FHA assessment report

7/30/2019 FHA assessment report

1/13

F H A A S S E S S M E N R E P O R

Te Role o the Federal Housing Administrationin a Recovering U. S. Housing Market

Te FHA Assessment Report

is a series of reports produced by

the George Washington University

Center for Real Estate and Urban

Analysis designed to analyze

and interpret the role of and

reforms to the Federal Housing

Administration (FHA) as we

emerge from the Great Recession.

3 R D E D I T I O N (November, 2011)

Keeping in mind the substantial history o FHA, rather than simply ocusing on current

conditions, we seek to provide greater understanding o FHAs place in the market. Tos

who ail to learn rom history are doomed to repeat past mistakes. Tese reports evaluat

FHA residential mortgage activity and examine steps the agency is taking, or may consid

to ensure its long-term viability while ullling its historical goals. In this series, we con-

sider some o the dicult questions acing Congress and FHA, as well as a number o FH

reorms intended to ulll its mission and limit taxpayer risk, including:

FHAloanlimits

Lowdownpaymentloans

Otherunderwritingguidelinesandpolicies

Tis third report ocuses on mortgage risk in the broader context o the overall

housing market.


2/13page 2

F I R S R E P O R F I N D I N G S

Our rst report, released in February 2011, analyzed FHAs current policies, particularly its loan limits, historical

mission and growing market share. Te main conclusion rom that report was that FHA served the market well

during the recession, and FHA mortgages, while continuing to have higher deault rates than conventional loans

(and charging a ee to cover the costs), did not experience as large a surge in deaults as did conventional and other

(e.g., subprime) mortgage types. However, FHA has moved into riskier territory as its market share and ocus on

higher balance mortgages have increased sharply over the last ew years.

Specically, our analysis ound that the 2008 expansion o FHAs loan limits gave the program, which previously had

ocused on low- to moderate-income and rst-time homebuyers, the ability to insure nearly 97 percent o the availablelow down payment market or home purchase. As a result, FHAs share o the home purchase market increased rom 6

percent in 2007 to more than 56 percent in 2009. Additionally, we ound that FHA-insured loans that were more than

$350,000 had deault rates that were approximately 20 percent worse than those on smaller loans. Tus, it is not clear

that enlarging FHA market share by maintaining high loan limits is a good way to recapitalize the insurance und; nor

is it clear that FHA is exible enough to operate or long periods o time with a large market share.

o view the rst report, click here.

In this third edition o the FHA Assessment Report, we seek to identiy the actors that are determinants

o mortgage risk to ensure that FHA doesnt layer on excessive risk. Tese determinants include downpayments and credit scores, as well as debt-to-income ratios, which help determine deault risk. We look at

past experience with attempts to inuence the down payment constraint aced by borrowers. Our interest is

in general principles o credit risk, and we use mostly non-FHA data to provide a broad

perspective. Central points are:

Low down payment loans tend to deault more,

but not in a disastrously dierent manner.

Down payment and equity are not the only things

that matter; risk can be mitigated by other actors,

such as good credit scores.

While low down payment loans have always had

higher deault rates, during the crash their deault

rates did not increase by more than those o loans

with higher down payments. This suggests that

risks were not much dierent.

The true value o a borrowers equity in a house

can be masked by the source o the down payment.

For instance, loans with seller-assisted nancing are

apt to be associated with infated house prices and

less real equity in the property.

THIRD REPORT FOCUS

1

2

3

4
http://business.gwu.edu/files/fha-assessment-report-02-2011.pdfhttp://business.gwu.edu/files/fha-assessment-report-02-2011.pdfhttp://business.gwu.edu/files/fha-assessment-report-02-2011.pdf


3/13page 3

CAUSES OF DEFAULT, EVALUATING LOW

DOWN PAYMENTS

Mortgage underwriting standards have tightened

signicantly since 2008, and several discussions have

emerged regarding raising standards urther or FHA,

private insurers, Fannie Mae and Freddie Mac, private

label securities and other mortgage market partici-

pants. Among the proposals under consideration is

the elimination or sharp curtailment o low down

payment loans. Tis is an overreaction. I all we knew

about loans was their down payment, we should want

to control it closely. But we know more than that.Credit risk in mortgage lending is more complicated

than a single dimension o risk can capture, and much

o the current discussion has not measured risk well. It

is possible to do low down payment lending protably,

but not in a vacuum, and it need not be much riskier

than higher down payment lending.

DEFINING MORTGAGE RISK

Credit risk in mortgage lending is determined bya combination o actors, and there are tradeos

among them. Moreover, there is a tendency to

conuse risk with expected loss. Dierences in

expected loss should not pose a threat to a mortgage

insurance und i they are priced correctly within

the insurance premium. Dierences in risk (i.e.,

the variation around expected loss) present more

dicult problems or mortgage insurance and have

important implications or capital requirements.

In this report we examine more closely three

specic issues:

First,weconsidersomeimportanttradeos

among readily observable loan characteristics.

Historically, low down payment loans tend to have

higher deault rates. However, other actors, such as

credit score, can oset the risk rom low down pay-

ments; whereas other actors such as high payment

burdens can exacerbate it.

S E C O N D R E P O R F I N D I N G S

Te second report noted that current FHA loan limits were larger than necessary to serve its targeted market o rst-

time and low- to moderate-income borrowers. Te report demonstrates that FHA still could serve 95 percent o its

historic targeted market even i the maximum FHA loan limits were reduced by nearly 50 percent. Tereore, an FHAlimit o $350,000 in the high cost markets and $200,000 in the lowest cost markets is sucient to satisy more than 95

percent o FHAs target constituency. Furthermore, in order to serve its target population, FHA only needs to maintai

a market share o somewhere between 9 to 15 percent o total mortgage originations, well under its current share o

more than 30 percent. Te report also ound that the recent reductions in loan limits will aect only 3 percent o loan

endorsed in calendar year 2010. Recommendations rom this report included:

GraduallyreducingtheFHAsloanlimitinthelowestcostmarketsfromthecurrent

$271,050to$200,000.

Overtime,returningtheFHAloanlimitinhighcostmarketsfrom$729,750to$363,000

(FHAloanlimitswerereducedto$625,500onOctober1,2011).

Returningtotheuseofcurrentareamedianhomepriceincalculatingthelocalloan

maximum,movingawayfromtheinated2008medianhomepriceestimate.

o view the second report, click here.
http://business.gwu.edu/creua/research-papers/files/FHA2011Q2.pdfhttp://business.gwu.edu/creua/research-papers/files/FHA2011Q2.pdfhttp://business.gwu.edu/creua/research-papers/files/FHA2011Q2.pdf


4/13page 4

Second,weconsidertheriskofdierentloantypes

as measured by their variability or sensitivity over

the housing cycle, in particular by how much their

deault rates increased during the recession.

Third,weconsiderfactorsnotsoreadilyobserv-

able, such as the source o the down payment, that

are important and need to be considered careully

when judging the relation between loan-to-value

(LTV) and expected uture deault.

DOWN PAYMENTS (LOAN-TO-VALUE):

STRIKING THE RIGHT BALANCE ON SIZE

AND VOLUME

Tere is plenty o evidence that a lack o equity in a

property (aka skin in the game) is a determining

actor o when a borrower is more likely to deault.

Low down payment loans are more likely to become

under water when property values decline because

they start out closer to being under water. But can we

conclude that low down payment loans are riskier

than others based on LV alone?

Determining i low down payment loans are riskier

than others is dicult precisely because it hingeson the notion o risk. Casual inspection o the

data indicates that low down payment loans have

higher deault rates. However, higher deault rates

and oreclosure losses are not the same as higher

risk. Te reason or this is that risk is about devia-

tions rom the mean, not the size o the mean. High

LV deault rates are priced by FHA and by private

insurers. I deaults on high and low LV loans turn

out to be as expected, prots rom the two will be

about the same because the higher price will osetthe increased losses. Te risk question is whether or

not the prots are more or less unstable when there

are shocks to deault rates.

Here, the answer is much more mixed. Both high

and low LV loans had much higher losses than

usual in the recent and ongoing slump in house

prices. Consequently, we need to compare the in-creases in losses or dierent types o loans and loan

underwriting characteristics to get a better sense or

the risk o high LV loans.

Te ollowing three tables are rom data supplied

by the Federal Housing Finance Agency (FHFA).

Here we ocus on xed rate mortgages, including

those bought by Fannie Mae and Freddie Mac (the

Government Sponsored Enterprises [or GSEs]), and

those put into private label securities (PLS). Te datado not include FHA loans, although they do include

a wide range o mortgages, in particular, a wide

range o LV ratios and care in underwriting. Note

that FHA loans have perormed better than the PLS

loans, but worse than the GSE loans.

Te our LV classes include:

75andbelow,thesafestcategory

75-85,themostcommoncategory,whichclusters

around80(basic20percentdownloans)

85-95,highLTVloans,whichclusteraround90

95,whichcontains95andhigher

Perormance is measured by the share o loans dur-

ing that years originations that were ever deemed to

be in serious trouble (90 days delinquent or more)

through 2009.

able 1 depicts deaults on loans originated in 2003,

a good year because the economy was strong andproperty values rose rapidly in the ollowing three

years. able 2 looks at the same measure or loans

originated in 2006, a bad year with sharply declining

housing prices, and able 3 presents the dierences

between the two, showing the sensitivity o various

categories and how they compared to the mortgage

meltdown.


5/13page5

2003GSEFixedRateMortgages(FRMs)

2003PLSFRMs

720

95 20.7% 10.7% 6.5% 3.2%

720

95 25.2% 14.8% 9.9% 6.1%

FICOBucket

FICOBucket

LTVBucket(%)

LTVBu

cket(%)

TABLE 1 Default Rates: 2003 Vintage (ever seriously delinquent)


6/13page 6

TABLE 2 Default Rates: 2006 Vintage (ever seriously delinquent)

2006GSEFRMs

2006PLSFRMs

720

95 40.1% 25.6% 18.0% 9.5%

720

95 50.1% 39.0% 31.0% 22.4%

FICOBucket

FICOBucket

LTVBucket(%)

LTVBu

cket(%)


7/13page7

TABLE 3 Default Rates Between the 2006 and 2003 Vintages

Differences:20062003GSEFRMs

20062003PLSFRMs

720

95 19.4% 15.1% 11.5% 6.5%

720

95 25.9% 25.2% 21.1% 16.3%

FICOBucket

FICOBucket

LTVBucket(%)

LTVBu

cket(%)

Lookingat2003defaultrates,highLTVloansmeanthigher deaults i FICO scores were held constant,

butnotifthescorevaried.Forinstance,forGSE

data,95percentorgreaterLTVloanswithvarying

creditscores(680-720)hadaboutthesamedefault

ratesasthoseloansbelow75percentLTVwith

lowcreditscores(6.5percentvs.6.9percentrates).

There is clearly a trade-o among these two impor-

tant characteristics. What looks to be worse is not

simply high LTV, but rather high LTV combined with

low FICO score. This is an example orisklayering.

Economicconditionsareimportant.The2006vintagehadworsedefaultsthanthe2003vintage

or all categories, and the story is worse than the

tablesuggestsbecausethe2006loanshadonly

threeyearsofexposureuntil2009;whereasthe

2003loanshadsix.Werethesedataupdated,they

would illustrate starker dierences between the

two vintages.

Theoriginationchannelmatters.Forbothorigina-

tion years, holding constant the FICO and LTV on

the loans, deaults were signicantly higher or the

PLSloans,especiallyfor2006.

We make the ollowing observations:


8/13page8

Tese stylized acts tell us to be careul about

presumptions regarding single dimension causes

o deault (i.e., the rst issue noted above). High

LV is not the only source o losses, and while de-aults certainly do increase with LV, the relation is

relatively smooth.

But what about risk? Consider the last table, which

depicts dierences between the rst two. It shows

sensitivity to very poor economic conditionsa

natural stress test, which is a simple way o capturing

what economists usually think o as riskhow ar

rom the norm things vary when conditions change,

especially when they get bad.

Here the results are perhaps surprising. For instance,

look at the three highest FICO categories and two

highest LV categories or the GSE data. For those

FICO scores rom 680-720, the increase in deaults

or the highest LV loans was 11.5 percent. Howev-

er, or 75 to 85 it was actually a bit higher at 12 per-

cent. Tis is because when dealing with large shocks

to the market, such as prices alling by 40 percent

in some regions, even down payments o 20 percentprovide much less protection than usual. It was the

seemingly saest, lower LV loans that perormed

disproportionately worse. What did help were high

FICO scores and the avoidance o risk layering.

Tis doesnt end the analysis. Within these previous

tables are percentages or serious delinquencies, not

actual oreclosures, and they do not tell us about

dierences in loss severities. High LV loans tendto lose more when there is a deault, which will have

the eect o making the total loss per loan higher,

even i the increase in the share that deault is the

same.1 Nonetheless the data above suggest that high

and medium LV loans might be dierent in degree

but are not rom dierent planets.

Te data set in the rst three tables does not contain

inormation about borrower debt payments relative

to income, the debt to income ratio (DI). We turnto a separate data set, in able 4, rom CoreLogic,

Inc., which shows serious delinquencies by 2011 or

loans originated in 2008 and 2009, or FHA, GSE

and PLS loans or various loan-to-value ratios.2

Te table consists o our matrices, each or an LV

range, depicting serious delinquency by FICO score

and DI.

1 More generally, loss per loan is the product o deault rate per loan multiplied by the severity rate per deaulted loan, or rs, where r is the deault rate

(percent o loans on the category that deault) and s is the severity rate (loss per l oans that deault). The change in loss per loan is given by d(rs) where d

indicates dierence, and it is (approximately) given by d(rs)=rds+sdr. Our best guess is that dr and ds, the increases in deault rates and severity rates,

were about the same or high LTV and low LTV loans. However, because both r and s started out higher or high LTV loans, they will still have higher overall

increases in losses per loan.

2 AnalysisprovidedbyGenworthFinancialusingCoreLogicServicingDataSet.


9/13page9

TABLE 4 Default Rates: 2008 and 2009 Vintages (ever seriously delinquent)

350-639 640-679 680-719 720-950

DTI20.1-25 14.8% 5.2% 3.0% 0.5%

DTI25.1-30 16.7% 6.7% 3.5% 0.8%

DTI30.1-35 17.6% 7.8% 4.3% 1.0%

DTI35.1-40 19.9% 9.0% 5.6% 1.5%

DTI40.1-45 22.2% 11.2% 6.4% 1.8%

DTI45.1-50 23.0% 11.5% 7.2% 21.%

350-639 640-679 680-719 720-950

DTI20.1-25 17.5% 8.9% 5.4% 1.3%

DTI25.1-30 19.0% 9.3% 5.9% 1.6%

DTI30.1-35 22.2% 11.0% 7.2% 2.2%

DTI35.1-40 24.2% 13.2% 9.2% 3.2%

DTI40.1-45 27.1% 15.4% 10.9% 3.9%

DTI45.1-50 27.6% 16.5% 11.4% 4.6%

350-639 640-679 680-719 720-950

DTI20.1-25 17.4% 7.8% 5.0% 2.1%

DTI25.1-30 19.4% 9.0% 5.5% 2.5%

DTI30.1-35 22.4% 10.7% 6.7% 3.1%

DTI35.1-40 25.0% 12.7% 9.4% 4.3%

DTI40.1-45 27.2% 15.0% 11.1% 5.8%

DTI45.1-50 28.2% 16.2% 12.1% 6.8%

350-639 640-679 680-719 720-950

DTI20.1-25 19.0% 6.8% 3.1% 1.3%

DTI25.1-30 21.0% 7.4% 3.5% 1.4%

DTI30.1-35 22.2% 8.5% 4.1% 1.7%

DTI35.1-40 25.1% 10.1% 5.0% 2.1%

DTI40.1-45 26.5% 11.6% 6.3% 2.8%

DTI45.1-50 26.2% 12.7% 6.8% 3.6%

FICO

FICO

FICO

FICO

LTV20-75

LTV

75.1-85

LTV85.1-95

LTV>95

(Source: CoreLogic Servicing Data Set)


10/13page10

3 There is, o course, a substantial academic literature consisting o statistical models o delinquency, deault, and deault loss that supports the results presented

here and substantiates our concern with risk layering.

4 HUDMortgageeLetter10-29

Tese data extend the notion o layering. Mortgages

with higher DI ratios deault more requently

across the board. In this distressed period, LV was

a much less important protection against deaultthan were credit score and DI: Tere was much

more variation moving rom southwest to north-

east parts o the matrices in the table, holding LV

constant, than rom moving rom one matrix to the

other holding FICO and DI constant.

Te above is a quick survey o recent results or

deault rates.3 O course, it is not the last word, but it

does suggest that simply limiting high LV loans is

not as important a risk control as a careul analysiso all the components o risk and avoidance o risk

layering.

CONTROLLING RISK

Te FHA, i maintained by trained management

utilizing sound economic principles, can remain

sound and ulll its joint missions o: (1) serving as

the guarantor o last resort in a signicant housing

market crisis, and (2) insuring access to mortgage

credit or rst-time and minority home buyers.

Te problem remains that political oversight limits

the exibility o FHA management to adjust to the

trade-os involved in risk management. Historically,

FHA has responded to challenges only afer signi-

cant lags, during which losses and ineciencies were

rampant.

For example, today FHAs guidelines permit it to

give maximum nancing (96.5 percent LV) to an

individual with a 580 FICO score3 and a DI that

can reach 48 percent. Tis is an example o risklayering. While FHA continues to insure these types

o loans, other market participants have acknowl-

edged the layering associated with such allowances.

For instance, many lenders have imposed stricter

guidelines above those o FHA or high LV loans,

requiring a higher FICO score and limiting DI ra-

tios. In the private sector, 97 LV loans are generally

limited to a borrower with FICO scores o around

720 and more manageable DI ratios, such as being

below 41 percent. FHA does not have to mimic theprivate sector, but it does have to be able to adjust to

changes or it risks being selected against.

THINGSYOUCANTSEE:

Below we present a case study o hidden credit risk

and responses to inormation about it.

Seller-funded assistance to buyers

Seller-unded assistance takes two orms: up-ront or

closing cost assistance and down payment assistance

(note: 100 percent seller-unded down payments are

no longer permitted). Te FHAs experience with

each o these types o assistance is best understood

by discussing them separately, even though the les-

sons or the uture operation o FHA are similar.


11/13page 11

5

Iftherequiredminimumdownpaymentisdierentfrom3.5percent,justadjusttheexampleaccordingly.Notethatrelatives,closefriends,etc.maycontributeto the required down payment i they provide an appropriate git letter.

6 The monthly payment-to-income ratio may change as a result o this seller nance arrangement but, i purchase price rises to nance the added down

payment that will not be the case. Current HUD rules provide that the seller or a third party may contribute up to 6 percent o the sales price toward buyers

cost(See,BruceE.Foot,TreatmentofSeller-FundedDownPaymentAssistanceinFHA-InsuredHomeLoans,CongressionalResearchServiceReport7-5700,

RS22934.)

7 SeeRobertF.Cotterman,FinalReport:SellerFinancingofTemporaryBuydowns,Washington,D.C.:U.S.DepartmentofHUD.(1992).

8 SeeAustinKelly,SkinintheGame:ZeroDownpaymentMortgageDefault,JournalofHousingResearch,Vol17,#2,(2008)

9 Documentationofthestruggletoeliminateunsoundsellernancepoliciesisprovidedin:GeneralAccountabilityOce,(November2005)Mortgage

Financing:AdditionalActionNeededtoManageRisksofFHA-InsuredLoanswithDownpaymentAssistance(GAO-06-24)andSeller-FundedDownpayment

AssistanceinFHA-InsuredHomeLoans,CongressionalResearchServiceReport7-5700,RS22934.

10 Ibid, pg. 6

Seller-funded down payment assistance

to buyers

Seller-unded down payment assistance occurs whenthe seller either directly pays part o the required

minimum down payment or the seller pays a third

party who, in turn, helps the buyer pay the required

down payment. It is possible or a buyer to pay the

minimum required 3.5 percent down payment and

or the seller to pay another 1 percent so that the total

down payment is 4.5 percent o the purchase price.5 In

this case, the seller has nanced part o the down pay-

ment, but the borrower has met the minimum down

payment rom his/her own unds and could presum-ably have completed the transaction without the

sellers help.6 Such cases will not be counted as seller

nancing o down payments here.

Sellers would not willingly contribute to buyer down

payments or up-ront costs i they could sell their

houses or the same price without making these

payments. For example, i one buyer is willing to pay

$200,000 or a home and does not need or require

a seller contribution, most sellers will choose to sell

to that buyer rather than another who will only pay

$200,000 i the seller pays $6,000 in up-ront costs.

Tus, economic theory suggests that sellers who pay

a part o the down payment or up-ront costs are re-

ceiving a compensating increase in sales price above

what other buyers were willing to pay or the unit. In

this second case the property is probably only really

worth $194,000.

Research conducted by Robert Cotterman (1992)

ound that, or FHA mortgages, seller unding

resulted in a rise in deault rate that was equivalent

to the increase in LV associated with the amounto seller assistance, as compared to down payment

and up-ront cost.7 For example, the deault rate

on a mortgage with an LV o 98 percent with no

seller assistance was equivalent to the deault rate

on a mortgage with an LV o 95 percent that had

3 percent seller assistance. Te ndings o Cotter-

mans 1992 study have been conrmed by a number

o more recent papers, including a most persuasive

study by Kelly (2008).8

Afer several ailed attempts the practice o provid-

ing seller unding o the required down payment

was ended on January 1, 2009.9 However, it has had

a cost. Future homebuyers ultimately pay the price

given that the program is a mutual mortgage insur-

ance und, and losses rom periods o problematic

management imply higher premiums or uture buy-

ers. According to the latest annual audit, the Mutual

Mortgage Insurance (MMI) und had an economic

value o $5.2 billion at the end o scal year 2010,compared to an economic value o $21.3 billion at

the end o scal year 2007. Te audit report attri-

butes this 75.6 percent decline in value to two things:

1. A weakening o the housing market since

scalyear2007,and:

2. The concentration o loans receiving down

payment assistance rom nonprots10


12/13page 12

SELLER-FUNDED UP-FRONT OR

CLOSING COST ASSISTANCE TO BUYERS

Sellers or other parties may also contribute up to 6percent o the sales price to cover either additional

down payment (above the required amount), or

closing costs (homeowners insurance, basis points,

FHA up-ront insurance payment up to 1 percent

o the loan amount, ees, and taxes). Tese will all

be termed up-ront costs. Just as in the case o

down payment assistance, the same economic logic

concludes that sellers inate the sales price to cover

some portion o these up-ront costs.

On January 20, 2010, the FHA announced its intent

to reduce the allowable seller assistance rom 6

percent to 3 percent. At the same time, it announced

other measures to reduce expected losses, such as

increasing the minimum FICO score to qualiy or

3.5 percent down payment to 580 and increasing en-

orcement on FHA lenders. Te announcement also

included measures to enhance revenue by raising

the mortgage insurance premium. Te rationale or

reducing seller assistance rom 6 percent to 3 percent

was that the current level exposes the FHA to excess

risk by creating incentives to inate appraised value.

Tis change will bring FHA into conormity with

industry standards on seller concessions.11 Recently,

FHA announced that it hopes to publish a proposed

rule by year-end, afer postponing in midsummer.

WHAT CAN WE LEARN FROM THIS

REVIEW OF FHA POLICIES TOWARD

SELLER ASSISTANCE WITH DOWN

PAYMENTS AND CLOSING COSTS?

It has been dicult or FHA to adopt economically

sound policies to limit seller assistance even afer

overwhelming evidence demonstrated that it re-

sulted in misrepresented property values and led to

substantial losses or the insurance und. Te down

payment assistance policies were only changed in the

ace o losses that eroded reserves o the insurance

und. Te FHA has been unable to lower closing cost

assistance even at a time in which it has had to takeother dramatic steps to control losses and raise rev-

enues. Tis presents another reason or controlling

the volume o FHA insured lending in order to limit

the damage that these policies can do to current and

uture FHA borrowers (who pay higher premiums

and ace higher underwriting standards) as FHA at-

tempts to recover past losses at their expense.

11HUDNO.10-016,FHAAnnouncesPolicyChangestoAddressRiskandStrengthenFinances,(January20,2010)

We wish to thank Genworth Financial for contributing data tabulations for this report upon request,

as well as for support to our respective research centers.


13/13

ROBERT VAN ORDER

Robert Van Order is a member o the Department o Finance at Te George

Washington University School o Business. He is a proessor o nance, the

Oliver . Carr Proessor o Real Estate and Chair o the Center or Real Estate

and Urban Analysis. His areas o expertise are in nancial markets, housing

nance, housing policy and real estate economics.

Proessor Van Order has taught at Purdue University, the University o Southern

Caliornia, Queens University (Canada), American University, the University

o Caliornia, Los Angeles, Ohio State University, the University o Pennsylva-

nia, the University o Michigan and the University o Aberdeen (Scotland). Hewas an economist at the U.S. Department o Housing and Urban Development

(HUD), serving as director o the Housing Finance Analysis Division, and he

was chie economist at Freddie Mac.

Van Order has been a consultant to USAID, HUD, the World Bank and other

corporations, agencies and organizations, both public and private. A member o the American Economic Association and

the American Real Estate & Economics Association, he sits on the editorial boards o several prestigious journals. His work

on economics, housing and real estate both academic and general has been widely published. He received his Bachelors

degree rom Grinnell College, his Masters degree rom the University o Essex (England) and his PhD at Johns Hopkins

University.

ANTHONY YEZER

Anthony Yezer is a member o the Department o Economics o Te George

Washington University where he directs the Center or Economic Research.

He teaches courses in regional economics, urban economics, microeconomics,

and the economics o crime. He has been a Fellow o the Homer Hoyt School o

Advanced Studies in Real Estate and Urban Economics since 1991.

His articles have appeared in theJournal of Finance, Journal of Real Estate

Finance and Economics, Economic Inquiry, Journal of Law and Economics,

Real Estate Research, Journal of Economic Perspectives, Land Economics, Journal

of Economics and Business, Journal of Urban Economics, Journal of Regional Sci-

ence, Regional Science and Urban Economics, and Southern Economic Journal.

He edited the monograph Fair Lending Analysis: A Compendium of Essays on

the Use of Statistics, currently serves on the editorial boards o six journals and is

editor o the American Real Estate and Urban Economics Association

monograph series.

Proessor Yezer received a Bachelors degree rom Dartmouth College, continued his studies at the London School o

Economics and Political Science where he earned a M.Sc. degree, and holds a Doctoral degree rom the Massachusetts

Institute o echnology. He was a Rhodes Scholarship nalist and received a National Collegiate Athletic Association

Scholar-Athlete Fellowship.

AUHORS

FHA assessment report

Documents