Journal of Monetary Economics 76 (2015) 173–190

Contents lists available at ScienceDirect

Journal of Monetary Economics

journal homepage: www.elsevier.com/locate/jme

Mortgage defaults
Juan Carlos Hatchondo a,n, Leonardo Martinez b, Juan M. Sánchez c
a
Department of Economics, Indiana University, 100 South Indiana Ave. Bloomington, IN 47405, United States
b
International Monetary Fund, 700 19th Street NW, Washington, DC 20431, United States
c
Federal Reserve Bank of St. Louis, 1 Federal Reserve Bank Plaza, St. Louis, MO 63102, United States

a r t i c l e i n f o abstract

Article history: A life-cycle model is developed in which households face income and house-price risk and
Received 1 August 2014 buy houses with mortgages. This model, which accounts for key features in U.S. data, is
Received in revised form used as a laboratory for prudential policy. Recourse mortgages increase the cost of default
28 September 2015
but also lower equity and increase payments. The effect on default is nonmonotonic. Loan-
Accepted 30 September 2015
Available online 22 October 2015
to-value (LTV) limits increase equity and lower the default rate, with negligible effects on
housing demand. Combining recourse mortgages and LTV limits reduces the default rate
Keywords: while boosting housing demand. Together, they also prevent spikes in default after large
Mortgage declines in aggregate house prices.
Default
& 2015 Elsevier B.V. All rights reserved.
Recourse
LTV
Housing risk

1. Introduction

The increase in U.S. mortgage defaults observed since 2006 moved the stability of mortgage markets to the center of
policy debates (Campbell, 2012; FED, 2012; Treasury, 2009), invigorating academic and policy discussions about prudential
policies that could prevent defaults. Two prudential policies have received widespread consideration: recourse mortgages,
which allow lenders to collect from debtors after a default, and loan-to-value (LTV) limits on new mortgages.1
This paper evaluates recourse mortgages and LTV limits in the light of a life cycle model with housing and non-durable
consumption, idiosyncratic shocks to labor earnings and the price of housing, and mortgages. Households can consume
housing services by renting or owning the house they live in, and they can buy houses of different sizes. They can borrow to
buy a house using a long-term collateralized defaultable mortgage. A defaulting household must move out of the house used
as collateral. There is a deadweight cost of liquidating houses in foreclosure. Households can also refinance their mortgage
loans and save using a risk-free asset. There is room for policy interventions because households have limited commitment
and markets are incomplete.
The model generates plausible predictions for the households' demand for housing, demand for mortgages, and mortgage
default decisions. Idiosyncratic income and housing-price stochastic processes are parameterized using previous estimations

n
Corresponding author. Tel.: þ 1 812 856 1202.
E-mail address: [email protected] (J. Carlos Hatchondo).
1
IMF (2011) discusses the widespread use of these policies across countries. It is often argued that recent housing-price declines had a much larger
effect on mortgage defaults in the United States than in Europe in part because of soft U.S. recourse policies (Hatchondo et al., 2013; IMF, 2011; Feldstein,
2008). Wong et al. (2011) present empirical evidence that, for a given fall in house prices, the incidence of mortgage default is higher for countries without
an LTV limit than for countries with an LTV limit. Several studies document the important effects of the origination LTV on the probability of mortgage
defaults (Mayer et al., 2009; Schwartz and Torous, 2003).

http://dx.doi.org/10.1016/j.jmoneco.2015.09.011
0304-3932/& 2015 Elsevier B.V. All rights reserved.
174 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190

obtained with U.S. data. The calibration targets the homeownership rate, the median house price, the median ratio of financial
assets to income, and the median down payment. The model also generates plausible implications for other indicators of the
demand for housing (the life cycle profiles of ownership and house prices), the use of mortgages (home equity, mortgage
payments, and the distribution of mortgage down payments), and the mortgage default rate. The overall match between the
model predictions and the data makes the model a good laboratory for the quantitative evaluation of policies.
Recourse mortgages are assumed to allow lenders to garnish some of the cash-in-hand wealth (income and financial
assets) of a household that defaults. Results are presented for 15 different recourse policies that differ in three dimensions:
the level of cash-in-hand wealth exempt from attachment, the maximum percentage of cash-in-hand wealth subject to
attachment, and the (expected) duration of the attachment period.
These results indicate that it may be difficult to reduce the rate of mortgage defaults significantly with recourse mort-
gages. Furthermore, an implementable recourse policy (mild enough to be consistent with bankruptcy law) may fail to
prevent a sharp increase in the mortgage default rate after a large aggregate decline in the price of housing and may even
exacerbate the increase in the default rate.
The effect of recourse on the equilibrium default rate is nonmonotonic.2 On the one hand, a harsher recourse policy
makes defaults more costly, reducing the probability of a default in any mortgage. On the other hand, in our model with
endogenous choice of down payment and equilibrium pricing of mortgage interest rates, a harsher recourse policy may
increase the LTV chosen by households and, therefore may increase the default rate.
The effect of recourse on the demand for housing is also nonmonotonic. This effect follows the effect of recourse on LTVs,
as higher (lower) LTVs allow for more (less) housing consumption. On the one hand, a harsher recourse policy lowers the
cost of high-LTV mortgages and thus may lead households to choose higher LTVs. On the other hand, a harsher recourse
policy may force households with adverse income shocks to reduce their consumption in order to stay current on their
mortgages payments and avoid a default. This may lead households to choose mortgages with lower LTVs to prevent these
costly adjustments. The latter effect dominates for very harsh recourse policies, dampening the demand for housing.
The relationship between recourse and welfare follows the one between recourse and the demand for housing. In our
model, the households' ability to default implies endogenous borrowing constraints. Recourse mortgages may relax these
constraints, producing welfare gains (default decisions need not be optimal from an ex-ante perspective).
The findings described above indicate that the implementation of recourse mortgages may present difficulties. A
recourse policy that is not harsh enough would increase default and a recourse policy that is harsh enough to significantly
lower the default rate may end up reducing the boost to housing consumption implied by recourse mortgages (as pointed
out by Campbell, 2012, the main stated goal of much U.S. housing policy is to increase the homeownership rate). Fur-
thermore, bankruptcy laws could prevent the implementation of very harsh recourse policies.
Since the difficulty of using recourse mortgages to reduce the default rate is the result of high origination LTVs, this
problem could be mitigated by imposing LTV limits. It is first shown that LTV limits lower the default rate with negligible
effects on the demand for housing. For instance, comparing simulations for the benchmark economy with those for a model
economy with an 80 percent LTV limit shows negligible differences in housing consumption, while the LTV-limit economy
features a default rate 70 percent lower than the one in the benchmark. Nevertheless, an 80 percent LTV limit may be
insufficient to prevent a sharp increase in the mortgage default rate after large declines in the aggregate price of housing.
The mild effect of LTV limits on homeownership sheds light on important policy debates. For instance, in the U.S.,
qualified residential mortgage rules make higher down payments necessary for originators to fully securitize and sell the
mortgage, which in turn would result in lower interest rates for borrowers. Critics argue that these rules could have sig-
nificant negative effects on housing demand (see, for example, MBA, 2011). Since these rules can be viewed as a flexible LTV
limit for new mortgages, our results cast doubt on these arguments.
There may also be important complementarities between recourse mortgages and LTV limits. Economies with both
recourse mortgages and LTV limits feature a lower default rate and higher housing consumption than the benchmark, thus
achieving the two most-cited goals of mortgage policies. Furthermore, combining recourse mortgages and LTV limits is
necessary to greatly reduce the increase in the mortgage default rate that follows after large declines in the aggregate price
of housing.

1.1. Related literature

Our model incorporates housing, idiosyncratic housing-price risk, and mortgages into the class of models used in quantitative
studies of households' earnings risk (see Kaplan and Violante, 2010, and the references therein). Our model also extends the
equilibrium default model used in quantitative studies of credit card debt (Athreya, 2005; Chatterjee et al., 2007) by studying
collateralized long-term debt (mortgages) and shocks to the price of the collateral (housing). Some studies of credit card debt
focus on the effects of changes in the severity of bankruptcy penalties or income garnishment, which is comparable to our
discussion on the effects of recourse mortgages (Athreya, 2008; Athreya et al., 2011; Chatterjee and Gordon, 2012; Li and Sarte,

2
This nonmonotonicity may account for the mixed evidence on the effect of recourse on mortgage defaults. See Clauretie (1987); Ghent and Kudlyak
(2011), and the references therein.
 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190 175

2006; Livshits et al., 2007). Studying collateralized debt allows us to consider LTV limits as an alternative default-prevention
policy and discuss important complementarities between recourse mortgages and LTV limits.
Recent studies discuss the effects of recourse mortgages. Quintin (2012) shows that recourse mortgages may increase
mortgage defaults by changing the pool of borrowers in a model economy with asymmetric information. In this paper, there
is a nonmonotonic relationship between the degree of recourse and mortgage default. Furthermore, the mechanism through
which a harsher recourse policy increases the default frequency in our environment is different from the one presented by
Quintin (2012). In addition, while Quintin (2012) presents a theoretical discussion of the effects of recourse mortgages, this
paper shows that it is possible for recourse to increase mortgage defaults in a quantitative model that matches several
features of the data.
Corbae and Quintin (2015) study the role of the introduction of mortgage contracts with low down payments in
accounting for the recent rise in U.S. mortgage defaults. They present an exercise showing that introducing recourse
mortgages lowers the mortgage default rate and increases home ownership. Analyzing a larger set of recourse policies in a
richer model of origination LTVs allows us to present the nonmonotonic effect of the degree of recourse on default and
ownership. Corbae and Quintin (2015)'s recourse mortgages are such that a fraction of financial assets can be confiscated in
the default period. In contrast, this paper assumes that cash in hand (both financial assets and earnings) can be confiscated.3
Mitman (2012) presents a quantitative study of the interactions among recourse mortgages, bankruptcy, and mortgage
defaults across U.S. states. He finds that recourse mortgages have only a small effect on U.S. mortgage defaults because
households can file for bankruptcy. This is consistent with using a benchmark model without recourse mortgages to study
the U.S. economy as done, for instance, by Corbae and Quintin (2015) and in this paper.4 Mitman (2012) also finds that
nonrecourse is the optimal policy. This is in sharp contrast to the gains from introducing recourse mortgages presented here
and by Corbae and Quintin (2015). In contrast with Mitman (2012), this paper (and Corbae and Quintin, 2015) assumes that
it is costless for a lender to obtain and enforce a judgment against a defaulting household and that the debt of the defaulting
household implied by the recourse policy is no dischargeable in bankruptcy. This paper also departs from Mitman (2012) on
the assumed duration of mortgage contracts and the assumed nature of housing shocks (following previous studies,
Mitman, 2012, assumes one-period mortgages and models shocks to the house value as depreciation shocks that affect the
services a household obtains from its house without affecting the price of housing in the economy).
Campbell and Cocco (2015) study a framework with an exogenous origination LTV and present comparative statics with
respect to that variable. They show that higher origination LTVs are related to higher probabilities of mortgage defaults. This
paper presents a model that features endogenous LTVs, and shows that the distribution of LTVs generated by this model is
consistent with the one in the data. Thus, our model is well suited to study the effects of LTV limits (because these limits do
not change the LTV chosen by all households in the model economy). For instance, our model allows us to discuss the effects
of LTV limits on homeownership, a key element of policy debates.
The rest of the paper is organized as follows. Section 2 presents the model. Section 3 presents the recursive formulation
of the model. Section 4 discusses our calibration. Section 5 shows that predictions of the benchmark model (without
prudential regulations) fit the data. Section 6 compares economies with different prudential regulations in place. Section 7
concludes.

2. The model

In contrast to models presented in previous quantitative studies of idiosyncratic earnings risk, in this model (i) in
addition to consuming nondurable goods, the household consumes housing; (ii) in addition to idiosyncratic earning shocks,
the household faces idiosyncratic shocks to the price of housing; and (iii) borrowing options are endogenously given by
lenders' zero-profit conditions on mortgage contracts. The household lives T periods and works until age t ¼ W r T.5 Let
β denote the subjective discount factor.
At the beginning of the period, the household observes the realization of its earnings and housing-price shocks. After
observing these shocks, the household makes its housing and financial decisions.

2.1. Housing

This is a stylized model of housing that follows closely that of Campbell and Cocco (2003). The household must live in a
house and, in any given period, it may own up to one house.
In contrast to the model presented by Campbell and Cocco (2003), this model (i) allows the household to choose whether
to own or rent the house it lives in and (ii) incorporates houses of different sizes. The latter allows us to account for the

3
Limiting confiscation to only assets or income would imply introducing an additional state variable into our model, increasing the computational cost
significantly. Confiscation of income is often a legal possibility with recourse mortgages. For instance, in U.S. states with recourse mortgages, an employer
can be forced to withhold an employee's income exceeding exempted wages.
4
Chatterjee and Eyigungor (2015); Guler (2015), and Jeske et al. (2013) present other recent quantitative studies of mortgage defaults but do not
discuss policies that could mitigate defaults.
5
Previous versions of this study assume stochastic death and find essentially the same results.
176 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190

increasing life cycle profile of the mean house price observed in the data. If the household owns a house, it must live in the
house it owns. For simplicity, the household does not need to pay rent if it chooses to be a renter. This assumption guar-
antees that the household is always nable to affordohousing. There are M þ1 house sizes and the house with the smallest size
R
is the one available for rent. Let hA h ; h1 ; …; hM denote a house size, where hR is the size of the house available for rent.
The utility u derived from consumption c and from living in a house of size h displays a constant elasticity of substitution
between the two goods:
h ið1  γÞ=ð1  1=αÞ
1  1=α
ð1  θÞc1  1=α þ θh
uðc; hÞ ¼ ;
1 γ
where γ denotes the risk aversion parameter, α governs the degree of intra-temporal substitutability between housing and
nondurable consumption goods, and θ determines the expenditure share for housing.
The price per housing unit for household i is given by pit. This price changes stochastically over time. The transaction cost
i i
of buying a house of size h is ξB hpt , and the transaction cost of selling a house of size h is ξS hpt .

2.2. Earnings and housing-price stochastic processes

Both the price of housing and earnings follow exogenous processes. Each period, household i receives income yit. During
working age, income has a fixed effect fi, a life cycle component lt, an i.i.d component εit, and a persistent component zit:
i
logðyit Þ ¼ f þ lt þ εit þzit ;
where εit is normally distributed with variance σ 2ε , and

zit ¼ zit  1 þ eit :

After retirement, the household receives a fraction of the last realization of the persistent component of its working-age
income. To mimic the U.S. Social Security retirement benefits, this fraction is a decreasing function of pre-retirement
income:
i
replacement ratio ¼ maxfA0 þ A1 expðf þ lW þ ziW Þ; A2 g:
As is standard in the housing literature, housing-price shocks are an autoregressive process, and earnings and the price of
housing are correlated.6 In particular, following Nagaraja et al. (2011), the log of the housing price is assumed to follow an AR
(1) process:

log ðpit Þ ¼ ð1  ρp ÞlogðpÞ þρp logðpit  1 Þ þνit ; ð1Þ

2
where p is the mean price, and eit and νit are jointly normally distributed with correlation ρe;ν and variances σe and σ 2ν .

2.3. Mortgage contracts and savings

Financial intermediaries are risk neutral and make zero profits in expectation. Their opportunity cost of lending is given
by the interest rate r. The household can save using one-period risk-free asset that pays a constant interest rate and can
finance housing consumption with mortgages.
A mortgage for a household of age t is a promise to make payments for the next T t periods or to prepay its debt in any
period before T. Mortgage payments decay at rate δ. This allows us to account for the decline in the real value of mortgage
payments due to inflation. To prepay its mortgage, the household must pay the value of the remaining payment obligations
discounted at the rate r, q ðnÞb, where n ¼ T t, b denotes the current-period mortgage payment, and
 
1δ nþ1
1
1þr
q ðnÞ ¼ for n Z 1: ð2Þ
1δ
1
1 þr
Computing the prepayment amount using the opportunity cost of lending r instead of using the mortgage interest rate
allows us to economize a state variable (which is important given the high computational cost of our exercises). This is
unlikely to have significant quantitative effects: the mortgage interest rate is given by the opportunity cost of lending plus a
default premium, and the majority of households choose down payments high enough to make the default premium
negligible (which is consistent with the data). Note that since every period the household can prepay its mortgage and ask
for a new mortgage, the household can choose a decreasing or increasing pattern of mortgage payments and change the
effective duration of its mortgages. There is a limit to the mortgage origination LTV and a fixed cost ξM of signing a mortgage
contract. Mortgage loans are the only loans available to the household.

6
Thus, there is predictability in house prices as modeled by Corradin et al. (2014) and Nagaraja et al. (2011).
 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190 177

Fig. 1. Household choices. Note: R; B; SH ; SR ; F and P denote interim value functions.

The household can default on its mortgage. If the household chooses to default, it hands its house over to the lender, who
sells it with a discount at price phð1 ξ S Þ, with 0 r ξ S r 1. The household must rent while in defaults.
Recourse mortgages are such that if the proceeds from the foreclosed house sale are not enough to pay the household's
debt (phð1 ξ S Þ oq ðnÞb), the lender may be able to garnish some of the household's cash-in-hand wealth (income and
financial assets). Each recourse period in which the household has not paid the totality of its debt, the lender can garnish the
lesser of (i) a share κ of the household's cash-in-hand wealth, and (ii) the household's cash-in-hand wealth in excess of an
exempted amount.7 The exempted amount ϕw ~ is expressed as a share ϕ of the median cash-in-hand wealth w.~ Thus, each
period in which the household is in default it must transfer to the lender

~ q ðnÞb phð1  ξ S Þg; 0g;

Φðh; b; w; p; nÞ ¼ maxfminfκw; w  ϕw;

where w ¼ expðf þln þ z þεÞ þ a Z 0 denotes the household's cash-in-hand wealth (labor income plus savings) at the
beginning of the period.
To economize a state variable, the duration of the recourse period is stochastic. A defaulting household starts making
recourse payments Φ in the default period, and each period it may be freed from recourse payments with probability ψ. The
household also exits recourse when it finishes paying its debt. After exiting recourse, the household regains the option of
becoming a homeowner.

3. Recursive formulation

The household can enter each period as (i) a nonhomeowner with clean credit who can choose whether to buy a house,
(ii) a homeowner, or (iii) a defaulter (who defaulted in a previous period and did not pay its debt yet). Fig. 1 presents
household choices in each of these three situations and the corresponding value functions.

3.1. Nonhomeowner

If the household does not own a house and is not in default, it must choose whether to stay as a renter or buy a house.
Thus, the lifetime utility of this household is given by

Nðw; z; p; nÞ ¼ max fI rent Rðw; z; p; nÞ þ ð1  I rent ÞBðw; z; p; nÞg; ð3Þ

I rent A f0;1g

where R denotes the lifetime utility of a nonhomeowner who decides to stay as a renter during the period, and B denotes the
lifetime utility of a nonhomeowner that buys a house in the period.

7
In most countries, recourse allows for collection of personal assets and future income. To economize a state variable, our recourse rule does not
distinguish assets from income and thus uses cash-in-hand wealth. Our recourse rule features both means testing through exemptions and a limit to the
share of resources that can be taken by lenders, both common characteristics of attachment rules. For instance, in the U.S. wages below a legal amount
(30 h per week multiplied by the prevailing minimum wage) cannot be withheld. If wages exceed that amount, the employer is obligated to withhold the
lesser of the wages in excess of the exempted amount and 25 percent of the wages. Similar exemptions exist for attachment of personal assets.
178 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190

3.2. Renter

A household that enters the period not owning a house and chooses to continue renting can choose only its next-period
savings a0 Z 0. Thus, the value function R is determined as follows:
n   o
u c; h þβE½Nðw0 ; z0 ; p0 ; n  1Þ∣z; p ;
R
Rðw; z; p; nÞ ¼ max
0 a Z0
0
a
s:t: c ¼ w
1þr
w0 ¼ exp ðf þln  1 þz0 þε0 Þ þ a0 : ð4Þ

3.3. Buyer

0
A household that decides to buy a house must choose the size of the house (h ), the amount of savings (a0 ), and the
0 0 0 0
amount borrowed. The latter is given by b qðb ; a0 ; z; p; h ; nÞ, where b denotes the next-period mortgage payment and q is
defined in Section 3.5. The expected discounted lifetime utility of a buyer satisfies

 0  0 0 
Bðw; z; p; nÞ ¼ 0 max 0 u c; h þβE Hðh ; b ; w0 ; z0 ; p0 ; n  1Þ∣z; p ð5Þ
f b Z 0;a0 Z 0;h
g

0  0 0  a0 0
s:t:c ¼ wþ b q h ; b ; a0 ; z; p; n  I b0 4 0 ξM   ð1 þ ξB Þph ;
1 þr
0 0 0 0
w ¼ exp ðf þln  1 þz þε Þ þ a ;
0 0 0 0
b qðh ; b ; a0 ; z; p; nÞ r λph ;
0

h A h1 ; …; hM ; ð6Þ

where the indicator I b0 4 0 takes a value of 1 if the household buys the house with a mortgage and of 0 otherwise, and H
denotes the expected discounted lifetime utility of a household that enters the period as a homeowner. Eq. (6) imposes a
mortgage LTV limit (λ).

3.4. Homeowner

A household that enters the period as a homeowner can (i) pay its current mortgage (if any), (ii) refinance its mortgage
(or ask for a mortgage if it does not have one), (iii) default on its mortgage, or (iv) sell its house (and buy another house or
rent). Thus, the value function H is given by the maximum of the values of these four options denoted by P, F, D, and S,
respectively:
Hðh; b; w; z; p; nÞ ¼ max fI P PðÞ þ I F FðÞ þ I D DðÞ þ I S SðÞg:
I P A f0;1g;I F A f0;1g;I D A f0;1g;I S A f0;1g
s:t: 1 ¼ I P þ I F þI D þ I S : ð7Þ
0
Payer: If the household makes the current-period mortgage payment, its only remaining choice is a . Then, the value of
making the mortgage payment is given by

 
P ðh; b; w; z; p; nÞ ¼ max
0
uðc; hÞ þ βE Hðbð1 δÞ; w0 ; z0 ; p0 ; h; n  1Þ∣z; p
a Z0
a0
s:t: c ¼ w b  ;
1þr
w0 ¼ exp ðf þ ln  1 þz0 þ ε0 Þ þ a0 : ð8Þ
Refinancer: To refinance, the household must prepay its mortgage and choose the next-period payment of its new
0 0
mortgage b Z0 (the household can also choose to not have a mortgage, b ¼ 0). The household is also free to adjust its
financial wealth. Thus, the value of refinancing is given by

 0 
Fðh; b; w; z; p; nÞ ¼ 0 max uðc; hÞ þ βE Hðh; b ; w0 ; z0 ; p0 ; n  1Þ∣z; p ð9Þ
b Z 0;a0 Z 0

 0 0  0 a0
s:t: c ¼ y q ðnÞbþ q h ; b ; a0 ; z; p; n b I b0 4 0 ξM  ;
1 þr
w0 ¼ exp ðf þln  1 þz0 þε0 Þ þ a0 ;
0 0 0
b qðh ; b ; a0 ; z; p; nÞ r λph: ð10Þ
Defaulter: If the household defaults, it is still free to adjust its financial wealth. A defaulting household becomes a renter
until it exits the recourse period. Thus, the value of defaulting is given by
8  
> R 0
< u c; h þβE½Nðw0 ; z0 ; p0 ; n  1Þ∣z; p if I rm b ¼ 0
Dðh; b; w; z; p; nÞ ¼ max    ð11Þ
a0 Z 0 > R 0
: u c; h þβE ψNðw0 ; z0 ; p0 ; n  1Þ þ ð1 ψÞDð0; b ; w0 ; z0 ; p0 ; n  1Þ∣z; p otherwise;
 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190 179

a0
s:t: c ¼ w  Φðh; b; w; p; nÞ  ; ð12Þ
1 þr

w0 ¼ exp ðf þ ln  1 þ z0 þ ε0 Þ þa0 ;
0 ½q ðnÞb phð1  ξ S Þ  Φðh; b; w; p; nÞð1 þ rÞ
b ¼ ; ð13Þ
q ðn  1Þ
where Irm is an indicator function equal to 1 (0) if the mortgage is (not) a recourse mortgage.
Seller: If the household sells its house, it can become a renter or it can buy another house. Thus, the value of selling the
house is given by
n o
Sðh; b; w; z; p; nÞ ¼ max I rentS SR ðh; b; w; z; p; nÞ þ ð1 I rentS ÞSH ðh; b; w; z; p; nÞ ;
I rentS A f0;1g

where SR denotes the expected discounted lifetime utility of selling the house and becoming a renter, and SH denotes the
expected discounted lifetime utility of selling the house and buying another house.
If the seller chooses to become a renter, it can adjust only its financial wealth. Thus, its lifetime utility is given by
n   o
u c; h þ βE½Nðw0 ; z0 ; p0 ; n  1Þ∣z; p
R
SR ðh; b; w; z; p; nÞ ¼ max
0
ð14Þ
a Z0

a0
s:t: c ¼ w  q ðnÞb þ phð1  ξS Þ  ;
1 þr
0 0 0 0
w ¼ exp ðf þ ln  1 þ z þ ε Þ þ a : ð15Þ
If the seller buys another house, it must also choose the size of the new house and the new mortgage. Thus, the seller's
lifetime utility is given by

 0  0 0 
SH ðh; b; w; z; p; nÞ ¼ 0 max 0 u c; h þ βE Hðh ; b ; w0 ; z0 ; p0 ; n 1Þ∣z; p ð16Þ
fb Z 0;a0 Z 0;h g

0  0 0  0 a0
s:t: c ¼ w  q ðnÞb þ phð1  ξS Þ þb q h ; b ; a0 ; z; p; n  I b0 4 0 ξM  ð1 þξB Þph  ;
1þr
0 0 0 0
w ¼ exp ðf þ ln  1 þ z þ ε Þ þa ;
0 0 0 0
b qðh ; b ; a0 ; z; p; nÞ r λph ; ð17Þ

0

h A h1 ; …; hM : ð18Þ

3.5. Mortgages

0
When the household asks for a mortgage that promises to pay b next period, the amount it borrows is given by
0 0 0
b qðh ; b ; a0 ; z; p; nÞ, where
 0 0  qpay þ qprepay þ qdefault
q h ; b ; a0 ; z; p; n ¼ ð19Þ
1 þr
and
h   0 0  i
qpay ¼ E I^P ðh ; b ; w0 ; z0 ; p0 ; n  1Þ 1 þ ð1  δÞq h ; b ð1  δÞ; a00P ; z0 ; p0 ; n  1 jz; p ;
0 0

hh i i
qprepay ¼ E I^F ðh ; b ; w0 ; z0 ; p0 ; n  1Þ þ I^ S ðh ; b ; w0 ; z0 ; p0 ; n  1Þ q ðn  1Þjz; p ;
0 0 0 0

2 h i 3
^ 0 0 0 0 0 0 0 0 0 0 0 ″ D ″ ″ 0
6I D ðh ; b ; w ; z ; p ; n  1Þ p h ð1  ξ S Þ þ Φðh ; b ; w ; p ; n  1Þ þ I rm bD q ðbD ; wD ; z ; n 1Þ 7
qdefault ¼ E4 0 jz; p5:
b

In the expressions above, I^ P , I^ F , I^ S , and I^D denote the optimal choice of a homeowner (i.e., the solution to problem (7) above),
0
a″P ¼ a^ P ðh; b ; w0 ; z0 ; p0 ; n  1Þ denotes the optimal saving choice of a household that pays its mortgage next period (i.e., the
0
solution to problem (8) above), w″D ¼ exp ðf þ ln  2 þz″ þ ε″ Þ þ a^ D ðh; b ; w0 ; z0 ; p0 ; n 1Þ denotes the cash-in-hand wealth two
periods ahead given next-period shocks and the optimal savings of a household that defaults next period
0
a″D ¼ a^ D ðh; b ; w0 ; z0 ; p0 ; n  1Þ,
0 0 0
″ ½q ðn  1Þb Φðh ; b ; w0 ; p0 ; n  1Þð1 þ rÞ
bD ¼
q ðn  2Þ
denotes the debt in two periods for a household that defaults next period, and
" #
0 ″ ″
D
 0 0  Φð0; b ; w0 ; 0; n 1Þ þ b qD ðbD ; w″D ; z0 ; n 1Þ
q b ; w ; z; n ¼ ð1  ψ ÞE 0
ð1 þ rÞb
180 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190

Table 1
Parameter values.

Parameter Value Definition Basis

a0 0:65y0 Initial wealth SCF

σ 2ν 0.302 Variance of ν Campbell and Cocco (2003)
ρe;ν 0.115 Correlation e and ν Campbell and Cocco (2003)
ρp 0.970 Persistence in p Nagaraja et al. (2011)
l – Income, life-cycle component Kaplan and Violante (2010)
σ 2ε 0.0630 Variance of ε Kaplan and Violante (2010)
2
σe 0.0166 Variance of e Kaplan and Violante (2010)
f þ  0:459 Income fixed effects Storesletten et al. (2004)
r 0.030 Risk-free rate Standard in the literature
γ 2.000 Risk aversion Standard in the literature
α 0.500 Elasticity of substitution Standard in the literature
ξB 0.030 Cost of buying, households Gruber and Martin (2003)
ξS 0.030 Cost of selling, households Gruber and Martin (2003)
ξS 0.220 Cost of selling, bank Pennington-Cross (2006)
ξM 0.150 Cost of signing mortgage U.S. Federal Reserve
δ 0.020 Payments decay Average inflation
A0 0.7156 Replacement ratio U.S. Social Security
A1  0.040 Replacement ratio U.S. Social Security
A2 0.140 Replacement ratio U.S. Social Security
ϕ 1 Recourse exemption No recourse
Irm 0 Recourse indicator No recourse
λ 1 LTV limit Positive down payment

0 0
is such that b qD ðb ; w0 ; z; nÞ denotes the secondary market price of a mortgage in default with current-period promised
0
payments given by b (i.e., the secondary market price of the right to collect the recourse payments corresponding to that
mortgage).

3.6. Equilibrium definition

A recursive equilibrium is characterized by

1. a set of value functions N, R, B, H, P, F, D, S, SR, and SH,

2. rules for nonhomeowners' renting I^rent ; renters' savings a^ R ; buyers' savings a^ B , borrowing b^ B and housing h^ B ; home-
owners' choices of paying the mortgage I^ P , refinancing I^F , defaulting I^D , and selling the house I^S ; mortgage payers' savings
a^ P ; mortgage refinancers' savings a^ R and borrowing b^ R ; defaulters' savings a^ D ; sellers' renting I^rentS ; seller-renters' savings
a^ SR ; seller-buyers' savings a^ SB , borrowing b^ SB and housing h^ SB ,
3. and a price function q,

such that

(i) given a price function q; the policy rules I^rent , a^ R , a^ B , b^ B , h^ B , I^P , I^F , I^ D , I^S , a^ P , a^ R , b^ R , a^ D , I^ rentS , a^ SR , a^ SB , b^ SB , h^ SB , and the value
R H
functions N, R, B, H,
n P, F, D, S, S , andoS solve the Bellman equations (3)–(16).
(ii) given policy rules a^ P ; a^ D ; I^P ; I^F ; I^S ; I^D , the price function q satisfies Eq. (19).

4. Calibration

The model is calibrated using U.S. data. Whenever possible, the 2001 Survey of Consumer Finances (SCF) is used as a
reference.8 Table 1 presents parameter values obtained without using simulations of the model.
As in Kaplan and Violante (2010), a period in the model refers to a year. Households enter the model at age 25, retire at
age 60, and die at age 95. A household's initial asset position is 65 percent of its initial income, which allows us to match the
mean net asset position at age 25 in the SCF.
The variance of housing-price innovations (σ 2ν ) and the correlation of income and housing-price innovations (ρe;ν ) are
pinned down to match the standard deviation of house-price growth and the correlation between house-price growth and

8
For households between 25 and 60 years of age that are not in the top 5 percentile of the wealth distribution. The year 2001 is used because the
calibration does not feature changes in the aggregate price of housing (Section 6 studies such changes) and the U.S. boom in real house prices had just
begun in 2001.
 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190 181

Table 2
Targets and fit.

Parameter Value Definition

β 0.935 Discount factor

hR 1.49 Size rental house
p 4.48 Mean house price
θ 0.11 Weight of housing

Variables Data Model

Median (saving/income) 0.80 0.78

Homeownership rate 0.66 0.66
Median house value/median income 2.80 2.91
Median down payment 0.18 0.17

Source: 2001 SCF and Paniza Bontas (2010).

income growth estimated by Campbell and Cocco (2003), 0.115 and 0.027, respectively. The persistence of house prices (ρp)
is given by the estimate of Nagaraja et al. (2011).
The parameters σ e ; σ ε and the life cycle component of the income process are calibrated following Kaplan and Violante
(2010). As in Storesletten et al. (2004), the fixed effect takes two values,  0.459 and 0.459. Parameter values for the
retirement income are chosen to make the replacement ratio decline with income, from 69 percent to 14 percent, con-
sistently with the U.S. replacement ratios (Aon, 2008). This implies an average replacement ratio of 47 percent, which is
close to the replacement ratio in other quantitative studies (see, for example, Conesa and Krueger, 1999). Mean income
is 5.74.
The values of the risk-free interest rate (r ¼0.02), household's risk aversion parameter (γ ¼ 2) and elasticity of intra-
temporal substitution (α ¼ 0:5) are within the range of accepted values. Hanushek and Quigley (1980) estimate elasticities of
0.5 for Phoenix and 0.6 for Pittsburgh. Siegel (2008) finds an elasticity of 0.5, which is also the point estimate presented by Li
et al. (2015).9
The cost of buying and selling a house are the ones presented by Gruber and Martin (2003) and Pennington-Cross (2006).
The cost of signing a mortgage is the average cost reported by the Board of Governors of the Federal Reserve System. The
depreciation of mortgage installments is set considering an inflation rate of 2 percent.
There is no recourse (ϕ is sufficiently high and I rm ¼ 0). The household cannot borrow more than the value of the house it
buys (λ ¼ 1). There are seven house sizes the household can buy f2; 4; 6; 8; 10; 15; 20g. This is sufficient for accounting for the
life cycle profile of the average house value.
The remaining four parameter values (the size of the house available for rent, the mean price of houses, the discount
factor, and the weight of non-durable consumption in the utility function) are calibrated to make four statistics from the
model simulations approximate their data counterparts.10 The size of the house available for rent is the key parameter to
match homeownership (SCF). The discount factor is the key parameter to match the median (nonhousing) savings-to-
income ratio (SCF). The mean price of housing is the key parameter to match the median house value-to-median income
ratio (SCF). The nonhousing consumption weight in the utility function is the key parameter to match the median down
payment (Paniza Bontas, 2010). Table 2 presents the fit of the targets obtained with our benchmark calibration and the
implied parameter values. The model matches the targeted moments closely.

5. Fit of nontargeted moments

This section describes model predictions not targeted in the calibration regarding the demand for housing, the use of
mortgage loans, and mortgage defaults. In terms of the demand for housing, our calibration targets (and matches reasonably
well) the homeownership rate and the median house price. Fig. 2 shows that the model also captures changes in the
demand for housing over the life cycle (SCF). Homeownership increases over the life cycle, since older households tend to be
richer and thus are more likely to be able to afford ownership. Furthermore, while the housing price is exogenous and
independent of age, older households tend to buy more housing, making the mean house price increase over the life cycle.
Regarding the use of mortgage loans, Fig. 3 shows that the model produces plausible implications for the distribution of
mortgage down payments.11 Table 3 shows that mortgage payments in the data are higher than those in the model

9
Previous versions of this study assume a Cobb–Douglas utility function and find essentially the same results.
10
The model is solved using linear interpolation with evenly distributed grid points, and 10 grid points for b, 15 grid points for w, 20 grid points for a,
and 10 grid points for z. Expectations are computed using 20 Gauss–Legendre quadrature points over p0 , 10 points over z0 , and 8 points over ϵ0 . Simulation
results are for the behavior of 10,000 households (5000 of each type) during their lifetime. Statistics are computed using Census data to assign population
weights to each cohort.
11
Down payment data are not available in the SCF. The empirical distribution of down payments uses data on combined LTV ratios at origination for
the 2000–09 period presented by Paniza Bontas (2010).
182 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190

Ownership rate House Price for Home Owners (normalized, mean = 1)

100 1.8
Model Model
90 Data Data
1.6

80
1.4
70
1.2
60

50 1

40 0.8

30
0.6
20
0.4
10

0 0.2
25 30 35 40 45 50 55 60 25 30 35 40 45 50 55 60
Age Age

Fig. 2. Demand for housing over the life cycle (nontargeted).

5
Data
4.5 Model
4

3.5

3
Density

2.5

1.5

0.5

0
0 0.2 0.4 0.6 0.8 1
Down-payment

Fig. 3. Distribution of down payments. Source: The empirical distribution is constructed using data presented by Paniza Bontas (2010).

Table 3
Model's fit of nontargeted statistics.

Variables Data Model

Median payment/median income 0.15 0.11

Mean home equity/mean house price, mortgagees 0.24 0.28
Default rate (%) 0.50 0.47
Insurance coefficient, permanent shock 0.36 0.40
Insurance coefficient, transitory shock 0.95 0.88

Source: Payments data are from the 2001 SCF. The data on home equity are from CoreLogic. The default rate data is the calibration target presented by Jeske
et al. (2013) and Mitman (2012). Insurance coefficients for earning shocks are computed for the data by Blundell et al. (2008). The insurance coefficient for
it Þ;xit Þ
shock xit is given by μx ¼ 1  covðΔlogðc
varðxit Þ , where the variance and covariance are taken cross-sectionally over the entire population. To compute insurance
coefficients, log consumption and log earnings are defined as residuals from an age profile. The insurance coefficient is interpreted as the share of the
variance of shock x that does not translate into consumption growth.

simulations. Notice, however, that mortgage payments in the data overstate the financial cost of mortgages because of the
tax deductibility of interest payments (which is not a feature of our model). Table 3 also shows that our model slightly
overstates the mean home equity reported by CoreLogic, which is lower than the 42 percent in the 2001 SCF.12

12
The CoreLogic measure of housing equity may be more accurate because it uses transaction data to estimate house values. The SCF measure relies on
self-reported house price data.
 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190 183

0.87 70
Defaulter in period 0
60 Non−defaulter in period 0
0.86
Income, defaulters/non−defaulters

50
0.85
40

Equity / Price, %
0.84 30

0.83 20

10
0.82
0
0.81
−10

0.8 −20
−4 −3 −2 −1 0 −4 −3 −2 −1 0
Periods, normalized default period = 0 Periods, normalized default period = 0

Fig. 4. Model dynamics for income (Left) and home equity (Right) before default. Note: Age differences between defaulters and non-defaulters are con-
trolled for by computing for each group first the mean for each age and later the mean across ages.

Fig. 3 also shows that (as in the data) the majority of households in our simulations choose to pay significant down
payments, making their mortgages virtually default-free (this is also reflected in the very low default rate presented in
Table 3). Thus, most mortgage loans originate at a rate very close to the one that reflects the lenders' opportunity cost,
independent of the household's characteristics. Nevertheless, modeling the mortgage rate as a function of the loan and the
household's characteristics is essential for obtaining a plausible distribution of down payments and for measuring how the
household's borrowing opportunities would change with changes in policies.
The model also generates a plausible default rate. In particular, the default rate generated by the model is close to 0.5 percent,
which is the value targeted by Jeske et al. (2013) and Mitman (2012). They explain that the quarterly foreclosure rate was
0.4 percent between 2000 and 2006, and the ratio of mortgages in foreclosure eventually ending in liquidation was 25 percent in
2005. They argue that since a default in their model (as in ours) implies that the household relinquishes its house to the lender,
the default rate in the simulations should be compared with the liquidation rate in the data. They also argue that since the
default rate in the data is for a period of strong appreciation of house prices, they should target a higher default rate.13
In addition, Table 3 shows that the model generates reasonable responses of consumption to earnings shocks. This lends
confidence to the model predictions for consumption adjustments in response to changes in policies. Section 6 shows that the
ability of households to self-insure against both earnings and housing-price shocks is not significantly affected by these changes.
Using data from Massachusetts, Foote et al. (2008) show that negative equity is a necessary but not a sufficient condition
for default: less than 10 percent of the homeowners with negative equity default on their mortgages. They also argue that
income shocks play the role of trigger events for default. Fig. 4 shows our model is consistent with their findings: for
households that default, both income and home equity are lower and decline in the periods preceding defaults.
Overall, the results presented above indicate that our framework is a reasonable quantitative model of the demand for
housing and mortgages and mortgage defaults. Thus, our framework could be a useful laboratory for the study of policies
that could mitigate mortgage defaults. The next section studies the effects of such policies.

6. Prudential policies

This section evaluates two policies: recourse mortgages and LTV limits. The section first studies the effect of each of these
policies separately and later shows that there may be significant gains from the combined implementation of both policies.

6.1. Recourse mortgages

This subsection studies model economies with recourse mortgages.14 Table 4 presents model simulations for different
values of the three parameters used to characterize the recourse policy (all other parameter values are those used in the
benchmark calibration): the level of cash-in-hand wealth exempt from attachment as a percentage of the median income
(ϕ), the maximum percentage of cash-in-hand wealth that can be attached (κ), and the (expected) duration of the
attachment period (ψ).

13
Fig. 6 illustrates how the model generates a lower default rate during a period of strong appreciation of house prices.
14
Since the model does not feature a labor supply decision, this subsection cannot study the effect of recourse mortgages on this decision. Results in
previous studies indicate, however, that this effect is negligible (Chatterjee and Gordon, 2012; Li and Han, 2007). This occurs in part because people would
choose to default for asset and income levels lower than the ones at which recourse becomes operative.
184 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190

Table 4
Long-Run effects of recourse mortgages.

Benchmark One year of attachment, 100% attachment limit

Exemption % of median income (ϕ)

Moment 100 50 25 10 5

Homeownership rate 0.66 0.68 0.70 0.74 0.73 0.72

Mean house size (owners) 1.00 1.04 1.06 1.07 1.07 1.07
Median down payment 0.17 0.10 0.00 0.00 0.00 0.00
Default rate (%) 0.47 0.68 0.41 0.10 0.00 0.00
Median payment/median income 0.11 0.14 0.16 0.16 0.16 0.16
Median (equity/price), mortgagees 0.24 0.09 0.04 0.01 0.01 0.02
IC permanent shock 0.40 0.42 0.42 0.42 0.40 0.40
IC transitory shock 0.88 0.88 0.88 0.88 0.89 0.89
IC housing-price shock 0.87 0.88 0.88 0.87 0.85 0.85
Ex-ante welfare gains 0.00 1.50 2.36 2.78 2.53 2.32

Benchmark One year of attachment, 25% attachment limit

Exemption as % of median income (ϕ)

100 50 25 10 5

Homeownership rate 0.66 0.68 0.69 0.70 0.70 0.70

Mean house size (owners) 1.00 1.03 1.02 1.01 1.01 1.01
Median down payment 0.17 0.12 0.08 0.06 0.06 0.06
Default rate (%) 0.47 0.55 0.34 0.23 0.20 0.20
Median payment/median income 0.11 0.13 0.14 0.14 0.14 0.14
Median (equity/price), mortgagees 0.24 0.11 0.09 0.09 0.09 0.09
IC permanent shock 0.40 0.41 0.41 0.41 0.41 0.41
IC transitory shock 0.88 0.88 0.88 0.88 0.88 0.88
IC housing-price shock 0.87 0.88 0.88 0.88 0.88 0.88
Ex-ante welfare gains 0.00 1.36 1.65 1.69 1.65 1.65

Benchmark Six years of attachment, 50% exemption

Attachment limit as % of wealth (κ)

10 25 40 50 65

Homeownership rate 0.66 0.75 0.76 0.76 0.76 0.76

Mean house size (owners) 1.00 1.05 1.05 1.05 1.05 1.05
Median down payment 0.17 0.00 0.00 0.00 0.00 0.00
Default rate (%) 0.47 0.43 0.42 0.40 0.38 0.37
Median payment/median income 0.11 0.16 0.17 0.17 0.17 0.17
Median (equity/price), mortgagees 0.24 0.01 0.00 0.00 0.00 0.00
IC permanent shock 0.40 0.42 0.42 0.42 0.42 0.42
IC transitory shock 0.88 0.88 0.88 0.88 0.88 0.88
IC housing-price shock 0.87 0.87 0.87 0.87 0.87 0.86
Ex-ante welfare gains 0.00 2.81 2.97 3.00 2.99 2.99

Note: The first (second) panel assumes ψ ¼ 1 (one year of attachment of defaulter's wealth), κ ¼ 1 (κ ¼ 0:25) implying that lenders can attach up to 100%
(25%) of the defaulter's wealth. The third panel assumes ψ ¼ 0:167 (six years of attachment in expectation) and an exemption of 50% of the median income.
House sizes are normalized to 1 in the benchmark. Ex-ante welfare gains are measured as the proportional increase in (housing and non-durable) con-
sumption that would compensate a new-born household for living in the benchmark instead of in the economy with a prudential policy.

The first panel in Table 4 presents different limits for the cash-in-hand wealth exempt from attachment, with an
attachment period of one year and no limit for the percentage of cash-in-hand wealth that can be attached. The second
panel presents the same exercises but with the assumption that only 25 percent of cash-in-hand wealth can be attached
(this is the attachment limit for wages in the U.S.). The third panel assumes that 50 percent of cash-in-hand wealth is
exempt from attachment (an attachment exemption for wages in the U.S.), an expected attachment period of six years (the
usual duration of attachment periods in the U.K.; Lea, 2010), and different limits to the percentage of wealth that can be
attached. Overall, Table 4 presents results for 15 different recourse policies.
Table 4 shows that the relationship between the degree of recourse and the default rate is nonmonotonic. Somewhat
surprisingly, recourse policies that increase the cost of defaulting may increase the default rate. For instance, the economies
with recourse mortgages with an exemption from attachment equal to the median income in the first two panels of Table 4
(third column) have a default rate higher than the one in the benchmark. For an exemption equal to 50 percent of the
median income and an attachment limit of 25 percent of cash-in-hand wealth, the default rate is higher with a harsher
recourse rule of six years of attachment than with one year of attachment (second and third panels of Table 4).
 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190 185

2
Benchmark

Mortgage rate − risk−free rate (in %)

Rec, exemption = median(y)
Rec, exemption = 0.25*med(y)
1.5

0.5

0
0 10 20 30 40
Down payment, %
0
Fig. 5. Mortgage spread with and without recourse. Note: The figure is for a nonhomeowner buying a house with h ¼ 2 (smallest house to buy), a0 ¼ 0,
f ¼  0:459 (low), w¼ 0.40, z ¼  0:14, p¼ 0.46, and n¼ 65. Recourse mortgages do not have an attachment limit, and have an attachment duration of one year.

Furthermore, it may be difficult to lower the default rate significantly with recourse mortgages. Table 4 shows that
recourse mortgages only imply a default rate that is half the one in the benchmark for very low levels of exempted wealth
(below 25 percent of the median income). The default rate is zero only with a very harsh recourse rule with an exemption
level of 10 percent of the median income and no limit on the percentage of wealth that can be attached.
Why can the default rate be higher in economies with a recourse policy that implies a higher cost of defaulting? Table 4
shows that a harsher recourse policy may increase the LTV chosen by households, increasing the level of mortgage pay-
ments and lowering housing equity. Therefore, since affordability and equity are key determinants of the default rate (Fig. 4),
a harsher recourse policy may increase the default frequency.
But why would a household choose a higher LTV and assume more default risk when the punishment for defaulting is
harsher? A household can use a higher-LTV mortgages to consume more housing sooner at the expense of exposing itself to
costly defaults. The household dislikes defaults because of the associated costs (including, for instance, the cost of moving to
a different house) and because future default decisions need not be optimal from an ex-ante perspective. Two forces explain
why a harsher recourse policy may increase a household's benefit from assuming default risk.
First, with a harsher recourse policy, an increase in the LTV results in a smaller increase in default risk. This occurs
because the LTV is less important for a household's default decision, as illustrated by the flattening of the mortgage spread
curve implied by harsher recourse policies in Fig. 5. Fig. 5 also illustrates how a household that is eager to borrow chooses a
higher LTV when the spread curve is flatter.
Second, households dislike default risk less when default is more likely to be triggered by income shocks. Households
would like to insure against negative income shocks, and mortgage defaults provide this insurance. Recall that Fig. 4
indicates that negative income shocks trigger defaults. Thus, mortgage defaults provide debt relief to households that suffer
these shocks. In contrast, declines in the price of housing that could also trigger a mortgage default may have small negative
welfare effects for households that do not plan to adjust their consumption of housing and may even increase welfare for
homeowners who expect to buy larger houses in the future (see Fig. 7). Thus, households are less eager to acquire contracts
that transfer resources to states with negative shocks to the price of housing (as defaultable mortgages do increasingly with
a softer recourse rule).
Table 4 shows that the effect of the degree of recourse in the demand for housing (as represented by ownership and
house sizes) and welfare is also nonmonotonic. The demand for housing may increase with the degree of recourse because a
harsher recourse rule lowers the mortgage interest rate households pay for each LTV (see Fig. 5), making households more
willing to ask for high-LTV mortgages (as reflected in the lower median down payment reported in Table 4). This allows
households to buy larger houses sooner in their life cycle. However, with the harsher recourse rules in Table 4, households
become more reluctant to ask for high-LTV mortgages. With these recourse rules, the cost of defaulting is so high that
households do not default after very adverse income shocks, and thus are obliged to make large adjustments in non-housing
consumption. Therefore, households choose lower LTV mortgages that imply lower payments, which they are more likely to
be able to afford without large adjustments in non-housing consumption. To be able to afford lower-LTV mortgages,
households delay home purchases and choose smaller houses (since changes in the recourse rule affect only high-risk
borrowers, there is no change in the median down payment reported in Table 4).
Table 4 also shows that the effect of recourse on ex-ante welfare closely follows the effect of recourse on housing
consumption. Table 5 presents the distribution of welfare gains from introducing recourse mortgages in the benchmark
economy computed considering the transition paths for each household, and shows that welfare gains from recourse
186 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190

Table 5
Welfare gains from prudential policies.

Welfare gain Recourse mortgages LTV limit Recourse mortgages with LTV limit

Ex-ante new-born gain 1.65  0.47 0.55

25th percentile gain 0.82  0.37  0.12
Median gain 3.44  0.11 3.17
75th percentile gain 8.99  0.05 9.02

Note: Recourse mortgages have an exemption equal to 50 percent of median income, an attachment limit of 25 percent of cash-in-hand wealth, and an
attachment duration of one year. Mortgages that existed before the introduction of recourse mortgages continue to be nonrecourse. The LTV limit is 80 percent.

mortgages may be significantly higher than the ex-ante new-born gains presented in Table 4. The insurance coefficients are
not significantly affected by the recourse policy.15
Fig. 6 shows that the increase in mortgage defaults triggered by a large decline in the aggregate price of housing may be
significant in economies with recourse mortgages, and it may even be larger than the one observed in an economy without
recourse mortgages.16 Because of the computational burden of exercises with large decline in the aggregate price of housing,
Fig. 6 only presents one recourse policy—with an exemption equal to 50 percent of median income, an attachment limit of
25 percent of cash-in-hand wealth, and an expected attachment duration of one year.17
Fig. 7 shows that many households benefit from a large decline in the price of housing.18 These are households that
expect to consume more housing in the future (renters and young owners who expect to buy larger houses).
Fig. 7 also shows that welfare gains from a large decline in the price of housing are smaller in the economy with recourse
mortgages. The average welfare gain is  0.33 percent in the economy with non-recourse mortgages and  0.72 percent in
the economy with recourse mortgages. This is not surprising and should not be interpreted as an argument against recourse
mortgages. Recourse mortgages deprive debtors from insurance against declines in the price of housing, and Fig. 7 presents
a very extreme case in which such insurance would be useful (a large and expected-to-be-permanent decline in the price of
housing). Furthermore, the economy with recourse mortgages features a higher ownership rate and thus, since declines in
the price of housing are good for renters and tend to be bad for homeowners, gains from price declines are smaller in this
economy. In addition, a lower housing price favors households because it improves housing affordability, which is less of a
problem with recourse mortgages (that allow for lower down payments).
In sum, our results indicate that while recourse policies have potential for mitigating mortgage defaults, the imple-
mentation of these policies presents difficulties. It may be difficult to significantly reduce the default rate with recourse
mortgages, and a recourse policy that is too mild may even increase default risk. Furthermore, a recourse policy that is too
harsh may reduce the boost to housing consumption and welfare implied by recourse mortgages. Since the increase in
default risk implied by mild recourse policies is the result of high origination LTVs, this problem could be mitigated by
imposing LTV limits for new mortgages. Section 6.2 presents the effects of introducing LTV limits with nonrecourse
mortgages and Section 6.3 presents the effects of combining LTV limits with recourse mortgages.

6.2. LTV limits

This subsection solves the benchmark model but changes the LTV limit λ in constraints (6), (10) and (16) of the
household's problem, allowing the household to borrow a smaller fraction of the value of the house it buys, instead of 100
percent as in the benchmark. All other parameter values are the ones in the benchmark calibration.
Table 6 shows that economies with a stricter LTV limit feature a significantly lower mortgage default rate. This occurs
because with a stricter LTV limit, households are forced to have more home equity when they buy their house and, thus, are

15
Insurance coefficients in Table 4 show that the average adjustment is nonhousing consumption after a housing-price shock is small and comparable
to the average adjustment in nonhousing consumption after a transitory shock to earnings. Households that do not expect to adjust their housing con-
sumption will not significantly adjust their nonhousing consumption after a housing-price shock, because they expect the housing price to revert to its
mean. This finding is consistent with the evidence presented by Sinai and Souleles (2005), who show that the risk of owning a house declines with the time
the household expects to stay in its house. Households that expect to buy (sell) housing in the future typically benefit from (are hurt by) a negative
housing-price shock and choose higher (lower) nonhousing consumption.
16
It is difficult to compare model predictions from these experiments with the recent behavior of mortgage defaults in the U.S. (the foreclosure rate
peaked slightly above 3 percent while the seriously delinquent rate peaked at 5.1 percent; Noeth and Sengupta, 2011). This is because (i) there were
massive government interventions to help homeowners with negative equity and (ii) banks may have delayed foreclosures to avoid recognizing losses.
Some important government interventions include the National Foreclosure Mitigation Counseling program, the Making Home Affordable programs, and
the Neighborhood Stabilization Program.
17
This is a relatively mild recourse rule that could be easier to implement. For instance, the U.S. Bankruptcy Abuse Prevention and Consumer Pro-
tection Act of 2005 establishes that if a debtor's income is above the median income of his state, the debtor is subject to a means test that could force him to
file under Chapter 13 (under which a percentage of debts must be paid over a period of three to five years) as opposed to Chapter 7 (under which debts are
paid only from existing assets).
18
Because declines in the aggregate price of housing are unanticipated, Fig. 7 presents welfare gains for one-time declines instead of considering
consecutive declines as in Fig. 6.
 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190 187

1.18 1

1.16
0.95
1.14
0.9
1.12
0.85
House price

House price
1.1

1.08 0.8

1.06
0.75
1.04
0.7
1.02
0.65
1

0.98
1 2 3 4 1 2 3 4
Time Time

0.5 7
benchmark benchmark
0.45
LTV limit
6 LTV limit
recourse recourse
0.4
both both
0.35 5
Default rate

Default rate

0.3
4
0.25
3
0.2

0.15 2

0.1
1
0.05

0 0
1 2 3 4 1 2 3 4
Time Time

Fig. 6. Large changes in the aggregate price of housing. Note: The starting point is the benchmark economy with a constant aggregate house price p, which
is normalized to 1. Changes in p are unexpected and, once they occur, are believed to be permanent. Recourse mortgages have an exemption equal to 50
percent of median income, an attachment limit of 25 percent of cash-in-hand wealth, and an attachment duration of one year. The LTV limit is 80 percent.

Table 6
Long run effects of LTV limits.

Benchmark LTV limit at

Moment 90% 85% 80%

Homeownership rate 0.66 0.66 0.66 0.66

Mean house size (owners) 1.00 1.01 1.00 1.00
Median down payment 0.17 0.17 0.17 0.20
Default rate (%) 0.47 0.40 0.25 0.14
Median payment/median income 0.11 0.11 0.11 0.10
Median (equity/price), mortgagees 0.24 0.24 0.26 0.28
IC permanent shock 0.40 0.40 0.40 0.40
IC transitory shock 0.88 0.88 0.88 0.89
IC housing-price shock 0.87 0.87 0.86 0.86
Ex-ante welfare gains 0.00  0.12  0.24  0.47

Note: House sizes are normalized to 1 in the benchmark. Ex-ante welfare gains are measured as the proportional increase in (housing and non-durable)
consumption that would compensate a new-born household for living in the benchmark economy instead of in the economy with a LTV limit.

less likely to have sufficient negative equity to trigger a default in the future. However, the economy with an 80 percent LTV
limit cannot avoid a very high default rate after the larger shock to the aggregate price of housing presented in Fig. 6.
Table 6 also presents a negligible impact of LTV limits on the demand for housing (both ownership and the average size
of houses). There are two reasons for this. First, in economies with LTV limits, young households save more to afford higher
188 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190

Welfare gains, % CE
0

−2

−4
Benchmark
Recourse
−6 LTV limit
Both

−8
0 20 40 60 80 100
Percentiles of welfare gains

Fig. 7. Welfare gains after a large aggregate decline in the price of housing. Note: The figure assumes a 28 percent unanticipated decline in p and the price
of housing for all households. Recourse mortgages have an exemption equal to 50 percent of median income, an attachment limit of 25 percent of cash-in-
hand wealth, and an expected attachment duration of one year. The LTV limit is 80 percent.

Table 7
Long run effects of recourse mortgages with LTV limits.

Benchmark One year of attachment Six years of attachment

LTV limit at LTV limit at

Moment 100% 90% 80% 100% 90% 80%

Homeownership rate 0.66 0.69 0.69 0.68 0.76 0.71 0.68

Mean house size (owners) 1.00 1.02 1.04 1.04 1.05 1.07 1.07
Median down payment 0.17 0.08 0.10 0.20 0.00 0.10 0.20
Default rate (%) 0.47 0.34 0.15 0.02 0.42 0.04 0.00
Median payment/median income 0.11 0.14 0.13 0.12 0.17 0.15 0.12
Median (equity/price), mortgagees 0.24 0.09 0.14 0.22 0.00 0.10 0.21
IC permanent shock 0.40 0.41 0.41 0.42 0.42 0.41 0.41
IC transitory shock 0.88 0.88 0.88 0.89 0.88 0.88 0.89
IC housing-price shock 0.87 0.88 0.87 0.85 0.87 0.85 0.85
Ex-ante welfare gains 0.00 1.65 1.23 0.55 2.97 1.57 0.60

Note: The assumed exemption is 50% of the median income and lenders can attach up to 25% of the defaulter's wealth. House sizes are normalized to 1 in
the benchmark. Ex-ante welfare gains are measured as the proportional increase in (housing and non-durable) consumption that would compensate a
new-born household for living in the benchmark instead of in the economy with a default-prevention policy.

down payments. Thus, in general, LTV limits do not prevent households from buying the house they want. Second, LTV
limits lower the interest rate households pay on their mortgage, making housing consumption more attractive. Mortgage
interest rates are lower when the default probability is lower. LTV limits make it harder for a household that defaulted to
buy a new house and, therefore, lower the default probability and the mortgage interest rate. However, the second reason is
not quantitatively important in the simulations: even with the 100 percent LTV limit in the benchmark, most households
choose to pay significant down payments and thus do not pay a significant default premium in their mortgages.
Table 6 also shows that LTV limits generate welfare losses. Recall that in our model endogenous borrowing constraints
prevent households from consuming more housing. A stricter LTV limit tightens these constraints. In addition, Table 6 shows
that LTV limits do not significantly change the households' ability to self-insure.

6.3. Combining recourse mortgages and LTV limits

Could the combination of recourse mortgages and LTV limits mitigate mortgage defaults and at the same time boost
housing consumption? Could it prevent large increases in the mortgage default rate after a collapse in the aggregate price of
housing? Previous subsections show that recourse mortgages could relax households' borrowing constraints and thus
increase housing consumption but could be ineffective in significantly reducing the rate of mortgage defaults. In contrast,
LTV limits would lower the default rate, but at the expense of worsening households' borrowing constraints. Furthermore,
neither of these two prudential policies could prevent large increases in the mortgage default rate after a collapse in the
aggregate price of housing. This subsection studies the effects of combining these two policies.
 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190 189

Table 7 shows there may be important complementarities between recourse mortgages and LTV limits. The table con-
siders relatively mild recourse rules that respect attachment limits observed in the U.S. and thus could be easier to
implement. Economies with both recourse mortgages and LTV limits feature a higher ownership rate with a lower default
rate, indicating that the combination of these two tools could succeed in the two most often cited goals of mortgage policies:
promoting homeownership and containing default. Tables 5 and 7 show that welfare gains from introducing recourse
mortgages together with LTV limits could be substantial.
Fig. 6 shows that complementarities between recourse mortgages and LTV limits could also be key for preventing large
increases in the mortgage default rate after a collapse in the aggregate price of housing. For the case with three consecutive
declines in the aggregate price of housing, the default rate in the economy with both LTV limits and recourse mortgages is
less than 25 percent of the default rate in economies with only one of these policies. With both policies, while the aggregate
shock produces a large increase in the share of owners with negative equity, the cost of defaulting implied by recourse
mortgages is large enough to prevent a large increase in the default rate. Without LTV limits, equity would be too low.
Without recourse mortgages, the cost of defaulting would be too low.

7. Conclusions

A life-cycle model is developed in which households face income and housing-price risk and buy houses with mortgages.
This model, which accounts for key features in U.S. data, is used as a laboratory for studying two prudential policies:
recourse and LTV limits.
Recourse policies show potential for mitigating mortgage defaults, but the implementation of these policies could
present difficulties. It may not be possible to significantly reduce the default rate with plausible recourse policies, and a
recourse policy that is too mild may even increase default risk. Furthermore, a recourse policy that is too harsh may reduce
the boost to housing consumption and welfare implied by recourse mortgages.
LTV limits reduce the default rate with negligible negative effects on housing consumption. Furthermore, there are
important complementarities between recourse mortgages and LTV limits. Concerns about the undesirable effects of
recourse mortgages could be mitigated by combining a relatively mild recourse rule with LTV limits. This produces a lower
default rate and a stronger demand for housing. In addition, combining recourse mortgages and LTV limits may be necessary
to prevent high default rates after sharp declines in the price of housing.

Acknowledgments

For comments and suggestions, we thank the editor, Urban Jermann, an anonymous referee, and seminar participants at
ASU, Georgetown U., Indiana U., McMaster U., SUNY at Stony Brook, U. of Toronto, York U., the FRB of Richmond and St. Louis,
the IMF, the 2008 and 2009 Wegmans conference, the 2010 and 2014 SED conference, the 2010 and 2013 HULM Conference,
the 2011 North America Summer Meeting of the Econometric Society, the 2011 SAET conference, the 2011 Philadelphia Fed
Conference on Consumer Credit, the 2013 NBER Summer Institute, and the 2013 Macro-Finance Workshop at NYU. We thank
S. Henly, C. Liborio, J. Tompkins, T. Hursey, and E. Yurdagul for excellent research assistance. We thank Jennifer Paniza Bontas
for sharing her data with us. We thank M. Michaux, M. Nakajima, and D. Schlagenhauf for useful discussions. Remaining
mistakes are our own. This research was supported in part by Lilly Endowment, Inc., through its support for the Indiana
University Pervasive Technology Institute, and in part by the Indiana METACyt Initiative. The Indiana METACyt Initiative at
IU is also supported in part by Lilly Endowment, Inc. The views expressed herein are those of the authors and should not be
attributed to the IMF, its Executive Board, or its management, the FRB of St. Louis, or the Federal Reserve System.

References

Aon, 2008. Replacement ratio study: a measurement tool for retirement planning. Global Corporate Marketing and Communications.
Athreya, K., 2005. Welfare implications of the bankruptcy reform act of 1999. Rev. Financ. Stud. 18 (1), 1567–1595.
Athreya, K., Tam, X.S., Young, E.R., 2011. Are Harsh Penalties for Default Really Better? Federal Reserve Bank of Richmond Working Paper 09–11.
Athreya, K.B., 2008. Default, insurance and debt over the life-cycle. J. Monet. Econ. 55 (4), 752–774.
Blundell, R., Pistaferri, L., Preston, I., 2008. Consumption inequality and partial insurance. Am. Econ. Rev. 98 (5), 1887–1921.
Campbell, J.Y., 2012. Mortgage market design. Rev. Financ. 17, 1–33.
Campbell, J.Y., Cocco, J.F., 2003. Household risk management and optimal mortgage choice. Q. J. Econ. 118, 1449–1494.
Campbell, J.Y., Cocco, J.F., 2015. A model of mortgage default. J. Financ. 70 (4), 1495–1554.
Chatterjee, S., Corbae, D., Nakajima, M., Ríos-Rull, J.-V., 2007. A quantitative theory of unsecured consumer credit with risk of default. Econometrica 75,
1525–1589.
Chatterjee, S., Eyigungor, B., 2015. A quantitative analysis of the US housing and mortgage markets and the foreclosure crisis. Rev. Econ. Dyn. 18 (2),
165–184.
Chatterjee, S., Gordon, G., 2012. Dealing with consumer default: bankruptcy vs garnishment. J. Monet. Econ. 59, S1–S16.
Clauretie, T., 1987. The impact of interstate foreclosure cost differences and the value of mortgages on default rates. J. Am. Real Estate Urban Econ. Assoc. 15,
152–167.
Conesa, J.C., Krueger, D., 1999. Social security reform with heterogeneous agents. Rev. Econ. Dyn. 2 (2), 757–795.
Corbae, D., Quintin, E., 2015. Leverage and the foreclosure crisis. J. Polit. Econ. 123 (1), 1–65.
190 J. Carlos Hatchondo et al. / Journal of Monetary Economics 76 (2015) 173–190

Corradin, S., Fillat, J., Vergara-Alert, C., 2014. Optimal portfolio choice with predictability in house prices and transaction costs. Rev. Financ. Stud. 27 (3),
823–880.
FED, 2012. The U.S. Housing Market: Current Conditions and Policy Considerations. White Paper, Board of Governors of the Federal Reserve System.
Feldstein, M., November 2008. How to help people whose home values are underwater. Wall Street Journal.
Foote, C., Gerardi, K., Willen, P., 2008. Negative equity and foreclosure: theory and evidence. J. Urban Econ. 6 (2), 234–245.
Ghent, A., Kudlyak, M., 2011. Recourse and residential mortgage default: evidence from U.S. states. Rev. Financ. Stud. 24 (9), 3139–3186.
Gruber, J., Martin, R., 2003. Precautionary savings and the wealth distribution with illiquid durables. International Finance Discussion Papers No. 773, Board
of Governors of the Federal Reserve System.
Guler, B., 2015. Innovations in information technology and the mortgage market. Rev. Econ. Dyn. 18 (July (3)), 456–483.
Hanushek, E.A., Quigley, J.M., 1980. What is the price elasticity of housing demand? Rev. Econ. Stat. 62 (3), 449–454.
Hatchondo, J.C., Martinez, L., Sánchez, J.M., July 2013. Europe may provide lessons on preventing mortgage defaults. The Regional Economist, 12–13.
IMF, April 2011. Global financial stability report.
Jeske, K., Krueger, D., Mitman, K., 2013. Housing, mortgage bailout guarantees and the macro economy. J. Monet. Econ. 60 (8), 917–935.
Kaplan, G., Violante, G.L., 2010. How much consumption insurance beyond self-insurance? Am. Econ. J.: Macroecon. 2, 53–87.
Lea, M., 2010. International comparison of mortgage product offerings. Research Institute for Housing America.
Li, W., Han, S., 2007. Fresh start or head start? the effects of filing for personal bankruptcy on work effort. J. Financ. Serv. Res. 31 (2), 123–152.
Li, W., Liu, H., Yao, R., 2015. Housing over time and over the life cycle: a structural estimation. Federal Reserve Bank of Philadelphia Working Paper
No. 15-04.
Li, W., Sarte, P.-D., 2006. U.s. consumer bankruptcy choice: the importance of general equilibrium effects. J. Monet. Econ. 53, 613–631.
Livshits, I., MacGee, J., Tertilt, M., 2007. Consumer bankruptcy: a fresh start. Am. Econ. Rev. 97, 402–418.
Mayer, C., Pence, K., Sherlund, S., 2009. The rise in mortgage defaults. J. Econ. Perspect. 23 (1), 27–50.
MBA, 2011. April 13, 2011 proposed qrm harms creditworthy borrowers and housing recovery. Mortgage Bankers Association. White Paper Prepared in
Advance of April 14, 2011 House Subcommittee on Capital Markets and Government Sponsored Enterprise Hearing.
Mitman, K., 2012. Macroeconomic effects of bankruptcy and foreclosure policies. University of Pennsylvania, Unpublished manuscript. Available at https://
sites.google.com/site/kurtmitman/.
Nagaraja, C.H., Browny, L.D., Zhao, L.H., 2011. An autoregressive approach to house price modeling. Ann. Appl. Stat. 5, 124–149.
Noeth, B., Sengupta, R., January 2011. Have the trends in housing bottomed out? Regional Economist, Federal Reserve Bank of St. Louis, 22–23.
Paniza Bontas, J., 2010. Combined LTV at origination and household default probability. Unpublished manuscript, University of Chicago.
Pennington-Cross, A., 2006. The value of foreclosed property. J. Real Estate Res. 28, 193–214.
Quintin, E., 2012. More punishment, less default? Ann. Financ. 8 (4), 427–454.
Schwartz, E.S., Torous, W.N., 2003. Mortgage prepayment and default decisions: a poisson regression approach. Real Estate Econ. 21 (4), 431–449.
Siegel, S., 2008. Consumption-Based Asset Pricing: Durable Goods, Adjustment Costs, and Aggregation. Working Paper, University of Washington.
Sinai, T., Souleles, N.S., 2005. Owner-occupied housing as a hedge against rent risk. Q. J. Econ. 120 (2), 763–789.
Storesletten, K., Telmer, C.I., Yaron, A., 2004. Consumption and risk sharing over the life cycle. J. Monet. Econ. 51, 609–633.
Treasury, 2009. Homeowner affordability and stability plan fact sheet. Available at http://www.treasury.gov/press-center/press-releases/Pages/
20092181117388144.aspx.
Wong, E., Fong, T., Li, K.-F., Choi, H., 2011. Loan-to-value ratio as a macroprudential tools–hong-kong's experience and cross-country evidence. Hong Kong
Monetary Authority Working Paper 01/2011.