Consumer Ruthlessness and Mortgage Default during the 2007 to 2009 Housing Bust

• Date: 01 December 2017
• Author: HUI SHAN,NEIL BHUTTA,JANE DOKKO
• Published date: 01 December 2017
• DOI: http://doi.org/10.1111/jofi.12523

THE JOURNAL OF FINANCE •VOL. LXXII, NO. 6 •DECEMBER 2017

Consumer Ruthlessness and Mortgage Default

during the 2007 to 2009 Housing Bust

NEIL BHUTTA, JANE DOKKO, and HUI SHAN∗

ABSTRACT

From 2007 to 2009 U.S. house prices plunged and mortgage defaults surged. While os-

tensibly consistent with widespread “ruthless default,” analysis of detailed mortgage

and house price data indicates that borrowers do not walk away until they are deeply

underwater—far deeper than traditional models predict. The evidence suggests that

lender recourse is not the major driver of this result. We argue that emotional and

behavioral factors play an important role in decisions to continue paying. Borrower

reluctance to walk away implies that the moral hazard cost of default as a form of

social insurance may be lower than suspected.

FROM 2007 TO 2009 HOUSE PRICES in the United States plunged, especially in

many areas of Arizona, California, Florida, and Nevada (Figure 1). At roughly

the same time, mortgage defaults surged, consistent with the predictions of

option-theoretic models of mortgage pricing where borrowers “ruthlessly” de-

fault to maximize their wealth (Foster and Van Order (1984,1985)). Numerous

media anecdotes suggest that ruthless defaults were widespread, and the pre-

sumed wave of such defaults has led economists and policy makers to propose

ways to address ruthless or strategic defaulters. In this paper, we assess how

closely borrower behavior conforms to neoclassical models of default using rich

microdata where we know the precise timing of defaults and the evolution of

house prices at the ZIP code level.

We find that home equity has to turn deeply negative before most homeown-

ers will exercise their default “option”—much more so than the neoclassical

models predict. In particular, we estimate that the median borrower in our

∗Neil Bhutta is from the Federal Reserve Board and Hui Shan is from Goldman, Sachs &

Co. The bulk of the work for this paper was conducted when Dokko was at the Federal Reserve

Board and the Brookings Institution. Cheryl Cooper provided excellent research assistance. We

thank Sarah Davies of VantageScore Solutions for sharing estimates of how mortgage defaults

impact credit scores. We also thank Brian Bucks, Glenn Canner,Ronel Elul, Kris Gerardi, Andrew

Haughwout, Jerry Hausman, Benjamin Keys, Andreas Lehnert, Michael Palumbo, Karen Pence,

Monika Piazzesi, Stuart Rosenthal, Shane Sherlund, Bill Wheaton, and Paul Willen for helpful

comments and suggestions. The findings and conclusions expressed are solely those of the authors

and do not represent the views of the Federal Reserve System or the U.S. Government. The authors

have no financial conflicts of interest to disclose in conducting this research. An early version of

this paper was voluntarily submitted to the Federal Reserve’s Finance and Economic Discussion

working paper series, and was reviewed by other Federal Reserve Board staff prior to release in

the series.

DOI: 10.1111/jofi.12523

2433

2434 The Journal of Finance R

Figure 1. Home prices in selected metropolitan areas, January 1987 to August 2010.

This figure shows home price indices from Case-Shiller for four major metropolitan areas in the

four states that we study. The price indexes are adjusted for overall inflation using the all-urban

Consumer Price Index (CPI ) from the Bureau of Labor Statistics and are set to 100 in January

1987.

sample does not exercise the default option until his housing equity drops to

−74% (i.e., the cumulative loan-to-home-value (CLTV) ratio is 174%, which

equates to a loan balance—including first and second mortgages—of about

$348,000 on a $200,000 house). In contrast, the traditional frictionless option-

theoretic model predicts that default is certain once equity reaches about −20%

(Kau, Keenan, and Kim (1994)).

Although many papers demonstrate that negative equity makes default more

likely (e.g., Deng, Quigley, and van Order (2000), Bajari, Chu, and Park (2008),

Foote, Gerardi, and Willen (2008)), understanding whether the ruthless model

effectively characterizes consumer behavior requires further analysis. We de-

velop a novel strategy to estimate the distribution of threshold equity values

implied by the empirical relationship between equity and default. In a recent

survey that Guiso, Sapienza, and Zingales (2013) draw on, homeowners were

asked a series of questions to gauge how deeply underwater they would have

to be to walk away from their mortgage. For example, the survey asks, “If the

value of your mortgage exceeded the value of your house by $50,000, would you

walk away from your house even if you could afford to pay your monthly mort-

gage?” This type of question provides the intuition for our analysis, but we infer

Consumer Ruthlessness and Mortgage Default 2435

the threshold level of negative equity needed to induce default by examining

actual payment decisions of people who bought their homes at the peak of the

housing market in 2006 and started falling “underwater” soon thereafter. Al-

though many borrowers say they would not default in certain situations, they

may act differently.

The housing crisis has reignited a longstanding debate about whether mort-

gage borrowers ruthlessly exercise the default or put option. On the one hand,

Cunningham and Hendershott (1984) propose that transaction and “psychic”

costs could be substantial, and Foster and van Order (1985) provide initial em-

pirical evidence of an underexercise of default by underwater borrowers.1On

the other hand, Kau, Keenan, and Kim (1994) show that rational borrowers

may not default until equity falls well below zero even in the absence of trans-

action costs, and argue that the available evidence supports the frictionless

model.

Research on this topic can inform not just mortgage pricing questions, but

also policy issues concerning the penalties for default, regulatory considera-

tions, such as caps on loan-to-value ratios, and loss forecasting in the event

that house prices severely decline again (e.g., stress tests). Moreover, consider-

able improvements in data on loan performance and house prices, as well as in

econometric methods, can help answer these questions more precisely.

We focus primarily on the monthly payment status of over 125,000 nonprime

home purchase loans originated in 2006 in Arizona, California, Florida, and

Nevada, as well as a monthly measure of home equity computed from highly

disaggregated ZIP code level house price indexes (HPI). All of the borrowers in

this sample put no money down (a combined loan-to-value of 100) and many

purchased homes in ZIP codes where house prices declined by over 50% be-

tween 2006 and 2009. These loans were packaged into private-label mortgage

securities and defaulted at an extremely high rate, and have been a central

focus of research on the triggers of the financial crisis. If borrowers typically

refrain from walking away until their equity falls deeply and they are severely

underwater, this would imply that borrower behavior largely deviates from

that predicted by traditional option-theoretic models.

Because our main sample is not representative of the population of home

buyers, we also examine a broader set of loans from an alternative data set

that includes loans sold to Fannie Mae and Freddie Mac, loans held in banks’

portfolios, and loans with modest down payments. If borrowers in this sample

also tend not to default until falling deeply underwater, this would further

support the view that mortgage borrowers’ behavior generally deviates from

the traditional model.

Our methodology addresses a key empirical challenge, namely, that default

can occur because of liquidity shocks that necessitate default and lead to in-

correct inferences about walk-away thresholds. If, for example, we observe an

individual defaulting when his CLTV ratio hits 125% (or, equivalently, equity

1Subsequent work by Quigley and van Order (1995) also points to an underexercise of the

default option.