Also Inside:

Child Care andWelfare Reform

HUD’s HousingDiscrimination Study

CCommunitiesommunities & BankingankingCommunitiesommunities & BankingankingSpring 2003, Vol. 14, No. 2

Federal Reserve Bankof Boston

PredatoryLendingAttempts to Plug the Money Drain

PredatoryLendingAttempts to Plug the Money Drain

THIS ISSUE:

Predatory Lending:Attempts to Plug the Money Drain....................................3Abusive lending practices have made headlinesaround the nation, and the issue remains hotly debat-ed. Stephen O’Sullivan gives an overview and anupdate on the issue. He focuses on how various regu-lators and legislators are trying to curb abuses with-out limiting access to the subprime market.

Factoring Child Care into the Welfare to Work Equation ......7Child care supports are important in helping welfarerecipients make the transitition to work. Two states inNew England enacted different welfare programs,and studies of these different state programs bothattest to how adequate and affordable child care caninfluence a mother’s decision to go to work.

Testing for HousingDiscrimination: Findings from a HUD Study of Real-Estate Agents.................. 10 Working with a real-estate agent is typically one of thefirst steps people take when looking to rent an apart-ment or buy a home. To determine whether people ofcertain races and ethnicities are treated differently dur-ing this process, HUD sponsored a nationwide HousingDiscrimination Study. Julia Reade highlights findings.

2 c & b

Communities & Banking

The mission of Communities & Bankingis to enhance community and econom-ic development by exploring effectiveways for lenders to work with public,private, and nonprofit sectors towardproactive compliance with the Com-munity Reinvestment Act.

EditorKristin Kanders

Editorial BoardPatricia AllouiseMarques BentonCarrie ConawayAnn EgglestonJane KatzJoanna StavinsBob TannenwaldPaul Williams

Community Affairs OfficerMarilyn Weekes

Contribute to Communities & Banking.If you would like to submit an articlefor a future issue of Communities &Banking, contact the editor.

The views expressed are not necessari-ly those of the Federal Reserve Bank ofBoston or the Federal Reserve System.Information about upcoming eventsand other organizations should be con-sidered strictly informational, not as anendorsement of their activities.

Articles may be reprinted on the condi-tion that the source is credited and thedisclaimer (paragraph above) is used.Please send copies to:

Kristin KandersEditor, Communities & BankingFederal Reserve Bank of BostonP.O. Box 2076Boston, MA 02106-2076(617) 973-3997kristin.kanders@bos.frb.org

For free subscriptions contact:Public and Community AffairsFederal Reserve Bank of BostonP.O. Box 2076Boston, MA 02106-20761-800-409-1333bostonfed.publications@bos.frb.org

Available on the web atwww.bos.frb.org/commdev/html/c&b.htm

banking trends

research review

research review

redatory lending continues to capture attentionnationwide. Since it entered the spotlight in the1990s, advocates, legislators, regulators, lawyers,and lenders have intensified their activities

around the issue. Over the past year, the Board ofGovernors of the Federal Reserve System initiatedamendments to the Home Mortgage Disclosure Act(HMDA); community groups, city councilors, and otherlawmakers proposed additional protections to curb abu-sive lending practices; Georgia’s predatory lending lawcame under scrutiny; and Household International andCitigroup settled lawsuits (worth $700 million in total)that alleged abusive lending practices.1

Why is predatory lending so heatedly debated? Besidesthe sometimes devastating consequences of predatorylending, the practice itself has evaded simple definitionand detection, allowing for a lot of debate about solu-tions. This article reviews some of the issues involved inisolating predatory lending and describes efforts underway to curb the practice.

Predatory versus SubprimeEstablishing an agreed-upon, standard definition for theterm “predatory lending” has not been easy. State andfederal regulators, financial institutions, mortgage indus-try associations, and community groups all have differentviews on what types of loan terms and activities they

consider to be traits of abusive lending. This lack of astandard definition has made it almost impossible for reg-ulators or legislators to determine what loans are, or arenot, predatory. As former U.S. Senator Phil Gramm hascommented on the issue, it is impossible to regulatesomething that cannot be defined. The box on page 4describes some loan terms and practices often associatedwith predatory lending.

Part of the difficulty in defining predatory lending isthat many people mistakenly equate it with subprimelending. In actuality, predatory lending is the roguecomponent of subprime lending. Regulatory guidelinesdescribe subprime lending as the extension of credit to“borrowers who exhibit characteristics indicating a sig-nificantly higher risk of default than traditional banklending customers.” To compensate for these higherrisks, lenders charge higher rates and fees. Borrowersbenefit by qualifying for credit they otherwise wouldn’tobtain, and lenders gain access to a new and potentiallyprofitable market.

Subprime lending started proliferating in the 1990s.Home Mortgage Disclosure Act (HMDA) data show thatfrom 1993 to 2000, the number of subprime loans forhome purchases shot up by a factor of 19, from 16,000to 306,000. The rise in subprime loans that are home-equity loans has been less steep, increasing from 66,000

P

Predatory LendingAttempts to Plug the Money DrainPredatory LendingAttempts to Plug the Money DrainBy Stephen O’Sullivan, Federal Reserve Bank of Boston

to 658,000, but these loans repre-sent a much larger segment of themarket. The subprime market con-tinues to expand; its value in 2002was estimated at $175 billion, orabout 10 percent of the total dollaramount of the residential mortgagemarket, according to the MortgageIndustry Directory.

Federal Reserve Board GovernorEdward Gramlich has suggested thatthe subprime market helped make itpossible for low-income and minor-ity consumers, groups that tradition-ally have had difficulty obtainingmortgage credit, to obtain housing

loans at record levels. During therise in subprime lending from 1993to 2000, the number of convention-al home-purchase mortgage loansgoing to lower-income borrowersnearly doubled, while those toupper-income borrowers increasedat a slower pace of two-thirds.Further, conventional home mort-gage loans increased 122 percent to

African-American borrowers and147 percent to Hispanic borrowers.Other factors, such as low interestrates, advances in technology, andgreater competition also worked tobroaden access.2

The dark side of subprime lending isthat it acts as the umbrella underwhich predatory lending finds cover.Subprime lenders have been foundto target people in particular com-munities and groups, regardless oftheir ability to qualify for betterloans. Fannie Mae and Freddie Machave pointed out that between 30and 50 percent of those who getsubprime loans could qualify forprime rate loans, or at least loanswith lower interest rates. In a studyof the subprime market in Bostonfrom 1999 to 2001, “BorrowingTrouble? III,” Jim Campen of theUniversity of Massachusetts –Boston finds the subprime loanshare to upper-income blacks is 7.5times greater than the subprime loanshare going to upper-income whites,and three times greater than theshare going to lower-income whites.

Both consumer advocates andlenders look to mortgage delinquen-cy and foreclosure data for valida-tion about whether loan costs andfees are warranted or excessive. Forexample, U.S. foreclosure datareleased by the Mortgage BankersAssociation of America show that0.86 percent for all conventionalloans were foreclosed on during thefourth quarter of 2002. Furtheranalysis reveals that only 0.54 per-cent of conventional prime loanswere in foreclosure, whereas 7.79

percent of conventional subprimeloans were. Consumer advocatespoint to this information and chargethat subprime lenders are saddlingborrowers with loans they cannotafford. Lenders assert these data pro-vide evidence of the higher risksassociated with subprime lending.

Detecting Predatory LendersPresumably, financial institutions tryto avoid being linked with the word“predatory.” Such a stigma would behard to erase, and the institutionwould be intensely scrutinized byregulators and others, potentiallyresulting in consumer lawsuits thatmight force the company to shutterits mortgage-lending operation.

Despite the risks of running suchan operation, predatory lendingcontinues. In part, this is because amajority of total residential mort-gage originations occur outside therealm of federal- and state-regulat-ed financial institutions. Mortgagebrokers, for instance, are typicallylicensed and examined by stateagencies, but the degree of moni-toring varies from state to state. Atthe federal level, the Federal TradeCommission, an entity chargedwith enforcing numerous federalantitrust and consumer protectionlaws, has oversight responsibilityfor mortgage brokers and othernonbank lenders.

Much of the information about andawareness of subprime lenders,therefore, comes from annual reportsproduced by the U.S. Department ofHousing and Urban Development(HUD). Since 1993, HUD has identi-

4 c & b

Defining Predatory Lending

In general, predatory lending refers to the actions of unscrupulous mortgage lenders and brokers in takingadvantage of poor, elderly, and unsophisticated borrowers. Here are some lending terms and practices typ-ically considered predatory:

Excessive or unnecessary interest rates, points and fees, or insuranceAsset-based lending or equity stripping: lending without regard for the borrower’s ability to repay the loan“Packing” a loan with concealed or unwarranted products and feesFinancing fees, charges, or insurance as part of the loan“Flipping” a loan by repeatedly refinancing it, charging further feesBalloon paymentsNegative amortizationIncreased interest rate after defaultCertain prepayment penalties; modification fees; mandatory arbitration clausesDeceptive or overly aggressive marketing tactics; engaging in fraudBroker fees tied to interest rates

HUD Subprime Lender TallyYear Number1993 491994 681995 1001996 1401997 2051998 2461999 2522000 1902001 178

fied lenders who appear to specializein subprime lending. The list is pri-marily compiled from industry tradepublications and HMDA data analy-ses.3 HUD cautions against assumingthat these lenders carry out predato-

ry activities, and adds that its selec-tion process is not systematic, leav-ing room for error and omissions.Regardless, many groups use the listas a starting point when attemptingto uncover lenders in the predatorymarket. The table on the facing pageshows HUD’s count of subprimelenders for the past nine years.

Attempts to Rein in Predatory Lending Given the infrastructure of mortgagelending today, local legislatures andregulators have been making theirown attempts at limiting predatorypractices. They strive for a delicatebalance. If the scales are tipped tooheavily in one direction with rigidregulations, legitimate subprimeproducts could be eliminated. Somelenders might move out of particulargeographic areas, crimping theavailability of subprime credit. But ifthe scales are tipped too much in theother direction, the most vulnerableborrowers could become victims oflending scams and abuses.

Across the country, various legisla-tures and regulatory groups havetaken different stances on the issue,resulting in a patchwork of rules andregulations. Financial institutionsand mortgage lenders have criticizedthis hodgepodge of regulations,claiming the rules hurt the borrow-ers they intend to assist. So far,about a dozen states and 10 citieshave laws and regulations (somepending) against predatory lending.

States have generally attempted tolimit predatory lending by expand-ing the requirements of the federalHome Ownership and EquityProtection Act (HOEPA). HOEPA,along with HMDA, the other rele-

vant federal regulation, have recent-ly been revised to allow for bettertracking of subprime lending activi-ty, among other aims. (For sum-maries of HOEPA and HMDA, see thesidebar “Federal Regulations” on

page 6.) States make HOEPA morestringent either by lowering theterms that trigger a high-cost loan ,thereby requiring additional disclo-sures, or by classifying additionalpractices as predatory.

The effects of these laws vary bystate, and have been interpreted dif-ferently. The Center for ResponsibleLending estimates that NorthCarolina’s anti-predatory law savedborrowers $100 million in total abu-sive lending costs in its first year ofoperation (the law was passed in1999). It reports that the new law didnot result in a mass exodus of sub-prime lenders from the state. Itclaims North Carolina ranked sixthhighest among the 50 states for sub-prime activity by the end of 2000.Other sources, however, such as theMortgage Bankers Association,attribute a major lender’s 2001 exitfrom the subprime market to stricterstate anti-predatory lending laws,including North Carolina’s.

The Georgia Fair Lending Act(GFLA), as first issued, seemed likelyto be an example of the mortgageindustry’s claim that firmer regula-tion would restrict credit overall. Butthe law was recently amended, andoutcomes of the revised law remainto be seen. When the GFLA becameeffective in October 2002, it was thestrictest anti-predatory law in thenation. Two specific provisions ofthe GFLA were especially tough. Thefirst provision was “pass-throughliability.” This meant that liability forviolations of the law moved withownership of the loan, from mort-gage lenders to loan servicers toinvestors in mortgage-backed securi-ties. The second provision concernedpenalties for violations of the law.

These provisions had unintendedconsequences. Numerous mortgagelenders either restricted their lendingactivities in Georgia or withdrew out-right from the state. Many of thosethat continued lending increased

rates and fees to compensate forgreater risks. In addition, lack ofcompetition within the state mayhave led to increased mortgage costs.

Credit-market tightening spread out-side the state as well. Standard &Poors stopped assigning credit rat-ings to asset-backed securitizationsthat included conforming mortgagesoriginated in Georgia because it saidit was unable to assess the risks.Fannie Mae and Freddie Macstopped purchasing high-cost loansoriginated in Georgia. Fortunately,the growing liquidity drought washalted in March 2003, whenGeorgia’s governor signed anamendment to the law, eliminatingthe detrimental provisions.

In addition to action at the statelevel, cities from Oakland to NewYork City have passed local ordi-nances intended to curtail predatorylending. The scope of these ordi-nances varies, but many take theapproach used in New York City. Thecity establishes what it considers tobe a predatory loan and then pro-hibits the city from doing businesswith lenders who originate thoseloans nor with the investment firmswith which they are associated.

What’s NextIndustry and consumer advocates arecurrently squaring off on federallegislation called the “ResponsibleLending Act of 2003,” introduced inFebruary by Representatives RobertNey, R-Ohio, and Ken Lucas, D-Kentucky. Consumer groups arguethe legislation intends to preempt alllocal action against predatory lend-ing without taking significant steps toend the practice. Industry groupsargue that they cannot continue oper-

5 c & b

Subprime lenders have been found to target people in particularcommunities and groups, regardless of theirability to qualify for better loans.

6 c & b

ating with different sets of rules indifferent places, and cite Georgia as acase study. Cities are not waiting toexpress their concerns. City Councilsin Boston and Bridgeport, among afew others, have passed resolutionsopposing federal preemption. �

Stephen O’Sullivan is a member ofthe Supervision and RegulationDepartment at the Federal ReserveBank of Boston.

1. Household settled a suit allegingpredatory lending practices brought by agroup of attorneys general and regula-tors from more than two dozen states onOctober 11, 2002. Citigroup settled acase brought by the Federal TradeCommission (FTC) that alleged predatorylending practices by the Associates, alender purchased by Citigroup inNovember 2000 and merged into theCitiFinancial unit. Household andCitigroup settled their cases withoutadmitting any wrongdoing in theamounts of $484 million and $215 mil-lion, respectively. The $215 million set-tlement is the largest in FTC history.2. Information in this paragraph (andHMDA statistics in the prior one) arefrom a speech on predatory lending madeby Federal Reserve Governor Gramlichbefore the Housing Bureau for SeniorsConference on January 18, 2002. Thespeech is available at www.federalreserve.gov/boarddocs/speeches/2002.3. This information was collected fromthe website, www.huduser.org/datasets/manu.html. HUD used a number ofHMDA analyses to screen potential sub-prime lenders. First, HUD assumed sub-prime lenders typically have higherdenial rates and lower origination ratesthan prime lenders. Second, HUDassumed that home refinance loans gen-erally account for higher shares of sub-prime lenders’ total originations thanprime lenders’ originations. To verifythis belief, HUD then called the lendersor reviewed their web pages to deter-mine if they specialized in subprimelending. A large number of lenders toldHUD they offer subprime loans, but thatthese loans do not constitute a large per-centage of their overall conventionalmortgage originations. In cases wherelenders offered both prime and subprimeloans, HUD identified lenders as sub-prime lenders if they reported at leasthalf of their conventional originations assubprime loans. This criteria eliminatedsome lenders who have very large (butnot chiefly) subprime businesses. Mostlenders identified themselves as primari-ly a subprime or a prime lender.

Federal Regulations

Home Ownership and Equity Protection Act (HOEPA)In 1994, Congress passed the Home Ownership and Equity Protection Act(HOEPA) in an attempt to curtail loan abuses stemming from excessivecosts. When the law passed, one of the biggest misconceptions was thatHOEPA abolished high-cost loans. HOEPA does not eliminate high-costloans or make them illegal. HOEPA was implemented as part of RegulationZ, Truth in Lending, which establishes four requirements for these typesof loans:* First, HOEPA establishes two separate thresholds for determining whattype of loan is considered a high-cost loan. One trigger is the annual per-centage rate (APR), and the other is the amount of points and fees. * Second, if a loan is determined to be high cost, the lender must providewritten notice informing the borrower of a mortgage on his home, and ofthe possibility that the borrower could lose his house. * Third, if high-cost provisions are triggered, certain loan terms are not per-mitted. These include balloon payments, negative amortization, prepaymentpenalties, increased interest rate after default, and rebates made by amethod less favorable than the actuarial method. * Fourth, three practices are not permitted with these types of loans: (1)making asset-based loans, (2) directly paying loan proceeds to home-improvement contractors, and (3) selling or assigning the loan without pro-viding a notice informing the purchaser or assignee that the loan is subjectto “special rules under the Truth in Lending Act.”

In 2001, the Federal Reserve Board amended Regulation Z to broaden thescope of loans subject to HOEPA’s regulatory protections. The Board adjust-ed the two triggers that define high-cost loans: APR and points and fees. Itlowered the APR threshold for first-lien loans from 10 percentage pointsover the rate of a Treasury bond of comparable maturity to 8 percentagepoints. (The trigger for subordinate lien loans remained at 10 percentagepoints.) In addition, the Board amended the method of calculating pointsand fees. The new regulation classifies optional single-premium creditinsurance as a fee that must be included in the calculation of total pointsand fees. The amendments, which became effective on October 1, 2002,also prohibit these loans from being refinanced within a one year period.

Home Mortgage Disclosure Act (HMDA)The Federal Reserve Board also amended Regulation C, which implementsthe federal Home Mortgage Disclosure Act (HMDA). The focus of theamendments (effective January 1, 2004) is to allow for more effective col-lection and monitoring of subprime and potential predatory lending trends.The changes require lenders to collect additional data for potential high-cost loans. The amendments:* set thresholds for determining the loans for which financial institutionsmust report loan pricing data. Institutions must report loan-pricing data ifthe rate spread (the difference between the APR on a loan and the yield oncomparable Treasury securities) equals or exceeds 3 percentage points forfirst-lien loans or 5 percentage points for subordinate-lien loans. * require lenders to report the lien status of applications and originated loans. * require lenders to ask applicants their ethnicity, race, and sex in applica-tions taken by telephone. (This became effective January 1, 2003.)

7 c & b

erhaps the largest barrier to moving welfarerecipients off cash assistance and into workand economic self-sufficiency is the lack ofadequate and affordable child care.Reauthorization of federal welfare law is

currently under way, and new requirements passedin February 2003 as part of House RepublicanWelfare Bill HR4, call for states to increase workactivity among parents who receive cash assistance.The bill aims to have 70 percent of parents partici-pate in work activity by 2008 and raises work hourrequirements to 40 hours per week. With a freezeon federal welfare funding, an uptick in caseloadsacross the nation, and state budgets under severefiscal strain, these new standards may force statesto shift funds away from already strapped supportprograms — including child-care programs andservices — that facilitate the welfare to work transi-tion for many families.

To provide a New England context for the currentreauthorization, we look at how welfare reform hasprogressed in Massachusetts and Rhode Island sincethe 1996 reforms. Rhode Island has one of the mostgenerous versions of welfare reform, whileMassachusetts has one of the most restrictive.

Child-care provisions under welfare programs inthese two states are very different, yet each pro-gram helps us see how important child care is inenabling women to work. Studying each of these

Factoring Child Care into the

Welfare to Work Equation: Lessons from Massachusetts and Rhode Islandby Ashley Maurier and Rebecca RussellFederal Reserve Bank of San Francisco and Wellesley College

P

8 c & b

Witte designed a study to measurehow child-care cost, quality, andavailability (referred to as the“child-care trilogy”) affect the prob-ability that a mother on cash assis-tance will work.

The results of this study provideimportant insights. The authorsfound that if the weekly cost of childcare were to double, a mother wouldbe five percent less likely to work.Enhancing the quality of providers,as measured by the number withnational accreditation, had a posi-tive, but small, effect on whether amother would choose to work. Theavailability of full-day kindergartenalso encouraged work, as womenliving in towns offering this type ofcare showed a three percent higherwork probability. However, increas-ing the stability of child care, byraising the number of years aprovider has been in operation, wasfound to have even larger effects.Increasing the stability of child-careproviders may have the most dra-matic impact on moving low-incomemothers from welfare to work.

Lessons from Rhode IslandIn contrast to Massachusetts, whichhas waiting lists for child-care sub-sidies, Rhode Island guaranteesvouchers to all eligible families tohelp them pay for child care. In1997, Rhode Island enacted majorreforms to encourage welfare recipi-ents to use those subsidies and makethe transition to work.

Among the reforms, Rhode Islanddecided to raise the rate at whichit reimburses child-care providers.Subsidies are now pegged to the 75thpercentile of market prices; this com-pares with Massachusetts, where thesubsidy is set at or below the median(50th percentile). Rhode Island’s rel-atively high reimbursement rateshave resulted in 90 percent of child-care providers accepting vouchers,while only 60 percent of providers inMassachusetts take them. RhodeIsland also increased the income capat which a family, and the age rangeat which a child, could qualify for asubsidy. In addition, the state offerssubsidized health insurance to fami-ly child-care providers and toemployees of centers that serve sub-sidized children. This has led to sig-

markedly different policies forchild care and welfare reform.

In Massachusetts, welfare recipientswith children over two years of agehave a maximum of 24 months ofassistance in any consecutive 60-month period. Time limits on assis-tance start ticking as soon as theyoungest child turns two years old,but the mother remains exempt fromwork requirements until the childturns six years old. Welfare recipi-ents with children over six years ofage must work at least 20 hours perweek two months after benefitsbegin. Failing to find a job, theymust fulfill 20 hours of communityservice per week to remain eligiblefor assistance.

Rhode Island enforces a 60-monthtime limit on benefits — two and ahalf times longer than the 24months allowed in Massachusetts(and the maximum allowable underfederal law). In addition, RhodeIsland recipients have the option ofcompleting up to 24 months ofeither post-secondary education orjob training that counts toward thework requirement. These 24 monthsallow for a gradual transition towork and the possibility of learningskills that might lead to a better pay-ing job.

Lessons from MassachusettsNumerous studies show that the costof child care is a significant factor inthe decision of a low-income moth-er to move from welfare to work. Yetthese studies typically do not con-sider the quality and availability ofchild care. Armed with comprehen-sive data from Massachusetts, econ-omists Lemke, Queralt, Witt, and

different approaches provides usefulinformation about child care andwork — especially as states every-where cut spending.

In Massachusetts, studies show thatthe cost, quality, and availability ofchild care play a major role in amother’s decision to choose workover welfare. In Rhode Island, stud-ies suggest that child-care subsidyreforms are important in encourag-ing mothers to use child-care bene-fits and make the transition to work.

Welfare ReformBy many measures, welfare reform —the Personal Work and ResponsibilityAct of 1996 — has been a remark-able success. The overall goal wasto end recipients’ dependence ongovernment benefits by promotingjob preparation, work, and theraising of children in two-parentfamilies. Welfare reform devolvedpower from the federal govern-ment to the states. Aid to Familieswith Dependent Children (AFDC)was replaced by TemporaryAssistance to Needy Families(TANF), which allows states toconstruct their own welfare pro-grams within federally mandatedparameters. This freedom for statesto construct their own programsexplains why, despite their geo-graphic proximity, Massachusettsand Rhode Island have adopted

Numerous studies show

that the cost of child

care is a significant factor

in the decision of a low-

income mother to move

from welfare to work.

Yet these studies

typically do not

consider the qualityand availability of

child care.

9 c & b

nificant increases in the availabilityof child care, especially in cities suchas Providence, Pawtucket, andCentral Falls.

In April 2003, Witte and Queraltcompleted a study of the impact ofRhode Island’s subsidized child-carereforms. They found that the reformsincreased by 11 percent to 13 per-cent the likelihood that a workingwelfare family would use a child-care subsidy. Further, the reformsincreased by 5 percent the probabil-ity that a single-parent family wouldleave welfare and work more than20 hours per week. Because of infor-mation lags, welfare recipients onlyslowly became aware of increasedbenefits as a result of Rhode Island’sreform — making it likely that thefull and final impact of the reform iseven greater than these estimates,which were calculated halfwaythrough year 2000.

Rhode Island’s numerous reformsalso show that raising provider reim-bursement rates indirectly affects thework decisions of low-incomewomen, ultimately increasing the

probability of work and the numberof hours worked. Rhode Island’s pol-icy made it more attractive forproviders to care for subsidized chil-dren. In turn, these providers foundit beneficial to disseminate informa-tion about the subsidy program tolow-income women — facilitatingrecruitment of children to their cen-ters. It is likely that these motherswere not aware of the full extent ofstate support, but when providedwith sufficient information, could belured off cash assistance by the rela-tive benefits of work.

Beyond the StatesLast year, the Congressional BudgetOffice predicted that the cost to

states of enforcing the 40-hour workrequirement and meeting theincreased participation rates couldreach as high as $11 billion over fiveyears. With more welfare recipientsworking and putting in more timeon the job, the costs of child-caresupports will increase. TheCongressional Budget Office esti-mates that $5 billion in child-carefunding will be needed to ensurethat TANF funds devoted to childcare keep pace with inflation. Thenew bill provides only $1 billionover the next five years in addition-al child care.

States have already had to make sub-stantial cuts in child-care programs.Compliance with the new bill willlikely require detrimental additionalcuts, particularly in light of the direfiscal situation in most states andbecause the federal government willnot be supplying additional funds.Across-the-board cuts in federalspending for child care and otherchildren’s services will result in30,000 children being dropped fromchild care in fiscal year 2003. TheBush Administration’s budget

acknowledges that, in addition tothese 30,000 children, over thecourse of the next five years at least200,000 more children will bedropped from child care.

Evidence from Massachusetts andRhode Island shows that as avail-ability of child care wanes andsupport for child-care programsdisappears, parents may decide tostay at home and work less. Thiswill hinder efforts to move familiesto self-sufficiency and threaten tocounter the progress made by wel-fare reform.�

Ashley Maurier and Rebecca Russellboth worked as research assistants to

Ann Witte at Wellesley College.Ashley Maurier is currently employedby the Federal Reserve Bank of SanFrancisco and Rebecca Russell isworking toward her economics degreeat Wellesley College. Ann Witte isdirector of Wellesley College’s ChildCare Research Partnership; she pro-vided oversight for the article.

Rhode Island’s policy made it more attractivefor providers to care for subsidized children.

In turn, these providers found it beneficial to

disseminate information about the subsidy

program to low-income women.

TESTING for HOUSING DISCRIMINATION:FINDINGS from a HUD STUDY of

REAL ESTATE AGENTSBy Julia Reade, Federal Reserve Bank of Boston

The study found that blacks andHispanics faced discriminationwhether they were renters or home-buyers. However, this study, con-ducted in 2000, found significantimprovements over an earlier studyconducted in 1989. Most of thegains occurred because bias againstblack and Hispanic homebuyers felldramatically. Bias among renters didnot show such improvement. Dis-crimination against black renters fellonly moderately, while Hispanicrenters saw no improvement.

Testing with PairsIn both 1989 and 2000, the HousingDiscrimination Study used “pairedtesting” to measure discrimination.Researchers bought the local newspa-pers in nearly two dozen metropoli-tan areas and selected a random setof advertised units for investigation.Then “testers” were sent to inquire atreal estate offices about these proper-ties. On paper, the two testers wereidentical — same family type, kind ofjob, education level, and financialstanding. The only difference wasthat one tester was white while theother was black or Hispanic.(Throughout this article, “white”refers to a “non-Hispanic white.”)

Because researchers wanted to detecteven indirect discrimination, theyemployed many measures of howeach tester was treated. Despite thesubtle issues addressed by the tests,most questions were easily answered.Simple questions like, “how manyrental units did the agent indicatewere available to you?” can illumi-nate patterns of discrimination ifsystematic differences occur by raceand ethnicity. The questions addressedissues of availability, inspections(going to see a home or apartment),costs, encouragement, and, for home-buyers, geographic steering. The boxon page 12 details the particularmeasures examined in the study.

11 c & b

ousing discrimination againstblacks and Hispanics is decliningbut still remains a significantnational problem. The U.S.

Department of Housing and UrbanDevelopment (HUD) reached this con-clusion using results of the nationalHousing Discrimination Study, conduct-ed by the Urban Institute and releasedthis past November.

The study focused on one of the earlieststeps in the process of finding housing— working with a real estate agent. Theresearchers examined whether rentersand homebuyers get different treatmentfrom real estate agents and rental man-agement companies depending on theirrace or ethnicity. The research did notassess credit issues, such as whetherpeople of certain races and ethnicitiesare disproportionately denied mort-gages, but did look at the extent towhich agents provide assistance withfinancing. Describing the importance ofthis research, the authors wrote,“Housing discrimination raises the costsof the search for housing, creates barri-ers to homeownership and housingchoice, and contributes to the perpetua-tion of racial and ethnic segregation.”

H

12 c & b

In their visits to real estate offices,testers inquired about an advertisedunit and then let the sales or rentalagent guide the next steps in theinteraction. If units were available,

testers asked to see the units.Afterward, they recorded their expe-rience on a survey form. (This setuphelped prevent real estate agentsfrom realizing they were being stud-ied.) After all the tests were complet-ed, researchers compared the infor-mation. For each measure, three out-comes were possible: the white testerwas favored, the minority tester wasfavored, or both were treated equally.

It’s All in InterpretationPaired-test studies leave a lot ofroom for interpretation. For example,in comparing the number of unitsshown to Hispanic renters versus the

number shown to whites, the whitetester received favorable treatment21 percent of the time. However, theHispanic tester received favorabletreatment 14 percent of the time.There are a couple of ways to inter-pret these results.

One interpretation focuses on a sin-gle measure of preferential treat-ment: the fact that whites are

favored in 21 percent of cases. Thisreflects the highest level offavoritism towards whites that mayhave occurred. Another interpreta-tion focuses on the net difference infavoritism between whites andHispanics, which was 7 percentagepoints. This reflects the lowest levelof favoritism towards whites thatmay have occurred.

So how much bias was there? Theresearchers stress that the underly-ing level is somewhere between 7and 21 percent. Although this is awide range, it indicates likelihood ofsome white-favored bias. The lower-bound estimate (7 percent) can betested for statistical significance. If itis found to be significant, which it isin this case, we can expect white-favored bias exists.

Discrimination among RentersBlack and White RentersBlack and white renters in the studywere treated differently. Whiteapplicants were more likely thanblacks to find better availability ofapartments (32 percent versus 28percent), mainly because agentsmore often told whites the adver-tised unit was available and recom-mended more units to them. Thedisparity in treatment, a 4-percent-age-point difference, was weaklysignificant. White applicants werealso more likely (by 8 percentagepoints, 28 percent versus 19 per-cent, taking rounding into account)to have preferential inspectionexperience, again because theymore frequently inspected theadvertised unit and saw a highernumber of comparable units.

These gaps were significantly lessthan those found in the 1989 study.The earlier gaps, 13 and 15 per-centage points, fell to 4 and 8 pointdifferences. Bias fell primarilybecause preferential treatment ofwhites declined.

Hispanic and White RentersHispanic renters experienced morediscrimination than blacks in thesame measures — availability andinspections. Rental comparisonsbetween Hispanics and whitesshowed some of the highest levels ofbias in the whole study. Whites were12 percentage points more likely to

RentersAvailabilityAdvertised unit available?Similar units available?Number of units recommended?

Overall availability

InspectionsAdvertised unit inspected?Similar units inspected?Number of units inspected?

Overall inspection

CostRent for advertised unit?Rental incentives offered?Amount of security deposit?Application fee required?

Overall cost

EncouragementFollow-up contact from agent?Asked to complete application?Arrangements for future?Told qualified to rent?

Overall encouragement

BuyersAvailabilityAdvertised unit available?Similar units available?Number of units recommended

Overall availability

InspectionsAdvertised unit inspected?Similar units inspected?Number of units inspected?

Overall inspection

Cost/FinancingHelp with financing offered?Lenders recommended?Down payment discussed?

Overall financing

EncouragementFollow-up contact from agent?Arrangements for future?Told qualified?

Overall encouragement

SteeringHomes recommended?Homes inspected?Editorializing?

How Discrimination Was MeasuredMany of the same measures were used to determine discrimination forrenters and buyers. The categories in common include availability, inspec-tions, cost, and encouragement. For buyers, steering was also studied.

Hispanic renters experienced more discrimination thanblacks in measures of availability and inspections. Rentalcomparisons between Hispanics and whites showed someof the highest levels of bias in the whole study.

about homes in neighborhoods whoseoccupants share the applicant’s raceor ethnicity. For example, a Hispanichomebuyer may be shown a highernumber of homes in neighborhoodswith more Hispanics than a whitehomebuyer would be shown. In thisstudy, the researchers focused on geo-graphic steering solely with respect tohomebuyers.

Three types of agent behavior wereconsidered indicators of steering:homes recommended, homes shown,and editorializing. Editorializing isagent commentary, both positiveand negative, on home locations.Even little comments, such as “greatschools” or “that restaurant can getnoisy,” can influence homebuyers.After testers turned in their surveys,researchers noted the address ofeach recommended and shownhome. Using data on each home’scensus tract, municipality, andschool district, they could determinewhether agents were more likely torecommend or show homes to blacksand Hispanics in areas with higher

ment was availability. This gap wasjust 3 percentage points — downfrom 17 in 1989 and one of thegreatest improvements since then.

Hispanic and White BuyersBias against Hispanic homebuyerswas confined principally to onearea: financing. (This pattern con-trasted with that of black homebuy-ers, who experienced low to moder-ate levels of bias across most meas-ures.) Whites were more likely toreceive better financing assistance(39 percent versus 24 percent), lead-ing Hispanics by 14 percentagepoints. This was the only major dis-criminatory gap to widen since1989. All other gaps shrank by 11 to15 points over the years since 1989,leaving them too small to be statis-tically significant.

Geographic Steering Exists, but at Low LevelsGeographic steering is another impor-tant but subtle form of disparatetreatment. It occurs when a renter orbuyer is given more information

find better availability (34 percentversus 22 percent) and 7 percentagepoints more likely to have betterexperience with inspections (24 per-cent versus 17 percent); both gapswere statistically significant andwere little changed from 1989because preference for both groupsfell at about the same rate.

Discrimination among BuyersBlack and White BuyersWhite preference was found morebroadly in the case of buyers. In allbut one of the areas measured, pref-erential treatment for white buyerswas statistically significant. At 9percentage points the largest gap (43percent versus 34 percent), was foroverall inspections. This gap wasdriven mainly by differences in theaverage number of units inspected.Weaker, but still statistically signifi-cant, bias was found for treatmentrelated to financing assistance (suchas when agents recommend lenders)and overall encouragement. Theonly area in which blacks andwhites received comparable treat-

13 c & b

Marital Status: MarriedFamily Size: ThreeOccupation: TeacherJoint Income: $75,000Credit Standing: Excellent

Have Same Background, Looking for Home (fictional example)

Marital Status: MarriedFamily Size: ThreeOccupation: AccountantJoint Income: $75,000Credit Standing: Excellent

14 c & b

Highest Pro-White BiasBirmingham, AlabamaOf all the metros, bias against black homebuyerswas highest in Birmingham. According to everyavailability measure, whites were favored with per-centage point gaps ranging from 14 percent to 26percent. Inspections were even more biased, withgaps over 22 percentage points. Whites inspectedmore homes than their black partners in 62 percentof tests, while blacks inspected more in only 18percent of tests, leaving a wide 44 percentage pointgap. Black renters in Birmingham also faced high-er levels of unfavorable treatment. (Data were notcollected in Birmingham for Hispanics.)

Austin, TexasBias against both black and Hispanic homebuyerswas higher in Austin than nationally. Both groupswere more likely to have unfavorable treatmentregarding inspections. They also experienced dis-parities related to availability and financing. Biasagainst black and Hispanic renters was weaker.

Lowest Pro-White BiasDenver, ColoradoBlack and Hispanic renters in Denver were treatedcomparably to whites according to almost allmeasures. Actually, the only two manifestations ofbias among renters showed that minorities receivedsignificantly favorable treatment. (Real estateagents were more likely to make future arrange-ments with black renters; Hispanic renters wereless likely to be told an application fee wasrequired.) Hispanic homebuyers were treated withlittle bias. Black homebuyers, however, faced biasat rates significantly higher than the national aver-age, particularly regarding availability, inspection,and encouragement.

Chicago, IllinoisTests in Chicago found that black and Hispanicrenters received virtually the same treatment aswhite renters according to all but one measure(Hispanics were 15 percent less likely to be offered

rental incentives). Black homebuyers were alsotreated comparably to whites according to all butone measure (blacks were 16 percent less likely tobe told they qualified for financing). Hispanichomebuyers, however, faced strong bias in allfinancing measures as well as most availability andinspection measures.

Detroit, MichiganFew measures for either renters or homebuyersshowed statistically significant net bias betweenwhites and blacks, making overall bias lowerthan average. (Data were not collected in Detroitfor Hispanics.)

Mixed BiasAtlanta, GeorgiaAtlanta’s black renters faced bias at rates muchhigher than the national average, but the city’sblack buyers faced bias at rates much lower thannationally. White renters were far more likely to bequoted a lower rent for each unit (28 percent ver-sus 6 percent) and also more likely to be offeredrental incentives (16 percent versus 5 percent).Most other rental measures were weakly biasedagainst blacks. For homebuyers, pro-black bias wasstrong according to many encouragement andfinance measures. (Data were not collected inAtlanta for Hispanics.)

Los Angeles, CaliforniaBlack and Hispanic renters were treated comparablyto whites in all but one measure (blacks were rec-ommended significantly fewer units). Black home-buyers faced bias in only two measures (whetheragents discussed down payment requirements andwhether agents told them they were qualified torent), but the gaps were high, 19 percentage pointsand a remarkable 56 percentage points, respective-ly. Hispanic homebuyers were treated with pro-Hispanic bias in availability and inspection meas-ures, but faced pro-white bias in financing.

Specific Metros Studied Bias varied considerably across metropolitan areas according to theHousing Discrimination Study. Compared with the average level ofdiscrimination nationwide, high levels of pro-white bias were foundin some metros, while low levels were found in others. Still others hadmore of a mix. Comments below summarize the researchers’ findingsof discriminatory behavior in some of these areas. Unfortunately,insufficient data often kept researchers from being able to say that thedifferences they found were statistically significant.

15 c & b

concentrations of blacks andHispanics, respectively. Did agentseditorialize on neighborhoods, steer-ing testers toward purchasing inareas where their race or ethnicitywas common?

In black/white and Hispanic/whitetests, whites were more likely thanboth blacks and Hispanics to beshown homes in census tracts withhigher concentrations of whites.Whites were also more likely thanblacks to be recommended such

homes. (These differences tended tobe weakly significant, and no differ-ences were found by municipality orschool district.) Bias in editorializ-ing was substantial; it was signifi-cant for blacks and weakly signifi-cant for Hispanics. Whites were 15percentage points more likely thanblacks and 6 percentage points morelikely than Hispanics to hear morepositive comments about homes incensus tracts with higher concentra-tions of whites.

The study also sought to determine ifagents steered along class lines,directing minorities to areas of lowersocioeconomic status. To investigatethis possibility, researchers usednumerous measures of the socioeco-nomic status of a geographic area,such as the percent of owner-occu-pied homes, median home price, percapita income, and percent of house-holds below the poverty threshold.

Evidence of class steering betweenblacks and whites existed only ineditorializing, with no differences inrecommendations or inspections.White testers were about 12 percent-age points more likely to hear posi-tive comments about areas wherefewer were poor. There was no evi-dence of socioeconomic steeringbetween Hispanics and whites.

Agents and AgenciesHaving established that bias clearlyexists when blacks and Hispanics

work with real estate agents,researchers then tried to determineif certain characteristics of agents orfirms were associated with a pro-white bias. The most consistentfinding was that older agents weremore likely to show bias against thetwo minority groups, particularlyblacks. Also, both blacks andHispanics had less favorable inspec-tion experiences with female agents.Hispanic agents were more likely togive Hispanic testers favorabletreatment, but also more likely to

give black testers less favorabletreatment. In addition, larger firmstended to exhibit higher levels ofpro-white bias.

What about Testers’ Other Characteristics?The paired-test method is often crit-icized because it measures the effectof only one characteristic: in thiscase, race/ethnicity. There is no wayto know the effects of other testercharacteristics. Is discriminationworse for people with less educa-tion? For immigrants? The authorsran some other analyses to examinethe role of these other characteris-tics. They mainly found that femaleminority testers tended to face lowerlevels of bias than male minoritytesters. The roles of other character-istics were unclear.

Future phases of the HousingDiscrimination Study will attemptto unravel these questions. Insteadof using paired testing, theresearchers will send triads of peo-ple. This way, with either two whitesand a minority or two minoritiesand a white, it will be easier to teaseout the effect of race and ethnicity.Future reports will also show thelevel of housing discriminationfaced by Asians, Native Americans,and persons with disabilities. �

Julia Reade is a community affairsanalyst with the Federal ReserveBank of Boston.

The study also sought to determine if agents steeredalong class lines, directing minorities to areas of lowersocioeconomic status.

FIRST CLASSU.S. Postage Paid

Boston, MAPermit No. 59702

Communities & Banking

Public and Community AffairsFederal Reserve Bank of BostonP.O. Box 2076Boston, MA 02106-2076

Change Service Requested

To be notified of upcoming conferences sponsored by the Boston Fed, and when the latest issue ofCommunities & Banking and other Boston Fed publications are available on the web, subscribe tothe free E-Mail Alert! service. The e-mail sent to you will give a direct link to the information orissue. To sign up, visit www.bos.frb.org/genpubs/email/index.htm

InsideTesting for HousingDiscrimination among RealEstate Agents