Credit Rationing, Income Exaggeration, and Adverse Selection in the Mortgage Market

• Author: BRENT W. AMBROSE,JAMES CONKLIN,JIRO YOSHIDA
• DOI: http://doi.org/10.1111/jofi.12426
• Date: 01 December 2016
• Published date: 01 December 2016

THE JOURNAL OF FINANCE •VOL. LXXI, NO. 6 •DECEMBER 2016

Credit Rationing, Income Exaggeration, and

Adverse Selection in the Mortgage Market

BRENT W. AMBROSE, JAMES CONKLIN, and JIRO YOSHIDA∗

ABSTRACT

Weexamine the role of borrower concerns about future credit availability in mitigating

the effects of adverse selection and income misrepresentation in the mortgage market.

We show that the majority of additional risk associated with “low-doc” mortgages

originated prior to the Great Recession was due to adverse selection on the part of

borrowers who could verify income but chose not to. We provide novel evidence that

these borrowers were more likely to inflate or exaggerate their income. Our analysis

suggests that recent regulatory changes that have essentially eliminated the low-doc

loan product would result in credit rationing against self-employed borrowers.

DURING THE GREAT RECESSION OF 2007 to 2008, the United States experienced

a massive increase in residential mortgage defaults and foreclosures not seen

since the Great Depression. For example, the Financial Crisis Inquiry Com-

mission reports that 9.7% of all mortgages were in default by the end of 2009,

compared to approximately 1% at the start of the decade.1While the decline

in house prices between 2007 and 2009 is obviously one of the primary causes

for the significant number of mortgage defaults registered during the crisis,

financial economists have only recently begun to examine the role of mortgage

fraud and adverse selection in exacerbating the consequences of the 2007 to

2009 housing bust. Evidence is mounting that the great mortgage expansion

that accompanied the rise in home prices coincided with increases in mortgage

fraud related to misrepresentations of borrower income (Jiang, Nelson, and

∗Brent W. Ambrose and Jiro Yoshida are with the Pennsylvania State University Smeal Col-

lege of Business. James Conklin is with the University of Georgia Terry College of Business. We

thank Itzhak Ben-David, Sumit Agarwal, Manuel Adelino, Christopher Palmer, Sam Kruger,Ken

Singleton (Editor), the anonymous Associate Editor and referees, and seminar participants at the

National University of Singapore, University of California–Berkeley,Research Institute of Capital

Formation, Housing Research and Advancement Foundation of Japan, Board of Governors of the

Federal Reserve System, University of Cambridge, Federal Reserve Bank of Atlanta Real Estate

Finance Conference, and the 2015 ASSA meeting in Boston for their helpful comments and sug-

gestions. We also thank the Penn State Institute for Real Estate Studies for providing access to

the New Century Mortgage database. We gratefully acknowledge research assistance from Sergio

Garate. The authors have no material financial interests related to this research, as identified in

the journal’s Disclosure Policy; furthermore, no party had the right to review this paper prior to

circulation.

1See U.S. Financial Crisis Inquiry Commission (2011, p. 215). Default is defined as “90-days or

more past due or in foreclosure.”

DOI: 10.1111/jofi.12426

2637

2638 The Journal of Finance R

Vytlacil (2014), Mian and Sufi (2015)), borrower assets (Garmaise (2015)), in-

flated appraisals (Ben-David (2011), Agarwal, Ambrose, and Yao (2014), Agar-

wal, Ben-David, and Yao (2015), Griffin and Maturana (2015)), and second

liens and owner-occupancy status (Piskorski, Seru, and Witkin (2015)).2As a

result, regulators and policy makers have implemented new rules to combat

perceived abuses in mortgage lending.3The purpose of this paper is to shed

light on how borrower heterogeneity with respect to employment status con-

tributed to income misrepresentation and adverse selection, and how lender

actions and borrower concerns about preserving future access to credit mit-

igated these risks. From a policy perspective, our results echo the concerns

raised by Keys et al. (2009), Rajan, Seru, and Vig (2010), and Piskorski, Seru,

and Vig (2010), among others, concerning the need to carefully weigh the costs

and benefits of new financial regulations.

With respect to income misrepresentation, we present several novel insights.

First, by comparing individual incomes within job titles, we provide new evi-

dence that strongly suggests that income misrepresentation was concentrated

primarily among borrowers who originated low-documentation loans but could

have easily originated full-documentation mortgages instead. Second, unlike

previous studies that show that misrepresentation in mortgage originations

resulted from lender actions at origination,4we find that income falsification

was essentially a borrower-level phenomenon.5Our results therefore suggest

that excesses in the mortgage market in the last decade resulted from both

borrower and lender actions. Third, we provide new evidence on lender actions

in response to potential borrower income falsification. Finally, in additional

analysis examining the role of borrower income falsification in facilitating the

expansion in mortgage credit, we provide new insights into one of the possible

causes of the Great Recession.

The role of borrower income misrepresentation leading up to the financial

crisis is a source of considerable debate. For example, Mian and Sufi (2009)and

Mian and Sufi (2015) argue that borrower income falsification was a leading

culprit in facilitating the expansion of mortgage credit during the 2002 to

2006 housing boom. Supporting this argument, Jiang, Nelson, and Vytlacil

(2014) show that income falsification occurred on low-documentation loans,

2In addition to mispresentation at the loan origination level, Piskorski, Seru, and Witkin (2015)

find that misrepresentation was endemic in the secondary market (between originators and in-

vestors) as well. Furthermore, Agarwal and Evanoff (2013) and Agarwal et al. (2014)provide

evidence of systematic predatory lending practices by loan originators. These practices may have

exacerbated the consequences of mortgage fraud.

3For example, the Consumer Financial Protections Bureau adopted the “Ability to Repay Rule”

that requires lenders to provide greater documentation of borrower income, and the Federal Hous-

ing Finance Agency in conjunction with the New York Attorney General’s office issued the Home

Valuation Code of Conduct (HVCC) that was designed to reduce the incidences of inflated ap-

praisals.

4Piskorski, Seru, and Witkin (2015) is a notable exception. The authors provide evidence that

borrowers misrepresented occupancy status on mortgage applications.

5Note that we are not arguing that the originator was unaware of the misrepresentation, but

rather that borrowers are complicit in falsifying income.

Credit Rationing, Income Exaggeration, and Adverse Selection 2639

resulting in elevated defaults, particularly for loans originated through the

wholesale channel. By focusing on differences in employment status, we show

that the majority of adverse selection and income falsification is confined to a

specific borrower group that was never intended to use the low-documentation

product. Our results therefore suggest that broad policies designed to eliminate

activities associated with excesses in mortgage originations during the housing

boom may have unintended consequences.

Since the potential for mortgage fraud and adverse selection has always been

present, lenders have long relied on underwriting guidelines to limit this risk.

However, Burke, Taylor, and Wagman (2012) illustrate how lender screening

to reject higher risk applicants results in greater adverse selection.6One such

underwriting metric is the debt-to-income (DTI) ratio, which limits the loan

amount based on the borrower’s income. This metric, in combination with the

loan-to-value (LTV) ratio, serves to limit the borrower’s housing consumption.

As a result, borrowers seeking to maximize their housing consumption or in-

vestment have an incentive to exaggerate their income in order to reduce their

DTI ratio and thereby qualify for a higher loan amount.

Recognizing the borrower’s incentive to circumvent these metrics, mortgage

lenders require proof of reported assets and incomes to verify that the borrower

is capable of repaying the debt. Of course, verification of borrower income and

assets comes at a cost. Not only do lenders bear the costs associated with

verification activities, but borrowers also bear costs of collecting and reporting

income and assets to the lender. For some borrowers, these costs are relatively

minor and involve simply submitting the prior two years’ W-2 tax documents

from their employer along with their past two months’ pay stubs. However, for

many other potential borrowers, the costs of verifying income and assets are

not so trivial. For example, self-employed individuals would need to provide

full tax returns for the previous two years. Yet, self-employed individuals often

file for tax return extensions due to the complexity of their tax situation, and as

a result, the returns are not available to the lender. Self-employed individuals

are also required to provide current profit and loss statements along with

bank statements for several months to prove sufficient cash flow to service the

debt. Furthermore, to comply with underwriting DTI guidelines, lenders may

require additional documentation from self-employed borrowers to determine

the nature of deposits and withdrawals and ascertain those expenses that are

personal versus those associated with their business.7

6However,the presence of adverse selection at mortgage origination is not universally accepted.

For example, Agarwal, Chang, and Yavas (2012) rely on differences in loan performance between

prime and subprime markets to claim that adverse selection was less severe in the subprime

market.

7Anecdotal discussions with mortgage brokers and other industry participants provide examples

of the verification costs self-employed borrowers face. For example, lenders may require that self-

employed borrowers provide written explanations for every deposit over the previous year. For a

business with just two transactions per week, that would necessitate over 100 separate written

explanations. Furthermore, many self-employed borrowers face serious confidentiality issues in

revealing client names as the source of deposits.