The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Working Paper Series
Fresh Start or Head Start? Uniform
Bankruptcy Exemptions and Welfare
WP 03-03
This paper can be downloaded without charge from:
http://www.richmondfed.org/publications/
Kartik Athreya
Federal Reserve Bank of Richmond
Fresh Start or Head Start? Uniform Bankruptcy
Exemptions and Welfarey¤
Kartik Athreya
Research Department
Federal Reserve Bank of Richmondxz
Working Paper No. 03-03R
July 26, 2004
Abstract
The 1990’s witnessed a historically unprecedented number of personal bankruptcy
…lings. In response, Congressional debate over bankruptcy law has recently led to sev
eral proposals aimed at making it more di¢cult to exempt wealth in a bankruptcy. In
this paper, I evaluate uniform exemption policy primarily within the context of the
recent congressional proposal H.R. 975. I develop an incomplete markets model where
secured and unsecured assets coexist and are treated di¤erentially in a bankruptcy
proceeding. I …nd that exemptions are associated positively with …ling rates and the
amount of equity held at the time of …ling. Conversely, exemptions are strongly neg
atively associated with the availability of unsecured credit. The welfare consequences
of exemptions, while small, are positive for high exemptions and negative for low ones.
Steady state welfare is maximized under a full exemption, and is worth $28.24 annually
to the average household. The results are robust, and show that increases in bank
ruptcy exemptions beyond current state averages are largely a matter of indi¤erence,
and do not merit the heated debate they have generated.
¤
JEL Classi…cation: D52; D58; G33
Keywords: Bankruptcy; Exemptions, Risk-Sharing; Incomplete Markets
z
E-mail: kartik.athreya@rich.frb.org, Ph: (804) 697-8225.
x
I thank the Editor, Wouter den Haan, and two anonymous referees for very helpful suggestions. I also
thank John Weinberg, Richard Hynes, Steve Williamson, William Zame, Dirk Krueger, Jesus Villaverde, and
y
colleagues at the Richmond Fed. For comments on earlier versions of this paper, I am grateful to seminar
participants at CFS-Frankfurt, Illinois, Iowa, Colorado, CSUF, the Federal Reserve Board of Governors, and
the Federal Reserve Banks of Atlanta, Cleveland, and Richmond. John Hejkal and Jon Peterson provided
able research assistance. The views expressed here are those of the author.
1. Introduction
Personal bankruptcy …lings have grown rapidly over the past decade. In 1990, there were
approximately 700,000 …lings. By 2003, …lings had more than doubled to 1.6 million. The
resulting losses to creditors have been estimated at over forty billion dollars annually (WEFA
(1998)). Consequently, there is now an intense public debate on the desirability of compre
hensive bankruptcy reform. In this debate, special emphasis has been placed on exemptions,
which are the rules governing the amount of wealth that may be retained by a debtor in a
bankruptcy …ling. In particular, several recent recommendations have advocated a uniform,
limited, and federally mandated exemption level, the most recent of which is the Bill in
Congress entitled “Bankruptcy Abuse Prevention and Consumer Protection Act of 2003”
(H.R. 975). The provisions of the bill with respect to exemptions concern the “homestead”
exemption, which applies to home equity, and is for most households by far the largest ex
emption. Exemption “reform” has been a divisive legislative issue for several years. In 1997,
a nine-member panel known as the National Bankruptcy Review Commission (NBRC) made
recommendations to increase exemptions and eliminates states’ rights to force households
to use more restrictive state exemptions. These recommendations proved contentious, and
survived only by a 5-4 margin. In a forceful reply to the majority opinion, dissenting Com
mission members (Jones and Shepard [1997]) argued that it was “...highly likely that these
liberal exemptions [would] translate into the …ling of more Chapter 7 liquidation cases”.
Secondly, the dissenters argued that the NBRC proposal gave “...debtors a head start, not
a fresh start” by enabling “...many Americans to escape their contractual obligations while
maintaining levels of wealth that the vast majority of Americans do not enjoy.”1
In this paper I ask three questions. First, as some have argued, will dramatic changes in
exemptions cause an equally dramatic change in bankruptcy rates, interest rates and con
sumer debt? Second, with respect to distributive e¤ects, will high uniform exemptions give
debtors a “head start” or just a “fresh start”, and what do we mean by these terms? Third,
what are the welfare consequences of exemptions, especially very low or very high ones? To
address these questions, I develop a dynamic equilibrium model with incomplete insurance
markets, secured and unsecured credit, and endogenous limits on unsecured borrowing. Most
importantly, I incorporate a well de…ned bankruptcy law that distinguishes between secured
and unsecured debt. I calibrate the model to the …ling behavior of U.S. homeowners, and
then study the e¤ects of uniform exemptions. The model gives quantitative predictions for
1
The NBRC also cites a Justice Department memorandum to it stating that it would favor “...more
modest exemption levels” in a June 18, 1997 letter to NBRC Commission Chairman Brady Williamson from
Francis M. Allegra, Deputy Associate Attorney General. The Justice Department made it known that it was
“concerned that the asset levels tentatively adopted by the Commission are too high in light of the historical
purposes of allowing property to be claimed as exempt”.
2
the extent to which exemptions limit unsecured borrowing and a¤ect welfare. Calculations
of the welfare implications of exemptions are notably absent in the ongoing policy discussion.
The central tension in the model is between the insurance that exemptions can provide,
and the increased costs of borrowing and bankruptcy they may generate. The model gener
ates …ve main results. First, large increases in uniform exemptions increase bankruptcy rates
and the equity held in bankruptcy non-trivially. Second, increases in uniform exemptions
increase the cost of obtaining unsecured debt, especially for households with low wealth.
Third, consumption smoothing is nearly invariant to exemptions. In other words, the increased use of bankruptcy under high exemptions o¤sets (and justi…es) the higher costs of
unsecured borrowing, leaving the equilibrium consumption process nearly una¤ected. This
leads to the fourth result, which is that welfare changes only minimally with exemptions.
The …fth result is that though the welfare gains are small, the optimal exemption level turns
out to be quite high.
Existing dynamic equilibrium analysis of exemptions is limited, but is growing quickly.
This paper is related to Li and Sarte (2002), who study bankruptcy exemptions in a model
with capital and exogenously limited unsecured debt. This paper also follows and extends
work originating in the general equilibrium models of bankruptcy of Dubey, Geanakopolos,
and Shubik (2001), (henceforth Dubey, et al. [2001]), Zame (1993), Zha (2001), and is related
to ongoing work of Chatterjee, Corbae, Nakajima, Rios-Rull (2002) (henceforth Chatterjee
et al. [2002]), Lehnert and Maki (2001), Livshits, MacGee and Tertilt (2003) [henceforth
Livshits et al. (2003)], and Pavan (2003). Chatterjee et al. (2002) evaluate bankruptcy
law in a small open-economy model of unsecured debt, and …nd a welfare improving role for
means-testing of bankruptcy …lers. Livshits et al. (2003) contrast bankruptcy codes in the
U.S. and Europe in a small open-economy, life-cycle setting, and …nd a welfare improving
role for bankruptcy in the presence of large “expense shocks”, such as health crises. A key
distinction between the work of Athreya (2002), Chatterjee et al. (2002) and Livshits et al.
(2003), and the present work is that I provide, to my knowledge, the …rst study of exemptions
in a quantitative incomplete-markets model where secured and unsecured debt coexist but
are treated di¤erently in bankruptcy.2
2
Note that the elimination of exemptions altogether does not in any way prevent bankruptcy, as it only
prevents the retaining of wealth by …lers.
3
2. The Model
2.1. Preferences
The environment consists of a continuum of households with CRRA preferences:3
E0
1
X
t=0
¯t
c1t ¡® ¡ 1
1¡®
(2.1)
2.2. Equity
In this paper, I will focus primarily on the behavior of homeowners. Homes are typically
the largest asset held by households, and “Homestead” equity exemptions are almost always
the largest exemptions o¤ered to households.4 Bankruptcy exemptions may therefore a¤ect
the decisions of homeowners in important ways.5 The value of a house is normalized to
ja s j units of the single perishable consumption good, and will provide the collateral for the
secured debt that households may take on. The labeling of this asset as a “house” is mainly
heuristic, and jas j should therefore be thought of as an aggregate measure of collateralizable
wealth. Total household equity, which I denote by “e", is then de…ned in the standard way,
as the di¤erence between the value of all collateralizable wealth and the value of debt held
against it. That is:
e ´ jas j + as
(2.2)
2.3. Endowments, Financial Intermediaries and Assets
Agents receive random endowments of a single perishable good each period. These endow
ments are assumed to take on discrete values, fy1 ; y2 ; :::; yN g, where yi > yj whenever i > j .
Household endowments follow a Markov process that is serially dependent over time but
independent across agents. There is therefore no aggregate risk. The transition function on
endowment realizations is given by µ ij ´ P (yt +1 = yj jyt = yi ), for i,j =1,..., N.
To smooth their stochastic incomes, agents are allowed to borrow as well as save. They
may borrow either on the collateralized/secured credit market or the uncollateralized/unsecured
3
The assumption of in…nite lives will not be important for welfare considerations. Beyond the standard
arguments (e.g., altruism, absence of annuities etc.), recent work speci…c to bankruptcy by Gross and Souleles
(1998) argues that “risk-composition”, a measure which adjusts for, among other things, changing account
age distributions, only accounts for a small amount of the observed rapid rise in bankruptcies and its
persistently high levels over the past …ve years.
4
See Gropp, Scholz, and White (1997).
5
Conversely, while sometimes large, exemptions seem by themselves extremely unlikely to a¤ect the
decision to own a home. Thus, assuming exogenous homeownership is likely to be a useful approximation.
4
credit market, and can save in a risk free asset. I follow Chatterjee et. al. (2002) and Livshits
et al. (2003), and assume that lenders price default risk for unsecured debt by being aware
of the overall debt position of a household, including the level of both secured and unsecured
debts held at any given time. Lenders are assumed to be unable to observe shocks to household endowments, and so face default risk on the unsecured contracts they enter. However,
lenders diversify default risk perfectly by holding a large loan portfolio of unsecured debt
contracts. That is, lenders are aware of the proportion of the loans of a given type that will
be defaulted on, which in turn allows loan pricing to be set such that pro…ts are zero with
probability one. By contrast, secured debt in the model is (by de…nition) risk-free for the
lender, and is therefore invariant to a borrower’s net-worth. All intermediation is assumed
to be costly, with proportional and di¤erential transactions costs for secured and unsecured
intermediation, denoted ¿ s , and ¿ u respectively.
Secured debt denoted as · 0, is debt that agents cannot default on in a bankruptcy
proceeding. Secured debt represents the sum of all mortgage debt and home equity loans
held by a household, and is o¤ered at interest rate Rs . Unsecured debt, denoted au · 0, may
be wholly discharged in a bankruptcy proceeding, provided that no equity is required to be
applied towards it. Unsecured debt represents credit card debt and other non-collateralized
loans, and carries an interest rate of Ru(º) that depends on the level and composition of a
household’s secured and unsecured debts, denoted by º = (a s; a u). Agents may also save in a
risk-free asset by choosing au ¸ 0, and will receive interest payments at a gross deposit rate
of Rd. By de…nition, secured debt must be backed by collateral, and therefore households
face …xed credit limits in secured debt of jas j.
2.4. Bankruptcy
Bankruptcy is modeled as a decision that removes unsecured debt in exchange for assets
above the exemption-mandated threshold. Bankruptcy is also assumed to impose a cost
on the household. Therefore, bankruptcy in the model is analogous to the U.S. Chapter 7
“Fresh Start” provision. This form of bankruptcy constitutes over 70% of all …lings, and
accounts for nearly all debt discharged.
In the model studied here, there are two types of costs associated with bankruptcy. First,
and of primary interest here, is the cost of giving up all non-exempt assets. Second, there
are explicit costs such as legal fees and time costs of court dates, and implicit costs, of which
“stigma” appears to be relevant (see Dubey, et al. [2001]) and Gross and Souleles [2000]).
Bankruptcy provides insurance, but beyond the transfer of assets triggered by exemp
tions, bankruptcy costs are “deadweight” in nature. The use of such penalties occurs in
part because it is often di¢cult to seize wealth, most obviously because nearly all …lers have
5
negative worth. Furthermore, penalties such as wage garnishing allow resource transfers but
act as a tax on labor e¤ort, and perhaps because of this have been severely restricted by law
in many states (see Baird [2001]).
The penalties for bankruptcy, given their deadweight nature, can most tractably be
represented as reducing the utility of a household that …les. This is the approach taken in
both Zame (1993) and Dubey et al. (2001). Additionally, because the marginal response
of consumption, welfare and interest rates to changes in exemptions may vary with the
…ling rate, I require that the model include all relevant costs of bankruptcy beyond those
explicitly related to exemptions. However, the precise composition of these various costs is
not necessary for understanding how changes in exemptions a¤ect outcomes. Let ¸ denote
all non exemption related costs of bankruptcy.
I will calibrate ¸ to match observed bankruptcy …ling rates among homeowners under
current exemptions. It should be made clear that the notion of costs used here includes
all the above penalties for bankruptcies. In particular, as it is true that households do
face at least temporary di¢culty in borrowing following a bankruptcy, ¸ implicitly includes
the imputed utility cost of being shut out of credit markets.6 In this way, credit markets
e¤ectively keep track of history in a way that does not require credit status to be retained
as a state variable.
2.4.1. Exemptions
Exemptions are rules governing the maximum amount of wealth that may be retained by a
bankruptcy …ler. Any wealth above the exemption must be surrendered and used to satisfy
unsecured creditors. Exemptions in this model have two e¤ects on borrowers. First, they may
provide risk-sharing bene…ts by keeping the consumption of agents smooth after bankruptcy.
Second, exemptions can damage risk-sharing and consumption smoothing, to the extent that
they increase the costs of borrowing on the unsecured credit market. The latter e¤ects will
occur for two reasons. First, conditional upon default, unsecured lenders may lose more in
environments with high exemptions than low exemptions. Second, bankruptcy rates may
increase with exemptions, thus increasing losses.
Let e > 0 denote the exemption level, such that any equity above e, is seized and used
6 The
approach taken here is also similar to Livshits et al. (2002), who employ costs of bankruptcy,
and do not explicitly model exclusion from credit markets. Additionally, I …nd in a related environment
(Athreya [2002]), that in order to match the data, penalties that are ex-ante restricted to only exclusion
from credit markets imply penalty periods in excess of twenty-…ve years under reasonable parameterizations
of preferences, market incompleteness, and idiosyncratic risk. This is not sensible as, among other things,
bankruptcy disappears entirely from one’s credit record after ten years. This indicates that the other costs
above play an important role.
6
to repay unsecured creditors. Given the pre-bankruptcy equity position of the household, e,
and the pre-bankruptcy value of unsecured debt, jauj, the term (e ¡ e) is therefore “excess”
equity, which must be applied towards paying o¤ unsecured debt in bankruptcy. Equation
2.3 below summarizes the post-bankruptcy equity position eb ¸ 0, as a function of prebankruptcy asset holdings.
eb =
(
e if e < e
max(e; e ¡ jau j) if e > e
(2.3)
Secured debt after bankruptcy, denoted asb is therefore given as:
8
>
<
as if e < e (No Excess Equity)
asb =
as ¡ (e ¡ e) if e > e and (e ¡ e) < ja uj (Excess Equity < Unsec. Debt)
>
:
a s ¡ ja uj if e > e and (e ¡ e) ¸ ja uj (Excess Equity ¸ Unsec. Debt)
(2.4)
The …rst line of 2.4 applies to cases where a debtor has equity below the exemption level
and so will not surrender any equity or repay any of his unsecured debts. This is the case
where the bene…ts of Chapter 7 bankruptcy exemptions are maximized. It is precisely this
aspect of Chapter 7 that demonstrates how exemptions, all else equal, discourage wealth
accumulation and encourage the use of unsecured debt. The second line of 2.4 refers to the
case when a debtor has enough equity to transfer some, but not all, of his unsecured debts
to secured debts. The unsecured creditor receives the equity in excess of the exemption.
Therefore, the household’s secured debt after bankruptcy will increase by the amount of
excess equity, but his unsecured creditors will not be fully repaid. The third line in 2.4
covers the case where a debtor’s equity exceeds the exemption by more than his unsecured
debts. In this case, a debtor will be required to transform all his unsecured debt into
secured debts, leaving him with secured debts that increase by the amount of his unsecured
debt, while his unsecured creditors are fully repaid. Chapter 7 bankruptcy is not useful for
households in this category.
Because the payo¤ to, and hence the likelihood of, bankruptcy depends on the particular
portfolio of debts held by a household, the price of unsecured debt will as well. A simple
example shows why. Consider an agent with equity of $18,000, no current unsecured debt
and who faces an exemption of $10,000. Assume that this agent wishes to borrow $8,000.
A total debt level of $8,000 can be achieved in a variety of ways. For example, obtaining
$8,000 of unsecured debt will leave the agent with equity of $10,000 following bankruptcy, as
the agent would be forced to liquidate the $8,000 in equity above the exemption and repay
unsecured creditors. This implies that the “…nancial bene…t” to bankruptcy is zero. In
equilibrium, creditors’ beliefs over repayment will internalize this, and such a portfolio will
7
generate an interest rate for unsecured debt that is equal (up to transactions costs) to that
on secured debt. Conversely, if the agent above began instead with equity of any amount
weakly less than the exemption, the zero-pro…t interest rate for an unsecured loan of $8,000
would be higher, as the …nancial bene…t from bankruptcy is $8,000 in the current period. In
other words, holding equity above the exemption allows an agent to reduce the number of
income states where bankruptcy is optimal, but of course commits him more strongly to a
repayment schedule that may not be appealing ex-post.
2.5. Timing and The Recursive Formulation
Labor income y is received at the end of each period, and is available for consumption at
beginning of the following period. Let the total beginning-of-period asset holdings of the
agent be de…ned by a ´ (as + au). Given a and y, the agent must choose consumption and
a level of savings or borrowing. If he chooses to save, he sets a0u > 0 and earns the risk-free
interest rate on savings Rd .7 If he chooses to borrow, he must choose a portfolio of secured
and unsecured debt º 0 = (a0s ; a0u). The interest rate charged on unsecured debt is denoted
Ru(º 0 ), and re‡ects default risk that depends on the debt portfolio. The asset choices then
residually determine current period consumption c through the budget constraint.
Once the consumption and savings decisions have been made, the income shock y0 is
realized, and the agent chooses either to remain solvent or …le for bankruptcy.8 The e¤ect of
remaining solvent is that the agent’s wealth remains una¤ected as the next period arrives,
with a0 = a0s + a0u. By contrast, the bene…t from …ling for bankruptcy is that the household
has its unsecured debts eliminated, in return for any wealth above the exemption. The
other cost for the removal of these debts is the penalty ¸. The alteration to wealth from
bankruptcy is de…ned by the map asb(:) de…ned in 2.4, and depends on the portfolio º 0 and
the exemption e. The period then ends.
The problem above can be expressed in a very compact recursive manner primarily because bankruptcy is modeled as altering only within-period variables. The income shock y0
and total wealth a0 constitute next period’s state vector because they jointly determine the
resources available for the agent as well as the expectation of next period’s income. Denote by V (a0 ; y0 ) the continuation value for solvent agents from ending a period with state
(a0 ; y0 ). Given the penalty, the value of …ling for bankruptcy at the end of the current pe
riod is V (a sb0(º0 ; e); y0 ) ¡ ¸. Taking these values as given, the agent chooses portfolio º0
7 As
I focus on a stationary representation of the agent’s problem, I drop time subscripts and use primes
to denote all variables one-period ahead.
8
For notational ease, because the income shock is assumed to be realized after the consumption/savings
choice is made, we drop primes from current consumption and beginning-of-period income, and instead
denote income received at the end of the period by y 0.
8
with consumption determined by 2.6, where the indicator function I(a0u < 0) determines the
applicable interest rate. Therefore, the value function V (a; y) satis…es the following.
V (a; y) = maxfu(c) + ¯Eº0 max[V (a0 ; y0 ); V (asb0 (º0 ; e); y 0) ¡ ¸]g
(2.5)
s.t.
c+
where
a 0s
a0u
+
·y+a
Rs Ru(º 0)I(a0u < 0) + Rd (1 ¡ I(a0u < 0))
º 0 ´ (a0s ; a0u)
(2.6)
(2.7)
The expectation in 2.5 can be expressed more precisely as follows. For any given portfolio
º 0, we can de…ne the set Y B (º 0) to be the set of income values that make bankruptcy
preferable to solvency. That is,
Y B(º 0 ) ´ fy0 2 Y jV (asb0 (º 0 ; e); y0 ) ¡ ¸ > V (a0 ; y0 )g
(2.8)
Given V (:), º 0 , and Y B (:), the term Eº0 max[V (a0 ; y0 ); V (asb0 (º 0 ; e); y0 ) ¡ ¸] can therefore
be easily calculated.
2.6. Equilibrium
I follow the recent work of Chatterjee at al. (2002), Livshits, MacGee and Tertilt (2003), and
Athreya and Simpson (2004) and study recursive equilibria where the risk-free rate of interest
on savings, Rd , is exogenous. To motivate this assumption, note …rst that cost of risk-free
funds applicable to the intermediaries in the model depends in principle on the aggregate
capital stock. However, the U.S. capital stock is overwhelmingly concentrated among a small
measure of households (e.g. the top 5% of households own 55% of the wealth). This small
group is unlikely to be in‡uenced by presence of personal bankruptcy law. Second, it is
well known that standard incomplete market models are typically incapable of matching the
concentration of wealth holdings.9 Therefore, general equilibrium analysis with such a model
would be potentially misleading, as it would tend to overstate the role of bankruptcy law
for aggregate outcomes, and in turn, to overstate the role of changes in costs of funds in
determining the level of borrowing. For these reasons, a better approximation is likely to
emerge by abstracting from the determination of the capital stock and implicitly assuming
that it is held exogenously by a small measure of households who are una¤ected by personal
9 See,
for example, the discussion in Castaneda, Diaz-Jimenez, and Rios-Rull (2003).
9
bankruptcy. Such a set-up is equivalent to the pure-endowment, exogenous cost-of-funds
model developed here.
Because households face idiosyncratic risk, and have an incomplete set of assets to insure
themselves, heterogeneity along asset holdings and consumption will emerge even if they
are ex-ante identical. More precisely, let Q(x; Z) be the transition function determining the
probability that, for any household, next period’s state lies in a set Z , given that the current
state is x. It is assumed that the heterogeneity across agents converges to a stationary
distribution, denoted ¹. This distribution yields the time-invariant fraction (measure) of
households over appropriate subsets of the state space, in our case levels of net asset holdings
and current income. By de…nition, ¹ is stationary under the transition function Q(x; Z ) if
R
it satis…es the condition ¹(Z) = x Q(x; Z)d¹.
Given a stationary distribution, equilibrium requires meeting two further conditions.
First, the decisions of agents, taking interest rate functions and bankruptcy law as given,
are optimal. Second, the intermediary must make zero pro…ts. The …rst condition is auto
matically satis…ed when decisions derive from the Bellman equation 2.5. I turn now to zero
pro…ts.
2.6.1. Zero Pro…ts
Because it is risk free, secured debt must only di¤er from the deposit rate by the cost of
intermediation ¿ s . That is, Rs = Rd + ¿ s. Unsecured loans are risky at the individual level,
but in the absence of aggregate risk, diversi…able. Credit intermediaries are assumed to be
competitive price-takers that hold diversi…ed loan portfolios. As stated earlier, lenders are
assumed to observe the amount of both secured and unsecured debt issued by the household
in a given period. This leads to the use of loan contracts that follow Livshits et al. (2003),
and Chatterjee et al. (2002).
Given a household’s debt portfolio º0 = (a0s ; a0u), the household obtains unsecured credit
in the current period by issuing one-period bonds. These bonds are discounted by the
market according to the likelihood of default. Given the interest rate on unsecured loans
Ru(º 0 ), the net interest rate is given by r u(º 0) ´ Ru(º 0) ¡ 1. Next, de…ne ¼bk (º 0) to be
the equilibrium probability of default given debt portfolio º 0 . In equilibrium, the zero pro…t
condition then implies that, given a net cost of funds r d ´ Rd ¡ 1, a default probability
¼bk (º 0 ) and transactions cost ¿ u, the net interest rate on unsecured loans must satisfy the
following:
ru(º 0 ) =
(rd + ¿ u )
(1 ¡ ¼bk (º 0 ))
(2.9)
The timing convention used here allows us to determine the probability of default under
10
a portfolio º 0 by simply computing the probability that end-of-period income will fall in
the set Y B(º 0 ). Because current income is not observable, the probability of bankruptcy is
computed by lenders using the unconditional distribution of income fy¤ .
X
¼bk (º 0) =
yj0 fy¤j
(2.10)
yj 2Y B (º0 )
with Y B (º 0 ) de…ned by equation 2.8. In summary, equilibrium is de…ned as follows.
De…nition 2.1. A recursive (partial) equilibrium of the model is a pro…le
fr d; V ¤(a; y); º0¤ (a; y); Y B¤(º 0¤); r u¤ (º0¤); ¼bk¤ (º0¤ ); ¹¤ (Z)g, whereby:
1. The decision rule, º 0¤(a; y), solves 2.5, subject to 2.6, and yields value function V ¤(a; y).
2. The bankruptcy probability satis…es: ¼bk¤(º 0¤ ) =
yj
3. The interest rate function satis…es: ru¤(º 0¤) =
P
2Y B¤(º0¤ )
yj0 fy¤j
(r d+¿ u )
(1¡¼bk¤ (º0¤ ))
4. ¹¤ (Z) is stationary, and therefore satis…es: ¹ ¤(Z) =
2.7. Welfare
R
x
Q(x; Z)d¹ ¤
The welfare criterion used here is ex-ante expected utility, is denoted by ¤ and is given
below.
¤=
Z
(2.11)
V (x)d¹
To compare how much better or worse o¤ households are under various exemption policies,
I use the measure above to answer the following question. What constant proportional in
crement/decrement to benchmark consumption at each date and state under the benchmark
exemption law would yield the same expected lifetime utility as the consumption allocation
under a proposed exemption policy? More precisely, let the pair (¤B ; fcB g) denote welfare
and the consumption process under the benchmark exemption, and let (¤p ; fcpg) denote
welfare and the consumption process under a proposed exemption policy. Next, let Á denote
the increment/decrement to consumption at each date that solves the following problem:
E0
1
X
t=0
¯
B 1¡®
t (Áct )
1¡®
¡1
= E0
1
X
t=0
¯
p 1¡®
t (ct )
¡1
1¡®
(2.12)
For ® > 1, using the de…nition of ¤B and ¤p , it is easily shown that Á is given by:
11
Á=
Ã
¤p +
¤B +
1
(1¡®)(1¡¯)
1
(1¡®) (1¡¯)
! 1¡®
1
(2.13)
Therefore, if, for example, Á > 1, households require an augmentation of the benchmark
consumption process, implying that a proposed policy improves on the benchmark.
3. Parameterization
The values of risk-aversion and discounting, denoted ® and ¯ respectively, are set to values
that are standard in the literature. I follow Huggett (1993), Aiyagari (1994), and others to
set ® = 3, and ¯ = 0:96. The transactions costs parameters, ¿ s and ¿ u are set such that, in
equilibrium, the model matches observed interest rate spreads between secured and unsecured
borrowing. Evans and Schmalensee (1999) estimate that among credit card issuers, costs of
servicing accounts are roughly 5.3% of total costs, but that these costs are partially o¤set by
interchange revenues of 1.9%, implying a net transactions cost of at least 3.4%. I therefore
set ¿ s =0.034. However, for unsecured lending, there appear to be other costs involved in
intermediation in addition to those faced by secured lenders. In particular, the observed
spread between credit card and home equity loans has averaged approximately 8 percentage
points, and is much larger than can be justi…ed by default risk alone. For example, the
annual fraction of credit card loans that are defaulted on has averaged 4.7% since 1991,
while that on all loans secured by residential real state has averaged roughly zero.10 In order
to analyze a benchmark economy where the cost di¤erential between secured and unsecured
data approximately accords with the data, I treat the di¤erential, 8%-4.7%¼ 0:03%, that
remains after default costs as an additional transactions cost for unsecured lending, and so
set ¿ u = ¿ s + 0:03 = 0:064.11
The exemption level, e, is the de…ning statutory restriction on Chapter 7 Bankruptcy
…lers.12 The single largest exemption available to …lers is nearly always the Homestead
10
The
default
rate
on
secured
debt
has
averaged
0.16%
in
the
same
period.
See
http://www.federalreserve.gov/releases/chargeo¤/chg_sm_sa.txt
11 Additional costs faced by unsecured lenders include the cost of recovering bad debts through court
procedures, as well as the frequent need to update information on borrower characteristics to price loans
according to risk. Edelberg (2003) carefully documents such risk-based pricing. Also, model outcomes are
robust to much lower levels of this cost (e.g. 0.01 units), but the present parameterization produces borrowing
at rates consistent with the data. Lastly, while a non-competitive markup might also “explain” the data,
the unsecured lending market appears quite competitive (see Evans and Schmalensee (1999)).
12
The other restrictions involve the number of years for which one cannot re-…le (6 years), and how
long bankruptcy can remain on one’s credit record (10 years). Virtually all other penalties are issued by
either credit markets (borrowing restrictions, and high interest rates), or through social sanctions and other
“stigma” e¤ects.
12
Exemption. This provision protects some or all of an individual’s home equity from seizure
by creditors, even when they have substantial uncollateralized debts. Other exemptions
include personal property, equity in automobiles, and exemptions for the “tools-of-trade”.
While there are federal guidelines governing exemptions, states were initially given the option
to opt out of these rules and impose their own. After 1978, when the federal law went into
e¤ect, all but eleven states “opted-out” of these federal regulations. However, although
almost all states opted out, the exemption provisions they chose varied enormously. For
example, the current federal homestead exemption is $16,150, while the state exemption
varies from $2,000 dollars in South Carolina, to essentially unlimited in Texas and Florida.13
Given the inter-state variation in exemptions documented above, it is more useful to
study exemptions in proportion to the population they apply to. In particular, Table 4 shows
that the median exemption, in terms of applying to 50% of U.S. households, is $12,500. In
1997, nominal income per person was $19,241 and per-capita nominal household income was
$50,411.14 Normalizing mean household income to unity implies that the baseline exemption
e be set at approximately 0.25.15
For the income process, I set N=3, whereby endowments/income, y,
e may take on three
values. Income is assumed to follow a markov chain with transition probabilities that are set
to capture the high level of serial persistence documented in recent empirical work, such as
Hubbard et al. (1995), Floden and Linde (2001), as well as Storesletten, Telmer, and Yaron
(2001). The benchmark process is given as: ye ´ fyl ; ym; yhg = f0:6; 0:8; 1:22g; with the
transition matrix de…ned as:
£=
yl0
y0m
0.700
0.175 0.125
ym 0.015
yh 0.000
0.920 0.065
0.070 0.930
yl
yh0
(3.1)
This process implies a serial correlation parameter of approximately 0.86 in an AR(1)
model, and generates a unconditional coe¢cient of variation of 0.22, which is in the range ex
plored by Aiyagari (1994), and others. The stationary distribution implied by this transition
matrix is given by fy¤ = [0:0242; 0:4836; 0:4922].
13
No exemptions are truly unlimited however. In states with unlimited exemptions, there are typically
restrictions on who may qualify. In Arkansas, for example, only those with a homestead smaller than 1/4
acre may qualify.
14
Source: Federal Reserve Board of Governors: http://www.bog.frb.fed.us/releases/h15data.
15 The Senate’s proposal, as it seeks to eliminate “opt-out”, and allows a range for state exemptions, does
not ultimately produce truly uniform nationwide exemptions. However, to study the e¤ects of uniform
exemptions, I study outcomes that obtain from varying the maximum allowable state exemption with “noopt-out”.
13
The …nal parameter, as , is the credit limit on secured credit. In 1997, the median price
of existing housing was approximately $120,000, slightly greater than twice annual mean
household income.16 This imposes a natural limit on secured credit at 2.0 units. The limit
on unsecured credit is endogenous, but the benchmark calibration is chosen to be close to
the median level of median credit card debt discharged in bankruptcy, which has remained
between 30% and 40% of median household income (see Sullivan et al. (2000), and Bermant
and Flynn (1999)). All model parameters beyond those governing the income process are
listed in Table 1.
With respect to bankruptcy, I target the benchmark …ling rate as follows. The shocks
hitting households in the model are to be interpreted as income shocks arising from the
labor market outcomes of job loss, overtime, displacement etc. I do not explicitly model
other shocks a¤ecting households such as catastrophic medical shocks or law suits. The
latter are in the nature of “expense shocks”, and are studied in some detail by Livshits et
al. (2002). Instead, I focus here on debt and equity positions arising from consumption
smoothing behavior in the face of non-catastrophic events.
Chakravarty and Rhee (1999) …nd that approximately 33% of total …lings occur amongst
households who have recently experienced catastrophic health events and/or lawsuits, while
Sullivan et al. (2000) …nd that 19% of …lings are associated with major health shocks.17
I therefore compromise, and assume that one-fourth of all bankruptcy are associated with
“expense” shocks. According the Administrative O¢ce of the U.S. Courts, the overall …ling
rate among U.S. households has averaged 1.3% for the ten-year period 1994-2003, of which
70% have consistently been Chapter 7 …lings. However the distinction between Chapter 7 and
Chapter 13 bankruptcies is not always clear in practice, for two reasons. First, households
are allowed to choose between them, and given that Chapter 7 yields a complete discharge of
unsecured debt, the debt rescheduling implicit in a Chapter 13 cannot be too strict. Secondly,
a Chapter 13 bankruptcy can always be converted to a Chapter 7 …ling. Therefore, the target
…ling rate for must lie between approximately 0.7% (if all Chapter 13 …lings are ignored)
and 0.97% (if all Chapter 13 …lings are included). For concreteness, I err on the side of
making a distinction between Chapters 7 and 13, and therefore set the target value for
bankruptcy at 0.80% (and implicitly allow for approximately 10% …lings to be Chapter 13
bankruptcies where debt is repaid). However, homeowners, are a subset of the population,
at approximately two-thirds. This implies that the relevant target for the benchmark model
is ¦=0.80%. 18 I measure all rates from the …rst quarter of 1994 through the fourth quarter
16
U.S. Dept.of Housing and Urban Development at: http://www.huduser.org/periodicals/ushmc/spring97
/histdata.html
17
See also Domowitz and Sartain (1999).
18
That is, the target …ling rate is: 1.3%x67%x90%¼0.8%.
14
The results are robust to the bankruptcy
of 2003.
4. Results
A …rst result is that exemptions appear to in‡uence …ling rates non-trivially, consistent with
the claims of NBRC dissenters. In Table 2, we see that in the benchmark case, …ling rates rise
from 0.66% under zero exemptions and nearly double to 1.24% under maximal exemptions.
What is striking is that this occurs despite the increased costs of borrowing associated with
more generous exemptions. Given that the price of debt and the probability of bankruptcy
are invertible maps of each other, as given in 2.9, I focus on the latter, as it provides more
direct intuition for the behavior of prices. We see from Figure 2 that for a given level of
unsecured debt, default likelihoods grow systematically with equity as exemptions rise. For
example, under maximal exemptions, the probability of default is substantial for even fairly
wealthy households, as de…ned by nontrivial equity holdings. In other words, exemptions
make unsecured credit costlier to obtain, a¤ecting most those with low or moderate equity
levels. This is consistent with the …nding of Gropp, Scholz, and White (1997) that exemptions
shift relatively cheap credit away from low-wealth households towards those with higher costs
of bankruptcy, namely, high wealth households.
From a welfare perspective, changes in consumer welfare arise in this model from changes
in the volatility of consumption. However, these changes are relatively minor. To evaluate
consumption volatility, I de…ne consumption “smoothing” to be the ratio of the coe¢cient of
variation of consumption relative to the coe¢cient of variation of income. Measured in this
way, Table 2 shows that consumption smoothing improves very slightly, but monotonically,
with exemptions. As exemptions rise from zero to the entire value of the collateral asset,
this ratio falls from 0.793 to 0.783. Therefore, again in Table 2, we see that welfare rises
monotonically with exemptions as well, from a welfare loss of $11.40 when exemptions are
eliminated, to a gain of $28.24, under a full exemption.
As stated above, the price of obtaining a given level of unsecured debt rises monotonically
with exemptions. However, as seen in Table 2 (in the column denoted “Avg. Uns. Debt”),
the willingness of households to borrow does not increase monotonically with exemptions,
but is rather an “inverted-U”. Namely, initially, as exemptions become more generous but
are still low in absolute size, households on average borrow less on the unsecured market,
choosing to save more to avoid bankruptcy. Subsequently, for more generous exemptions,
household unsecured debt rises with exemptions, as bankruptcy becomes less painful to
avoid. However the changes are minor, as most households do not hold unsecured debt at
all, leading average debt to change from -0.01636 to -0.01756 units, less than $50 annually.
target.
15
Nonetheless, despite the behavior of average unsecured debt, maximal unsecured debt
does increase monotonically with exemptions. Figure 3 and Table 2 (in the column denoted
“Max. Uns. Debt”), both document maximal unsecured debt held by agents across experi
ments. This behavior shows that low exemptions systematically discourage large individual
debt accumulation. It is striking that the value of maximal debt essentially doubles from
-0.19 units to -0.38 units as exemptions are raised from zero to the maximum. This change
is equivalent to an increase of $7,000.
Along with an increased willingness to hold unsecured debt under generous exemptions
is the increased willingness to simultaneously retain equity. First, note that while per-capita
unsecured debt is higher under maximal exemptions than it is under an exemption of zero,
per-capita equity holdings still rise. The column denoted “Avg. Equity” reveals the small
increase in average equity holdings in the population, which rises by approximately $1,200
on average, as exemptions go from zero to $120,000.
The behavior of per-capita values is suggestive, but more intuition for the results can
be gained from combining the information from Figure 3 with data on speci…c subsets of
the distribution of “net-worth” (the sum of debts and savings), given in Figure 4. It can
been seen here that under low exemptions (e.g. e=0.0, and e=0.05), there is more mass at
very low levels of net worth, than there is under both moderate and high exemption levels.
Combining this with the fact that maximal unsecured debt rises systematically with exemp
tions accounts for why the equity held in bankruptcy rises with exemptions. In other words,
for high exemptions, households with deeply negative net worth typically choose more unse
cured debt and less secured debt (i.e. more equity) than they would under low exemptions.
This illustrates the shift in …nancing away from secured debt and toward unsecured debt in
…nancing used by households to smooth consumption. Exemptions generate exactly these
incentives, with the only o¤setting factor being the increased cost of unsecured debt.
A summary of the results so far, and intuition for them, is as follows. While the cost
of bankruptcy is …xed in the model by ¸, higher exemptions steadily increase the bene…ts
of …ling. In turn, as seen in both Figures 2 and 6, unsecured loan interest rates must rise
for an increasingly large set of unsecured debt and equity levels to satisfy the zero-pro…t
condition 2.9. Despite this, Figure 3 shows that maximal equilibrium unsecured borrowing
still increases monotonically with exemptions. Therefore, even though most households
do not change their behavior substantially with exemptions (see Figure 4), the proportion
of households who …nd bankruptcy optimal grows and leads to higher bankruptcy rates.
In other words, what matters for bankruptcy rates is the behavior of maximal unsecured
debt, not average debt, and the former is strongly in‡uenced by exemptions. Moreover,
consumption, and consequently welfare, improve marginally as exemptions are increased.
This shows that households …nd that ex-ante, high exemptions are worth their price in
16
terms of more expensive unsecured credit. I turn now to the question posed at the outset.
Namely, does bankruptcy provides an “excessive” advantage to households that use it?
4.1. Head Start or Fresh Start?
There are two tangible dimensions along which to answer the question of whether bankruptcy
provides a “Fresh Start” or a “Head Start”. First, will many households “...escape their
contractual obligations while maintaining levels of wealth that the vast majority of Americans
do not enjoy”, as suggested by the NBRC dissenters? Second, for a given wealth position,
do households bene…t merely by increasing their holdings of unsecured debt, even when they
are unlikely to repay? The answer to the …rst question turns out to be a quali…ed “yes”,
while the answer to the second is “no”.
While exemptions do not a¤ect average welfare much, they do appear to encourage
bankruptcy and, ex-post, provide high-debt households with a “head start” by allowing them
to retain wealth in bankruptcy. Evidence for this is seen in the sequence of histograms in
Figure 1. These panels document the equity held by households at the time of a bankruptcy
…ling. As exemptions rise, there are increasing numbers of cases in which a bankruptcy …ler
retains equity, with the amount increasing in exemptions. Figure 1 documents that under
the maximal exemption, for example, beyond the fact that the bankruptcy is nearly double
the rate under zero exemptions is the fact that there is signi…cant mass on very high levels of
equity, at up to 1.5 units, or $90,000! This …nding also receives some empirical support. In
the data used by Berkowitz and Hynes (1999), …lers who did have assets were nearly always
from Texas, a state with a (nearly) unlimited homestead exemption.19
Another feature of the results is that the e¤ect of exemptions on the equity held at the
time of a bankruptcy …ling is strong for low exemptions, but weak for high exemptions. For
example, Figure 1 shows that when exemptions are raised from 0.0 to 0.57 (a value that is
one-half of the inter-state average of 1.14 or $68,000), maximal equity held in bankruptcy
grows substantially, while when exemptions are raised from 1.5 units to their maximal value
of 2.0 (in each case an increase equivalent to approximately $30,000), the equity held by
…lers increases only slightly.
19
The model developed here is meant to capture the e¤ects of uniform, nationwide exemptions, and is
calibrated to match country-level data for the U.S. Thereore, a inter-state interpretation of the model is not
without pitfalls. States di¤er not just by their exemptions, but also in many other ways such as: the income
processes of their inhabitants, the generosity of social insurance such as unemployment, welfare, housing
support etc, the rules placed on credit supply, such as interest rate ceilings, and by the di¤erential e¤ects
of aggregate shocks. Many of these factors vary substantially. Additionally, households may even move
to exploit these di¤erences, i.e. “Forum Shop”. Future work applying this model carefully to account for
interstate variation seems useful.
17
A natural way to address the second question above of whether increased unsecured debt
makes households better o¤, is to check if the value function of the household is increasing
in debt. Figure 5 displays the value function V (:) under benchmark exemptions. As is
easily seen, this function is strictly decreasing in debt. In fact, the function is most sharply
decreasing for low income households, who constitute the majority of bankruptcy …lers.
Secondly, while households are free to acquire unsecured debt regardless of their equity
position, the limit on secured debt is -2.0 units. In the area to the left of the dashed line all
reduction in net worth must therefore come from increases in unsecured debt alone. It is clear
from Figure 5 that such increases in unsecured debt reduce indirect utility monotonically,
and do so for all three values of the income shock. This turns out to be true for all exemptions
levels studied here, but for brevity, I do not present all cases. A second fact that suggests
that lax exemptions do not provide a “head start” is to see that from a welfare perspective,
the ex-ante consumption smoothing bene…ts of being able to discharge unsecured debt and
retain wealth ex-post appear to be nearly o¤set by the costs arising from increased use of
bankruptcy.
Therefore, bankruptcy is chosen by those with large debts, relatively low equity, and
low income. Furthermore, even large exemptions, while encouraging …ling and the retention
of wealth in …ling, do not appear to give debtors a “head start” along the dimensions of
household welfare. In particular, welfare increases with exemptions even though ex-post,
high exemptions will result in households …ling for bankruptcy while retaining wealth. 20
4.2. Risk-Aversion and Risk-Sharing
The primary con‡ict generated by exemptions is between the bene…ts of better risk-sharing
through the retention of wealth after bankruptcy and the impediment to risk-sharing created
by higher unsecured interest rates. Therefore, the measured gains or losses will be in‡uenced
most directly by the assumed level of risk-aversion. For robustness, in Table 3 I study
outcomes when risk aversion is lowered to 2, from the benchmark value of 3. By and large,
the results from the benchmark go through unchanged. Namely, that a high exemption is
(marginally) welfare improving relative to low exemptions. Second, and not surprisingly, the
optimal exemption is lower than under benchmark risk aversion. In other words, because the
bene…ts from smoothing are lower, the willingness of households to endure more restricted
borrowing is also lower. Thirdly, the “supply-side” response, as measured by the how interest
rate functions change with exemptions, is monotone, as before. In Figure 6, the set of debt
levels for which interest rates contain a default premium grows monotonically. Fourth, again
as before, equity holdings are increasing in exemptions, and increase by roughly the same
20
This feature is robust to the exemption level.
18
absolute amount (though a greater amount, proportionally) as in the benchmark.
There are also some di¤erences, though they result in only minor changes in the welfare
implications of exemptions. First, as in the benchmark, consumption smoothing improves in
general as exemptions are increased. However, the improvement is not strictly monotone as
in the benchmark. Moreover, unlike the benchmark case, the bankruptcy rate is no longer
monotone in exemptions. The intuition is as follows. In economies with very low exemptions,
fewer agents accumulate large, unsecured debts. This occurs despite the fact that relative
to settings with higher exemptions, unsecured credit is much cheaper to obtain. To see this,
compare the histogram of unsecured debt holdings in Figure 7, with its benchmark analog,
Figure 3. As exemptions are increased from zero to the full value of the collateral asset.
Both average unsecured debt and maximal unsecured debt incurred increase dramatically.
In fact, maximal unsecured debt nearly doubles relative to the benchmark case. Therefore,
bankruptcy rates are lower under very low exemptions than under very high ones. As
exemptions are increased from zero, this changes, and while average debt falls, maximal
debt and bankruptcy rates increase. However, for exemptions near the benchmark level,
per-capita unsecured debt falls su¢ciently to cause …ling rates to fall once again. Lastly, as
exemptions are increased substantially beyond the benchmark, the e¤ects of credit supply
dominate, and once again lead to increased borrowing and bankruptcy. This is seen in
Figure 7. The tension between incentives provided by exemptions to acquire unsecured debt
and the disincentives coming from high unsecured interest rates is resolved in favor of more
debt in the benchmark. Under lower risk aversion, the trade-o¤ varies more non-trivially.
However, as made clear earlier, the net welfare implications in favor of high exemptions
remain unchanged. Sensitivity analysis was also conducted with a lower-persistence income
process, transactions costs, and also with higher targets for the benchmark bankruptcy rate.
None of these alternative parameterizations changed the results substantively, and for brevity
are not reported here.
4.3. Related Work
Before concluding, it is useful to compare the results above with existing equilibrium ap
proaches to bankruptcy. The …nding here that welfare improves with higher exemptions is
consistent with the single-asset model of Zha (2001). By contrast, in a rich model of bank
ruptcy that includes labor supply, chapter choice and capital accumulation, Li and Sarte
(2003), …nd that welfare improves with lower exemptions. In their model, these gains come
in part for reasons other than pure risk-sharing, and arise instead from supply-side e¤ects.
Lower exemptions encourage labor supply and discourage the rescheduling of debt in “Chap
ter 13” bankruptcy, which in turn increases labor supply, capital accumulation, output, and
19
consumption. While the adjustment of capital to bankruptcy may be overstated in Li and
Sarte’s framework, the response of labor supply still moves in the direction of increased output and consumption. Conversely, because I allow loan prices to be conditioned on debt and
equity holdings, higher exemptions do not lead to uniformly higher loan prices, as they do
in Li and Sarte (2003). Therefore, the cost imposed on borrowers does not imply worsened
consumption smoothing for all households, but rather only for those with low wealth. This
acts to o¤set the welfare costs associated with high exemptions found in Li and Sarte (2003).
5. Conclusions
In this paper, I evaluate uniform exemption policy primarily within the context of the recent
congressional proposal H.R. 975. I develop an incomplete markets equilibrium model where
secured and unsecured assets coexist and are treated di¤erentially in a bankruptcy proceed
ing. I …nd that exemptions are associated positively with …ling rates, equity holdings in
bankruptcy, and welfare. I …nd however that exemptions are strongly negatively associated
with the availability of unsecured credit. Additionally, I …nd that the welfare bene…ts of high
exemptions, while positive, are small, at roughly $28 per household annually. The results
are robust, and show that, from an ex-ante welfare perspective, increases in bankruptcy
exemptions beyond current state averages are largely a matter of indi¤erence, and do not
merit the heated debate they have generated.
20
6. Bibliography
Aiyagari, S.R., “Uninsured Idiosyncratic Risk and Aggregate Saving”, Quarterly Journal of
Economics, v109, 1994, 659-684.
Athreya, K.B., 2002, “Welfare Implications of the Bankruptcy Reform Act of 1999”, Journal
of Monetary Economics, v 49, 2002, 1567-1595
Athreya, K.B., and Simpson, N., ”Personal Bankruptcy or Public Insurance?”, Working
Paper, March 2004, Colgate University.
Baird, D. G. “Elements of Bankruptcy”, 3rd Ed. , Foundation Press, 2001.
Bankruptcy Reform Act of 1998: Hearing Before the Subcommittee on Commercial and
Administrative Law of the Committee on the Judiciary, House of Representatives, OneHundred and Fifth Congress, second session.June 1998. Washington D.C., U.S. G.P.O.
Congressional Sales O¢ce, 1999. Superintendent of Documents.
Berkowitz, J., and Richard Hynes “Bankruptcy Exemptions and the Market for Mortgage
Loans”, Journal of Law and Economics, v42, 1999
Bermant, Gordon, and Flynn, Ed, “Incomes, Debts, and Repayment Capacities of Recently
Discharged Chapter 7 Debtors” Executive O¢ce for United States Trustees, January, 1999
Castaneda, A, J. Diaz-Jimenez, J.V. Rios-Rull, “Accounting for Earnings and Wealth Inequality”, Journal of Political Economy, v111, 2003, 818-857.
Congressional Budget O¢ce (CBO), “Personal Bankruptcy: A Literature Review”, CBO
paper, September 2000.Second and D streets, Washington D.C., S.W. 20515.
Chatterjee, S., D. Corbae, M. Nakajima, J. V. Rios-Rull, “A Quantitative Theory of Unse
cured Consumer Credit with Risk of Default”, 2002, Working Paper 02-6, Federal Reserve
Bank of Philadelphia.
Domowitz, I. and Sartain, R.L., “Determinants of the Consumer Bankruptcy Decision”,
Journal of Finance, v54, 1999, 403-420.
Dubey, Pradeep, Geanakoplos, J, and Shubik M, “Default in a General Equilibrium Model
with Incomplete Markets”, 2000, Cowles Foundation Discussion Paper.
Edelberg, W., “Testing for Adverse Selection and Moral Hazard in Consumer Loan Markets”,
2003, Finance and Economics Discussion Series, Federal Reserve Board.
Evans, D. and Schmalensee, R. Paying With Plastic: The Digital Revolution in Buying and
Borrowing, 1999, MIT Press, Cambridge, MA.
Floden, M., and Linde, J.,“Idiosyncratic Risk in the United States and Sweden: Is There a
Role for Government Insurance?”, Review of Economic Dynamics, v4, 2001.
Gropp, Reint, Scholz, J. K., and White M.J. “Personal Bankruptcy and Credit Supply and
Demand”, Quarterly Journal of Economics, v112, 1997, 217-251.
21
Gross, David B. and Nicholas Souleles: “Explaining the Increase in Bankruptcy and Delin
quency: Stigma vs. Risk Composition.” Wharton Financial Institutions Center, 98-28-B.
University of Pennsylvania.
Hubbard, G., Skinner, J., and Zeldes, S., ”The Importance of Precautionary Motives for Ex
plaining Individual and Aggregate Saving”, Carnegie-Rochester Conference Series on Public
Policy, v40, 1994, 59-126.
Huggett M., “The Risk-Free Rate in Heterogeneous-Agent Incomplete-Insurance Economies”,
Journal of Economic Dynamics and Control, v17, 1993, 953-969.
Jackson, T. H. “The Logic and Limits of Bankruptcy Law” 2001, Beard Books.
Jones, Edith, and James I. Shepard, National Bankruptcy Review Commission, “Recom
mendations for Reform of Consumer Bankruptcy Law by Four Dissenting Commissioners”,
1997, http://nbrc.gov/report/24commvi06.html
Lehnert, A,. and Maki, D. “The Great American Debtor: A Model of Household Consump
tion, Portfolio Choice, and Bankruptcy”, 2001, mimeo, Federal Reserve Board..
Li, W., and Sarte, P.-D. G. “Macroeconomic Implications of U.S. Consumer Bankruptcy
Choice: Chapter 7 or Chapter 13?, 2002, Working Paper 02-1, Federal Reserve Bank of
Richmond.
Livshits, Igor, MacGee, James, and Tertilt, Michele, “Consumer Bankruptcy: A Fresh
Start”, 2003, Working Paper 617, Federal Reserve Bank of Minneapolis.
Pavan, M., “Consumer Durables and Risky Borrowing: the E¤ects of Bankruptcy Protec
tion”, 2003, mimeo, Boston College.
Repetto, A., “Personal Bankruptcies and Individual Wealth Accumulation”, 1998, mimeo,
MIT
Storesletten K., Telmer, C., and Yaron, A., ”Asset Pricing with Idiosyncratic Risk and
Overlapping Generations”, 2000, CPER Working Paper.
Sullivan, T, Warren, E, and Westbrook, J. As We Forgive Our Debtors: Bankruptcy and
Consumer Credit in America, 1989, Oxford University Press, New York.
______, The Fragile Middle Class, Americans in Debt, 2000, Yale University Press, New
Haven.
“Memorandum Re: Expanded Proposal on Uniform Exemptions”, 1997,
http://www.abiworld.org/legis/review/proposals/97exempt.html
WEFA Group, The Financial Costs of Personal Bankruptcy (Feb.98) GAO B-279802 GAO/GGD98-116R.
Zame, William, R.“E¢ciency and the Role of Default when Security Markets are Incom
plete”, American Economic Review, v83, 1993, 1142-1164.
22
Zha, Tao, “Bankruptcy Law, Capital Allocation, and Aggregate E¤ects: A Dynamic Het
erogenous Agent Model with Incomplete Markets”, Annals of Economics and Finance, v2,
2001.
23
Table 1: Parameters
Parameter
®
¯
¿ s; ¿ u
¸
e
jas j
Benchmark Values
3
0.96
0.034,0.064
0.5208
0.25 ($15,000)
2.00 ($120,000)
Table 2: Benchmark Model
Bankruptcy
Max. Uns. Avg. Uns.
Avg.
Exemption
Smoothing
Welfare
Avg. Equity Savings
Rate
Debt
Debt
0
0.79317
0.66%
-$11.40 -0.19091 -0.016356
1.2437
0.21115
0.05
0.79219
0.71%
-$8.66 -0.20909 -0.016317
1.24249
0.21054
0.15
0.79164
0.73%
-$8.81 -0.20909 -0.016378
1.24261
0.20981
0.25 (Pop. Weighted Avg.)
0.79121
0.80%
$0.00 -0.22727 -0.014591
1.24281
0.21216
0.57
0.78843
0.94%
$5.70 -0.26364 -0.01486
1.24993
0.2136
1.14 (Interstate Average)
0.78624
1.06%
$12.30
-0.3 -0.016537
1.25288
0.21137
1.5
0.78502
1.16%
$18.75 -0.35455 -0.017272
1.25858
0.21164
2
0.78376
1.24%
$28.24 -0.37273 -0.017555
1.26322
0.21189
Table 3: Lower Risk Aversion
Bankruptcy
Max. Uns. Avg. Uns.
Avg.
Exemption
Smoothing
Rate
Welfare
Debt
Debt
Avg. Equity Savings
0
0.81236
0.78%
-$9.93 -0.26364 -0.02479
1.09529
0.12578
0.05
0.81218
0.86%
-$6.93 -0.28182 -0.024747
1.09476
0.12547
0.15
0.8146
0.96%
$1.00
-0.3 -0.023374
1.09756
0.12691
0.25 (Pop. Weighted Avg.)
0.81246
0.72%
$0.00
-0.3 -0.02375
1.09698
0.12652
0.57
0.80993
0.98%
$19.65 -0.35455 -0.022511
1.10738
0.12881
1.14 (Interstate Average)
0.80638
1.10%
$26.35 -0.39091 -0.022972
1.11296
0.12889
1.5
0.80636
1.10%
$26.58 -0.40909 -0.023075
1.1131
0.1288
2
0.80631
1.11%
$25.43 -0.40909 -0.023082
1.11285
0.12878
Table 4: Exemptions by State and Population
State
Population
Cumulative
Percent of Total Percent of
Population
Population
Homestead exemption
limit
Maryland
5,508,909
1.89%
1.89%
0
New Jersey
8,638,396
2.97%
4.86%
0
Pennsylvania
12,365,455
4.25%
9.12%
0
Rhode Island
1,076,164
0.37%
9.49%
0
Massachusetts
6,433,422
2.21%
11.70%
$3,500
Michigan
10,079,985
3.47%
15.17%
$3,500
Alabama
4,500,752
1.55%
16.71%
$5,000
Georgia
8,684,715
2.99%
19.70%
$5,000
Kentucky
4,117,827
1.42%
21.12%
$5,000
11,435,798
3.93%
25.05%
$5,000
South Carolina
4,147,152
1.43%
26.47%
$5,000
Tennessee
5,841,748
2.01%
28.48%
$5,000
Virginia
7,386,330
2.54%
31.02%
$5,000
Illinois
12,653,544
4.35%
35.37%
$7,500
Indiana
6,195,643
2.13%
37.50%
$7,500
Missouri
5,704,484
1.96%
39.47%
$8,000
Utah
2,351,467
0.81%
40.27%
$8,000
19,190,115
6.60%
46.87%
$10,000
8,407,248
2.89%
49.76%
$10,000
501,242
0.17%
49.94%
$10,000
Maine
1,305,728
0.45%
50.39%
$12,500
Nebraska
1,739,291
0.60%
50.98%
$12,500
Louisiana
4,496,334
1.55%
52.53%
$15,000
Oregon
3,559,596
1.22%
53.75%
$25,000
West Virginia
1,810,354
0.62%
54.38%
$25,000
Colorado
4,550,688
1.56%
55.94%
$30,000
Hawaii
1,257,608
0.43%
56.37%
$30,000
New Mexico
1,874,614
0.64%
57.02%
$30,000
Washington
6,131,445
2.11%
59.13%
$30,000
Wisconsin
5,472,299
1.88%
61.01%
$40,000
California
35,484,453
12.20%
73.21%
$50,000
Idaho
1,366,332
0.47%
73.68%
$50,000
New Hampshire
1,287,687
0.44%
74.12%
$50,000
648,818
0.22%
74.35%
$54,000
Connecticut
3,483,372
1.20%
75.54%
$75,000
Mississippi
2,881,281
0.99%
76.53%
$75,000
Vermont
619,107
0.21%
76.75%
$75,000
North Dakota
633,837
0.22%
76.97%
$80,000
5,580,811
1.92%
78.88%
$100,000
Montana
917,621
0.32%
79.20%
$100,000
Arkansas
2,725,714
0.94%
80.14%
$125,000
Delaware
817,491
0.28%
80.42%
$125,000
17,019,068
5.85%
86.27%
$125,000
Iowa
2,944,062
1.01%
87.28%
$125,000
Kansas
2,723,507
0.94%
88.22%
$125,000
Oklahoma
3,511,532
1.21%
89.43%
$125,000
764,309
0.26%
89.69%
$125,000
22,118,509
7.61%
97.30%
$125,000
Nevada
2,241,154
0.77%
98.07%
$125,000
Minnesota
5,059,375
1.74%
99.81%
$200,000
Ohio
New York
North Carolina
Wyoming
Alaska
Arizona
Florida
South Dakota
Texas
Note: Unlimited Exemptions are set to $125,000, and Washington D.C. (0.17% of pop.) has no welldefined Homestead exemption, but has other types of exemptions.
@FedFRASER