Mortgage Debt Overhang: Reduced Investment by Homeowners at Risk of Default

• DOI: http://doi.org/10.1111/jofi.12482
• Date: 01 April 2017
• Published date: 01 April 2017

THE JOURNAL OF FINANCE •VOL. LXXII, NO. 2 •APRIL 2017

Mortgage Debt Overhang: Reduced Investment

by Homeowners at Risk of Default

BRIAN T. MELZER∗

ABSTRACT

Homeowners at risk of default face a debt overhang that reduces their incentive

to invest in their property: in expectation, some value created by investments in the

property will go to the lender.This agency conflict affects housing investments. Home-

owners at risk of default cut back substantially on home improvements and mortgage

principal payments, even when they appear financially unconstrained. Meanwhile,

they do not reduce spending on assets that they may retain in default, including home

appliances, furniture, and vehicles. These findings highlight an important financial

friction that has stifled housing investment since the Great Recession.

ALONGSTANDING AND IMPORTANT IDEA in finance theory is that leverage can dis-

tort investment decisions. Myers (1977) introduces the notion of corporate debt

overhang, arguing that borrowing introduces an agency conflict that curtails

productive investments. In public finance, Keynes (1920), Krugman (1988),

and Sachs (1990) point out that heavy public debt loads reduce incentives for

public sector investments in infrastructure and private sector investments in

physical and human capital. This paper applies the same thinking to house-

hold financial decisions and provides evidence that homeowners reduce their

housing investments when faced with mortgage debt overhang.

The reasoning behind debt overhang is straightforward. For an incremental

investment in a debt-free asset, the owner captures the investment’s payoffs

∗Brian T. Melzer is with the Kellogg School of Management, Northwestern University. I thank

Gene Amromin, Efraim Benmelech, Daniel Cooper, Kevin Davis, Xavier Giroud, Erik Hurst, Ravi

Jagannathan, Jiro Kondo, Lorenz Kueng, Lindsey Leininger, David Matsa, Edwin Mills, Laudo

Ogura, Jonathan Parker, Alessandro Previtero, Timothy Riddiough, Kenneth Singleton (the Ed-

itor), Joseph Tracy, and two anonymous referees for helpful comments. Seminar participants at

Duke University,Ivey Business School, Kellogg School of Management, Loyola University Chicago,

Stanford GSB, and the Federal Reserve Banks of Boston and Chicago also provided valuable feed-

back, as did conference participants at the Summer Real Estate Symposium, NBER Summer

Institute (Real Estate and Public Economics), Consumer Expenditure Survey Users’ Workshop,

and the annual meetings of the Midwest Economics Association, Western Finance Association,

and Financial Intermediation Research Society. I gratefully acknowledge ICPSR for providing the

Consumer Expenditure data files used in this study, Fiserv for providing Case-Shiller Home Price

Index data, and Laura Paszkiewicz from the BLS for help with the Consumer Expenditure data.

Aaron Yoon and Kexin Qiao provided helpful research assistance. I received no external finan-

cial support for this research. I have read the Journal of Finance’s disclosure policy and have no

conflicts of interest to disclose.

DOI: 10.1111/jofi.12482

575

576 The Journal of Finance R

in all states of the world. In contrast, for a levered asset with some risk of

default, the investment’s payoffs accrue to the foreclosing lender when the

owner defaults. Faced with the same investment outlay today and the prospect

of sharing the future payoffs with the lender, the owner of a levered asset may

underinvest, forgoing economically efficient investments.

Following the precipitous decline in home values in the United States be-

tween 2006 and 2011, mortgage debt overhang has become an important issue.

During this period, home values fell by 26% nationwide and by more than 50%

in some states, leaving many homeowners with negative home equity, that is,

mortgage liabilities that exceed the value of their home.1At the end of 2011,

15% of homeowners had negative home equity and another 7.5% of homeown-

ers had minimal home equity (less than 10% of home value). Borrowers in this

position default at much higher rates in the future and thus are subject to debt

overhang.

This analysis uses data from the Bureau of Labor Statistics’ Consumer Ex-

penditure Survey (CE) to investigate whether highly leveraged homeowners

respond to debt overhang by investing less in their home. The data contain

property-specific information for a national sample of homeowners, including

mortgage balances and principal payments, home improvement and mainte-

nance expenditures, and property values as estimated by the homeowners.

Using these inputs to measure property-level expenditures and to construct

each property’s mortgage loan-to-value ratio, I test whether spending varies

with leverage and default risk.

I find that homeowners with negative home equity spend $200, or 30%,

less per quarter on home improvements and maintenance relative to home-

owners with positive equity. Homeowners with negative equity also pay down

less of their mortgage principal: controlling for differences in mortgage bal-

ances, they cut unscheduled principal payments by $180, or 40%. These reduc-

tions in investment and debt repayment are not explained by differences in

income, total expenditures and financial wealth, nor are they explained by a

rich set of controls for household demographics, property characteristics, and

mortgage traits. Even for a matched sample of positive- and negative-equity

homeowners—among whom key characteristics such as age, income, wealth,

nonmortgage indebtedness, and length of ownership are very similar—the pres-

ence of negative equity predicts similar declines in housing investment.

In an additional test that ties the empirical model more directly to debt

overhang theory, I confirm that homeowners respond to default risk. The CE

lacks information on borrowers’ delinquency and perceived default risk, so I use

an additional data source to calibrate a mortgage default model and estimate

each CE respondent’s probability of default as a function of their mortgage

loan-to-value ratio. While the default risk estimate and the negative-equity

indicator are defined using the same underlying variation in loan-to-value

1These changes in home values are calculated using the Case-Shiller Indexes between December

2006 and December 2011. State-level declines of 45% to 60% occurred in Arizona, California,

Florida, and Nevada.

Mortgage Debt Overhang 577

ratios, the default risk model refines the identifying variation in indebtedness

in two ways. First, it ascribes some default risk to households with limited

positive equity, and second, it attributes substantially greater risk of default

to a homeowner that is deeply “underwater” compared to a homeowner that is

barely underwater. Consistent with debt overhang theory, improvements and

debt repayment both show a strong negative correlation with estimated default

risk.

The patterns in housing investment described above are consistent with debt

overhang theory but do not rule out other possible interpretations. For example,

heavily indebted homeowners may prefer to invest in their properties but lack

the liquidity or borrowing capacity to finance those expenditures (Hurst and

Stafford (2004), Mian and Sufi (2011), Cooper (2013), Mian, Rao, and Sufi

(2013)). In the balance of the paper I make careful use of the CE data to

test a range of predictions that distinguish debt overhang from alternative

hypotheses such as financial constraints.

I begin by showing that homeowners with high incomes and substantial fi-

nancial assets—who are unlikely to be financially constrained—reduce their

housing investments when they are in a negative-equity position. These facts

are consistent with debt overhang but are difficult to reconcile with an inter-

pretation based on financial constraints. Foreclosure law also affects wealthy

homeowners as debt overhang theory predicts. Many states allow mortgage

lenders to claim borrowers’ financial assets when the collateral falls short of

the loan balance, and homeowners in such “recourse” states are less likely

to default (Ghent and Kudlyak (2011)). In the CE data, I find that wealthy

households in recourse states reduce their housing investments only modestly

when they have negative equity. By contrast, when lenders lack recourse and

“strategic” default is more likely, wealthy homeowners reduce their housing

investments substantially when they are underwater.

The next test strengthens the case for debt overhang by showing that

negative-equity homeowners reduce spending specifically on those durable in-

vestments that are sacrificed to the lender in default. After controlling for the

same set of covariates as in the main analysis, spending on home durables (ma-

jor appliances and furniture) is similar between positive- and negative-equity

homeowners. Likewise, underwater homeowners spend similar amounts on

entertainment-related durables (for example, televisions and recreation equip-

ment), jewelry,and vehicles after accounting for other household characteristics

such as income, total expenditures, financial wealth, and indebtedness. When

taken together with the main results, this evidence strongly supports debt over-

hang relative to other explanations, since it demonstrates within-household

variation in spending that matches the predictions of debt overhang theory. Al-

ternative hypotheses that postulate a household-level difference among heavily

indebted homeowners—like a binding financial constraint, the expectation of

low future income, or a preference, not captured by observables, to invest little

in home-related and other durables—cannot easily explain the full set of facts.

The final pair of tests control for omitted household characteristics through

household fixed effects. I incorporate fixed effects in two ways. First, I