vember 1995b, pp. 989–1016.

, “The relative profitability of com-mercial banks active in lending in lower-incomeneighborhoods and to lower-income borrowers,”Proceedings of a Conference on Bank Structureand Competition, Chicago: Federal Reserve Bankof Chicago, 1996 forthcoming, pp. 531–555.

Carr, James H., and Isaac F. Megbolugbe, “TheFederal Reserve Bank of Boston study on mortgagelending revisited,” Journal of Housing Research,Vol. 4, 1993, pp. 277–314.

Clair, Robert, “The performance of black-owned banks in their primary market areas,”Economic Review, Federal Reserve Bank ofDallas, 1988, pp. 11–20.

Coplan, Stephen, “HUD offers minority lendingdata on Internet,” American Banker, August 6,1996, p. 11.

Dahl, Drew, “Ownership changes and lending atminority banks: A note,” Journal of Banking andFinance, Vol. 20, August 1996, pp. 1288–1301.

Day, Ted, and Stanley J. Liebowitz, “Mortgages,minorities, and HMDA,” University of Texas atDallas, unpublished manuscript, March 18, 1996.


Dedman, William, “The color of money,” AtlantaConstitution, four-part series, May 1–4, 1988.

Esty, Benjamin C., “South Shore Bank: Is it themodel of success for community developmentbanks?” Proceedings of a Conference on BankStructure and Competition, Chicago: Federal Re-serve Bank of Chicago, May 1995, pp. 192–217.

Evanoff, Douglas D., and Lewis M. Segal, “Stra-tegic responses to bank regulation: Evidence fromHMDA data,” Journal of Financial Services Re-search, Vol. 11, 1997 forthcoming.

Ferguson, Michael F., and Stephen Peters, “Whatconstitutes evidence of discrimination?” Journal ofFinance, Vol. 50, June 1995, pp. 739-748.

General Accounting Office, Community Reinvest-ment Act: Challenges Remain to Successfully Im-plement CRA, Washington: Government PrintingOffice, 1995.

, Fair Lending: Federal Oversight andEnforcement Improved but Some Challenges Remain,Washington: Government Printing Office, 1996.

Garwood, Griffith L., and Dolores S. Smith,“The Community Reinvestment Act: Evolution andcurrent issues,” Federal Reserve Bulletin, Vol. 79,April 1993, pp. 251-267.

Glennon, Dennis, and Mitchell Stengel, “Anevaluation of the Federal Reserve Bank of Boston’sstudy of racial discrimination in mortgage lending,”Comptroller of the Currency, working paper, No.94-2, 1994.

Holmes, Andrew, and Paul Horvitz, “Mortgageredlining: Race, risk, and demand,” Journal ofFinance, Vol. 49, March 1994, pp. 81–99.

Horne, David K., “Evaluating the role of race inmortgage lending,” FDIC Banking Review, Vol. 7,1994, pp. 1–15.

, “Mortgage lending, race, and modelspecification,” Journal of Financial Services Re-search, Vol. 11, 1997 forthcoming.

Hula, Richard C., “Neighborhood developmentand local credit markets,” Urban Affairs Quarterly,Vol. 27, 1991, pp. 249–267.

Hunter, William C., and Mary Beth Walker,“The cultural affinity hypothesis and mortgagelending decisions,” Journal of Real Estate Financeand Economics, Vol. 13, 1996, pp. 57–70.

Hutchinson, Peter M., James R. Ostas, and J.David Reed, “A survey and comparison of redlin-ing influences in urban mortgage lending markets,”AREUEA Journal, Vol. 5, 1977, pp. 463–472.

Interagency Regulatory Task Force, “Policystatement on discrimination in lending,” FederalRegister, April 15, 1994, pp. 18266–18274.

Kwast, Myron L., “New minority-owned commer-cial banks--A statistical analysis,” Journal of BankResearch, Vol. 12, 1981, pp. 37–45.

Kwast, Myron L., and Harold A. Black, “Ananalysis of the behavior of mature black-ownedcommercial banks,” Journal of Economics andBusiness, Vol. 35, 1983, pp. 41–54.

Lacker, Jeffrey M., “Neighborhoods and bank-ing,” Economic Quarterly, Federal Reserve Bank ofRichmond, Spring 1995, pp. 13–37.

Lash, Nicholas, and Larry R. Mote, “Early evi-dence on banking in low- and moderate-incomeminority areas: A case study,” paper presented atthe Western Economics Association, VancouverBritish Columbia, June 29, 1994.

Liebowitz, Stanley J., “A study that deserves nocredit,” Wall Street Journal, September 1, 1993.

Lindsey, Lawrence B., “Home ownership opportu-nities in a deregulatory environment,” Marketwise,Federal Reserve Bank of Richmond, No. 1, 1996,pp. 20–27.

Macey, Jonathan R., and Geoffrey P. Miller,“The Community Reinvestment Act: Economicanalysis,” Virginia Law Review, Vol. 79, March1993, pp. 291–348.

Malmquist, David, Clifford Rossi, and FredPhillips-Patrick, “The economics of lending tolow-income mortgages,” Journal of FinancialServices Research, Vol. 11, 1997 forthcoming.

Megbolugbe, Isaac F., and Man Cho, “An empiri-cal analysis of metropolitan housing and mortgagemarkets,” Journal of Housing Research, Vol. 4,1993, pp. 191–223.

Munnell, Alicia H., Geoffrey M. B. Tootell, LynnE. Browne, and James McEneaney, “Mortgagelending in Boston: Interpreting HMDA data,” Fed-eral Reserve Bank of Boston, working paper, No.92-7, 1992.

, “Mortgage lending in Boston: Inter-preting HMDA data,” American Economic Review,Vol. 86, March 1996, pp. 25–54.

Perle, Eugene D., Kathryn Lynch, and JeffreyHorner, “Model specification and local mortgagemarket behavior,” Journal of Housing Research,Vol. 4, 1993, pp. 225–243.

Rosenblatt, Eric, “A reconsideration of discrimi-nation in mortgage underwriting with data from anational mortgage bank,” Journal of FinancialServices Research, Vol. 11, 1997 forthcoming.


Ross, Stephen L., “Flaws in the use of loan de-faults to test for mortgage discrimination and FHAloan performance,” Cityscape, Vol. 2, February1996, pp. 41–48.

Schill, Michael H., and Susan M. Wachter, “Atale of two cities: Racial and ethnic geographicdisparities in home mortgage lending in Boston andPhiladelphia,” Journal of Housing Research, Vol.4, 1993, pp. 245–275.

Seiberg, Jaret, “Efforts to halt mortgage bias seenfinally bearing fruit,” American Banker, July 31,1996a, p. 1.

, “HMDA: Five years later,” AmericanBanker, 4-part series, September 16–20, 1996b.

Shlay, Anne B., “Not in that neighborhood: Theeffects of population and housing on the distribu-tion of mortgage finance within the ChicagoSMSA,” Social Science Research, Vol. 17, 1988,pp. 137–163.

, “Financing community: Methodsfor assessing residential credit disparities, marketbarriers, and institutional reinvestment,” Journal ofUrban Affairs, Vol. 11, 1989, pp. 201–223.

Stengel, Mitchell, and Dennis Glennon, “Evaluat-ing statistical models of mortgage lending discrimi-nation: A bank-specific analysis,” Comptroller ofthe Currency, working paper, No. 95-3, 1995.

Tootell, Geoffrey, “Defaults, denials, and discrimi-nation in mortgage lending,” New England Eco-nomic Review, Federal Reserve Bank of Boston,September/October 1993, pp. 45–51.

U.S. Congress, Senate Committee on Banking,Housing, and Urban Affairs, “Discrimination inhome mortgage lending,” hearing before the 101stCongress, first session, Washington: GovernmentPrinting Office, October 24, 1989, p. 1.

White, Lawrence J., “The Community Reinvest-ment Act: Good intentions headed in the wrongdirection,” Fordham Urban Law Journal, Vol. 20,November 1993, pp. 281–292.

Wilke, John R., “Mortgage lending to minoritiesshows a sharp 1994 increase,” American Banker,February 13, 1996, p. 1.

Yezer, Anthony M. J., Fair Lending Analysis: ACompendium of Essays on the Use of Statistics,Washington: American Bankers Association, 1995.

Yezer, Anthony M.J., Robert F. Phillips, andRobert P. Trost, “Bias in estimates of discrimina-tion and default in mortgage lending: The effects ofsimultaneity and self-selection,” Journal of RealEstate Finance and Economics, Vol. 9, November1994, pp. 197–215.

Zandi, Mark, “Boston Fed’s bias study was deeplyflawed,” American Banker, August 19, 1993, p. 13.


Mor

tgag

e re

dlin

ing

stud

ies:

Cre

dit f

low

s to

geo

grap

hic

area

s

BO

X A

Au

tho

r cit

eP

eri

od

an

aly

ze

dD

ata

so

urc

es

De

pe

nd

en

t v

ari

ab

les

Ra

ce

va

ria

ble

sR

ed

lin

ing

co

nclu

sio

ns

Ah

lbra

nd

t (1

97

7)

19

73

–7

4 P

itts

bu

rgh

Cit

y P

lan

nin

g D

ep

t.M

ort

ga

ge

lo

an

s a

nd

do

lla

r%

bla

ck

% b

lack i

s n

ot

rela

ted

to

lo

an

sa

nd

19

60

–7

0 C

en

su

sv

alu

e o

f m

ort

ga

ge

lo

an

sC

ha

ng

e i

n %

bla

ck

or

do

lla

r v

alu

e o

f lo

an

s;

au

tho

rp

er

occu

pie

d h

ou

sin

g u

nit

co

nclu

de

s t

he

re i

s n

o e

vid

en

ce

of

in c

en

su

s t

ract

red

lin

ing

. H

ow

ev

er,

th

e c

ha

ng

e i

n%

bla

ck i

s p

osit

ive

ly a

sso

cia

ted

wit

h d

ep

en

de

nt

va

ria

ble

in

so

me

sp

ecif

ica

tio

ns.

Hu

tch

inso

n,

Osta

s &

19

75

su

rve

y o

f 4

S&

Ls i

nL

oca

l su

rve

y a

nd

No

. o

f m

ort

ga

ge

lo

an

s i

n%

bla

ck

Fin

ds e

vid

en

ce

co

nsis

ten

t w

ith

Re

ed

(1

97

7)

To

led

o O

hio

SM

SA

, 1

23

19

60

–7

0 C

en

su

so

wn

er

occu

pie

d h

ou

se

s,

% b

lack s

qu

are

dre

dli

nin

g.

To

tal

nu

mb

er

of

loa

ns

ce

nsu

s t

racts

No

. o

f in

su

red

lo

an

s,

Ch

an

ge

in

% b

lack

un

aff

ecte

d b

y r

ace

bu

t ra

cia

lly

% a

nd

no

. o

fm

ixe

d a

rea

s r

ece

ive

mo

reco

nv

en

tio

na

l lo

an

s,

go

ve

rnm

en

t in

su

red

lo

an

s a

nd

HI

% a

nd

no

. o

f h

om

elo

an

s.

Lo

we

r a

cce

pta

nce

ra

te f

or

imp

rov

em

en

t (H

I)H

I lo

an

s.

loa

ns a

cce

pte

d

Av

ery

& B

uy

na

k (

19

81

)1

97

7–7

9 C

lev

ela

nd

HM

DA

, 1

97

0P

erc

en

t o

f ti

tle

tra

nsfe

rs7

ca

teg

ori

es f

rom

Fin

ds n

o r

ed

lin

ing

wit

h t

ota

l sa

mp

le(C

uy

ah

og

a c

ou

nty

,C

en

su

s o

f H

ou

sin

g a

nd

fin

an

ce

d b

y b

an

ks,

S&

Ls,

larg

ely

wh

ite

to

bu

t th

is i

s d

ue

to

th

e l

en

din

g b

yb

y 3

35

ce

nsu

s t

racts

)C

ou

nty

Au

dit

or

reco

rds

mo

rtg

ag

e b

an

ke

rs, to

tal

larg

ely

bla

ck

mo

rtg

ag

e b

an

ke

rs, su

gg

esti

ng

th

at

mo

rtg

ag

es a

nd

ho

me

ba

nks a

nd

S&

Ls l

en

d l

ess i

n m

ino

rity

eq

uit

y lo

an

s (

tota

l v

alu

e o

fa

rea

s (

co

nsis

ten

t w

ith

re

dli

nin

g).

mo

rtg

ag

e a

nd

HI

loa

ns)

/(v

alu

e o

f o

wn

er

occu

pie

dh

ou

sin

g)

Ho

lme

s &

Ho

rvit

z (

19

94

)1

98

8–9

1 H

ou

sto

nC

en

su

s,

HM

DA

No

. o

f lo

an

s,

% b

lack,

% H

isp

an

icN

o r

ed

lin

ing

fo

r co

nv

en

tio

na

l lo

an

ow

ne

r o

ccu

pie

d u

nit

so

nce

ris

kfa

cto

rs a

re a

cco

un

ted

fo

r.

Pe

rle

, L

yn

ch

&1

98

2 D

etr

oit

Mic

hig

an

Fin

an

cia

lN

o. o

f lo

an

s/(

ye

ar-

rou

nd

% b

lack h

ou

se

ho

lds

Usin

g t

rad

itio

na

l m

eth

od

fin

ds

Ho

rne

r (1

99

4)

Insti

tuti

on

s B

oa

rd, C

en

su

sh

ou

sin

g u

nit

s)

ev

ide

nce

co

nsis

ten

t w

ith

re

dli

nin

g.

It d

isa

pp

ea

rs w

he

n a

mo

re c

om

pre

n-

siv

e m

od

el

is u

se

d.

Au

tho

rs a

rgu

eth

at

tra

dit

ion

al m

eth

od

s s

ay

lit

tle

ab

ou

t re

dli

nin

g.

Sh

lay

(1

98

8)

19

80

–8

3C

en

su

s,

Nu

mb

er

an

d d

oll

ar

% H

isp

an

icC

on

ve

nti

on

al:

% b

lack a

nd

Ch

ica

go

SM

SA

HM

DA

,v

alu

e o

f lo

an

s i

n m

ark

et

% b

lack

% H

isp

an

ic n

eg

ati

ve

ly in

flu

en

ce

sH

UD

the

nu

mb

er

of

loa

ns.

FH

A:

% b

lack n

eg

ati

ve

ly i

nfl

ue

nce

sn

um

be

r a

nd

do

lla

r v

alu

e o

f lo

an

s.

Be

nsto

n &

Ho

rsky

19

76

–8

2 In

dia

na

po

lis,

Sp

ecia

l S

urv

ey

Fin

ds n

o e

vid

en

ce

of

un

me

t d

em

an

d(1

97

9,1

99

2)

Cin

cin

na

ti, a

nd

for

mo

rtg

ag

e c

red

it.

No

ev

ide

nce

Na

sh

vil

le; 1

97

4–7

6o

f re

dli

nin

g.

Ro

ch

este

r, N

Y


Au

tho

r cit

eP

eri

od

an

aly

ze

dD

ata

so

urc

es

De

pe

nd

en

t v

ari

ab

les

Ra

ce

va

ria

ble

sR

ed

lin

ing

co

nclu

sio

ns

Bra

db

ury

, C

ase

19

82

–8

7 B

osto

nS

uff

olk

Co

un

tyN

um

be

r o

f lo

an

s i

n m

ark

et

% b

lack i

n t

he

Min

ori

ty s

tatu

s o

f N

SA

asso

cia

ted

& D

un

ha

m (

19

89

)R

eg

istr

y o

f D

ee

ds,

pe

r 1

00

ho

usin

g s

tru

ctu

ren

eig

hb

orh

oo

dw

ith

fe

we

r lo

an

ori

gin

ati

on

s.

Ce

nsu

s o

f P

op

ula

tio

nsta

tisti

ca

l a

rea

(N

SA

)1

0%

dif

fere

nce

in

th

e b

lack

an

d H

ou

sin

g S

urv

ey

% O

the

r m

ino

rity

race

va

ria

ble

co

rre

sp

on

ds t

oo

f C

on

su

me

r F

ina

nce

in N

SA

1.7

fe

we

r lo

an

s p

er

10

0 s

tru

ctu

res.

Sh

lay

(1

98

9)

Ba

ltim

ore

cit

yC

en

su

s,

HM

DA

Nu

mb

er

an

d d

oll

ar

va

lue

% b

lack

Le

nd

ing

in

ve

rse

ly r

ela

ted

an

d s

ub

urb

so

f lo

an

s i

n m

ark

et

to t

he

ra

cia

l v

ari

ab

le.

Ca

nn

er,

Ga

bri

el

&1

98

3 N

ati

on

al S

am

ple

Su

rve

y o

f C

on

su

me

rB

ina

ry f

or

loa

nB

lack o

r H

isp

an

icF

ind

s s

tro

ng

po

sit

ive

re

lati

on

sh

ipW

oo

lle

y (

19

91

)F

ina

nce

, 1

98

0 C

en

su

sd

eli

nq

ue

ncy

, b

ina

rya

pp

lica

nt

be

twe

en

FH

A u

se

an

d m

ino

rity

sta

tus.

of

Po

pu

lati

on

an

d H

ou

sin

gfo

r co

nv

en

tio

na

l%

min

ori

ty c

en

su

sH

ow

ev

er,

th

is i

s s

ho

wn

to

be

dri

ve

n b

yv

ers

us F

HA

lo

an

in c

en

su

s t

ract

eco

no

mic

ch

ara

cte

risti

cs o

f th

e a

pp

li-

ca

nt

an

d n

ot

the

ra

cia

l co

mp

osit

ion

of

the

ne

igh

bo

rho

od

. N

o e

vid

en

ce

of

red

lin

ing

.

Hu

la (

19

91

)1

98

1–8

7C

en

su

s,

No

. a

nd

do

lla

r v

alu

e o

fC

en

tra

l cit

y b

ina

ryL

ittl

e e

vid

en

ce

of

dis

cri

min

ati

on

.N

ati

on

al D

ata

HM

DA

loa

ns i

n C

en

su

s t

racts

% m

ino

rity

po

pu

lati

on

Me

go

lug

be

& C

ho

19

91

U.S

. M

SA

s (

less

Ce

nsu

s,

HM

DA

Nu

mb

er

of

loa

ns

% b

lack

Th

e r

ace

va

ria

ble

is p

osit

ive

ly(1

99

3)

NY

, L

A)

pe

r h

ou

sin

g s

tock i

nre

late

d t

o F

HA

vo

lum

e.

No

me

tro

po

lita

n a

rea

rela

tio

nsh

ip is f

ou

nd

fo

rco

nv

en

tio

na

l lo

an

vo

lum

e.

Av

ery

, B

ee

so

n,

19

90

–9

1 H

MD

A,

Ce

nsu

sA

cce

pt/

de

ny

Min

ori

ty b

ina

ryP

rob

ab

ilit

y o

f d

en

ial

is h

igh

er

for

all

Sn

ide

rma

n (

19

94

)N

ati

on

al D

ata

bin

ary

min

ori

tie

s, p

art

icu

larl

y b

lacks.

Ho

we

ve

r, t

he

pro

ba

bil

ity

of

de

nia

l is

no

t re

late

d t

o t

he

ra

cia

l co

mp

osit

ion

of

the

ne

igh

bo

rho

od

.

BO

X A

(C

ON

T.)


Mic

roec

onom

ic le

ndin

g st

udie

s: C

redi

t flo

ws

to in

divi

dual

s

BO

X B

Au

tho

r cit

eP

eri

od

an

aly

ze

dD

ata

so

urc

es

De

pe

nd

en

t v

ari

ab

les

Ra

ce

va

ria

ble

sC

on

clu

sio

ns

Bla

ck, S

ch

we

itze

r,1

97

6–7

7S

pe

cia

l S

urv

ey

Acce

pt/

reje

ct

bin

ary

Bla

ck b

ina

ryR

ace

ap

pe

ars

to

be

a d

ete

rmin

an

ta

nd

Ma

nd

ell

(1

97

8)

of

the

lo

an

de

cis

ion

.

Mu

nn

ell

et

al.

Bo

sto

n 1

99

0H

MD

A,

Sp

ecia

lA

cce

pt/

reje

ct

bin

ary

Bla

ck a

nd

His

pa

nic

Ra

ce

is f

ou

nd

to

in

flu

en

ce

th

e l

oa

n(1

99

2)

Su

rve

yb

ina

ryd

ecis

ion

. A

lern

ati

ve

sp

ecif

ica

tio

n (

B5

)re

jects

re

dli

nin

g h

yp

oth

esis

.

Ca

rr a

nd

Me

gb

olu

gb

eB

osto

n 1

99

0M

un

ne

ll e

t a

l. d

ata

Acce

pt/

reje

ct

bin

ary

Bla

ck a

nd

His

pa

nic

Co

nfi

rms M

un

ne

ll e

t a

l. r

esu

lts.

(19

93

)b

ina

ry

Sch

ill

& W

ach

ter

19

90

Bo

sto

n &

HM

DA

, C

en

su

s,

Acce

pt/

reje

ct

bin

ary

Pe

rce

nt

of

ho

use

ho

lds

Ev

ide

nce

of

red

lin

ing

dis

ap

pe

ars

(19

93

)P

hil

ad

elp

hia

FF

IEC

he

ad

ed

by

a b

lack.

By

aw

he

n r

isk v

ari

ab

les a

re in

clu

de

d.

His

pa

nic

. A

bin

ary

fo

r

bla

ck a

pp

lica

nts

. F

or

His

pa

nic

ap

pli

ca

nts

.

Be

rko

ve

c, C

an

ne

r,1

98

6–8

9H

UD

, C

en

su

sD

efa

ult

bin

ary

, %

lo

ss

Ra

cia

l b

ina

rie

sD

efa

ult

is n

ot

rela

ted

to

ra

cia

l co

mp

o-

Ga

bri

el

& H

an

na

nin

ev

en

t o

f d

efa

ult

sit

ion

of

ne

igh

bo

rho

od

. T

he

re i

s a

(19

94

)h

igh

er

like

lih

oo

d o

f d

efa

ult

fo

r b

lack

bo

rro

we

rs. L

osse

s a

re g

rea

ter

for

loa

ns e

xte

nd

ed

to

bla

cks a

nd

fo

r lo

an

sin

are

as w

ith

a l

arg

er

pro

po

rtio

no

f b

lacks.

Gle

nn

on

& S

ten

ge

l (1

99

4)

Bo

sto

n 1

99

0M

un

ne

ll e

t a

l. d

ata

Acce

pt/

reje

ct

bin

ary

Bla

ck a

nd

His

pa

nic

“C

lea

ne

d”

Mu

nn

ell

et

al.

da

ta a

nd

bin

ary

fou

nd

th

eir

re

su

lts w

ere

ro

bu

st.

Hu

nte

r &

Wa

lke

r (1

99

6)

Bo

sto

n 1

99

0H

MD

A a

nd

Su

rve

y d

ata

Acce

pt/

reje

ct

bin

ary

Min

ori

ty b

ina

ry (

als

oR

ace

is i

mp

ort

an

t (1

99

6)

for

the

cre

dit

fro

m M

un

ne

ll e

t a

l.in

tera

cte

d w

ith

va

rio

us

de

cis

ion

, b

ut

on

ly w

he

n t

he

ca

nd

ida

teb

orr

ow

er

ch

ara

cte

risti

cs).

is m

arg

ina

lly

(u

n)q

ua

lifi

ed

.

Bla

ck,

Co

llin

s,

Cy

ree

19

92

–9

3H

MD

A,

Ce

nsu

s,

Acce

pt/

reje

ct

bin

ary

Bla

ck b

ina

ryH

MD

A d

ata

su

gg

est

tha

t b

oth

bla

ck-

(19

97

)C

all

Re

po

rts

an

d w

hit

e-o

wn

ed

ba

nks m

ay

uti

lize

ap

pli

ca

nt

race

in

th

e m

ort

ga

ge

pro

-ce

ss. H

ow

ev

er,

aft

er

inco

rpo

rati

ng

ne

igh

bo

rho

od

ch

ara

cte

risti

cs f

rom

the

ce

nsu

s d

ata

fro

m t

he

ca

ll r

ep

ort

s,

on

ly b

lack-o

wn

ed

ba

nks a

pp

ea

r to

uti

lize

ra

ce

in

th

e c

red

it d

ecis

ion

.

.

cra and fair lending regulations: resulting trends in .../media/publications/... · additional...

Documents