:
;;•:;
Jul
>':
>i
"
mil
imwk
.
''/!{''-'-
.
...
iirfili
HB31
.M415
i
Massachusetts Institute of Technology
Department of Economics
Working Paper Series
INTERNATIONAL AND DOMESTIC
COLLATERAL CONSTRAINTS IN A MODEL OF
EMERGING MARKET CRISES
Ricardo
J.
Caballero,
MIT Dept of Economics
Arvind Krishnamurthy, Northwestern University
Working Paper 00-21
September 2000
Room
E52-251
50 Memorial Drive
Cambridge, MA 02142
This paper can be
downloaded without charge from the
Network Paper Collection at
Social Science Research
http://papers.ssiTi.com/paper.taf7abstract _ id=244
1
44
DEWEY
MASSACHUSETTS INSTITUTE
I6Y
OCT
3
2000
L
LIBRARIES
Massachusetts Institute of Technology
Department of Economics
Working Paper Series
INTERNATIONAL AND DOMESTIC
COLLATERAL CONSTRAINTS IN A MODEL OF
EMERGING MARKET CRISES
Ricardo
Caballero,
J.
MIT Dept
of Economics
Arvind Krishnamurthy, Northwestern University
Working Paper 00-21
September 2000
Room E52-251
50 Memorial Drive
Cambridge, MA 02142
downloaded without charge from the
Social Science Research Network Paper Collection at
http //papers ssrn com/paper taf?abstract_id=2441 44
This paper can be
:
.
.
.
Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium
Member
Libraries
http://www.archive.org/details/internationaldomOOcaba
International and Domestic Collateral Constraints
a Model of Emerging Market Crises
in
Ricardo
J.
Arvind Krishnamurthy*
Caballero
September
12,
2000
Abstract
We
build a
constraints.
model of emerging markets
Firms
in
other.
We
financial
collateral
real variables.
A
binding international constraint
and
to a sharp rise in interest rates
fire sales
can lead to wasted international
can interact in important ways. The
assets causes
banks to
fail in their
economy with
affect
aggregate leads
of domestic assets, while limited domestic
collateral.
first is
in the
These two
disintermediation: a
collateral constraints
fire sale
of domestic
economy
function of reallocating resources across the
leading to wasted international collateral.
that firms in an
in-
also have limited borrowing capacity with respect to each
study how the presence of and changes in these collateral constraints
and
collateral
which features two types of
a domestic economy have limited borrowing capacity from
They
ternational investors.
crises
The second
is
a dynamic
effect.
We
show
limited domestic collateral and a binding international
collateral constraint will not adequately precaution against adverse shocks, increasing
the severity of these shocks.
on the
literature
Our approach
is
distinctive in that, while
role of financial constraints in
within either of these collateral deficiencies,
macroeconomics draws
we argue that
their static
much
of the
their insights
and dynamic
interactions have important consequences for emerging markets' performance.
Keywords:
Capital flows,
straints, contractual
eral,
fire sales,
microeconomic and aggregate financial con-
and corporate governance problems, balance
sheets,
wasted
collat-
domestic and foreign spreads, excessive leverage, collateral under provision,
real
depreciation, banks.
"Respectively:
MIT
and NBER; Northwestern University.
Craig Burnside, V.V. Chari,
Zingales,
Bhagwan Chowdhry.
We
thank Daron Acemoglu, David Backus.
Oliver Hart, Bengt Holmstrom, Andres Velasco, Luigi
two anonymous referees and seminar participants at the Fed (Board of Governors), Harvard.
IMF, MIT, Northwestern, UCLA,
Finance (Buenos Aires)
for their
meetings, WFA meetings, the CEPR Summer Symposium
LACEA/UTDT Summer Camp in International Economics and
EFG-NBER
on Financial Markets (Gerzensee), and the
comments and
suggestions. Ricardo Caballero thanks the
support, and the
AEI and Research Department
their hospitality
and support. E-mails: caball@mit.edu, a-krishnamurthy@nwu.edu. This
revised version of "Emerging Market Crises:
An
at the
IMF, where part
of this
Assets Market Perspective,
"
NSF
for financial
work was accomplished,
(1999).
is
for
an extensively
1
Introduction
The
recent episodes of emerging market crises have increased awareness, both in the aca-
demic and policy
formance. In
flows
circles,
many
that financial factors play a central role in macroeconomic per-
accounts, the sudden stop and then withdrawal of international capital
was the spark that
set off these crises.
and exchange rates were
asset prices
"well
The IMF
asserts that the ensuing declines in
beyond what was
assessment of economic fundamentals, even in the light of the
Outlook,
May
1998, p.4.) Falling asset prices
while widespread bankruptcies
among
justified
by any reasonable
re-
(IMF World Economic
crisis."
compromised weak domestic banking systems,
over-leveraged firms
all
contributed to significantly
aggravating the downturn.
While often blurred, there are essentially two
models
in the
macroeconomics
veloped economies" literature
Holmstrom and Tirole
literature.
(e.g.
In the
distinct types of financial constraints
first
group, best represented in the "de-
Bernanke and Gertler 1989, Kiyotaki and Moore 1997,
1997), widespread microeconomic financial constraints either
among
firms or banks limit a country's ability to reallocate resources, exacerbating the amplitude
and
costs of recessions.
In the second group, as emphasized in the "emerging economies"
literature, a constraint
on the aggregate borrowing of a country binds and leads to real
dislocations.
for explicit
model of
Examples include Bulow and Rogoff (1987) and Atkeson and Rios-Rull (1996)
models of this international
Calvo (1998),
for
an implicit
it.
The purpose
some
collateral constraint, or
of this paper
is
to construct a simple equilibrium
of the stylized facts surrounding emerging markets' crises,
on the respective
roles of these
model to account
and thereby shed
for
light
two types of financial constraints. The analysis allows us
to delineate the distinct roles of these financial constraints, as well as uncover important
interaction effects. Firms in our
the
amount
model have limited international
collateral that determines
we imagine
of financing that foreign investors extend to these firms. Concretely,
that international collateral consists of export sector revenues that international investors
domestic collateral that determines
We
assume that firms within an economy have
how much
financing they can obtain from each other.
think of real estate as a prime example of domestic collateral.
A
in
We
also
can seize in the event of loan default.
tightening of the international constraint at the country level can generate a large rise
domestic interest rates and a
in real activity.
to finance
all
The binding aggregate
investment needs. The
these resources.
collateral.
fire sale
An
of domestic assets, and a corresponding contraction
constraint
means the economy runs out
rise in interest rates reflects
alternative channel behind a real contraction
Even when the economy has
the high
is
of resources
shadow value of
a shortage of domestic
sufficient resources in the aggregate,
if
firms in
need of funds have insufficient domestic collateral they
These two
resources.
sector
is
collateral constraints
be able to access these
will not
can interact via a credit crunch.
If
the banking
responsible for reallocating resources across the economy, and this capacity
compromised by a
is
asset prices, the tightening international financial constraint leads
fall in
to a contraction in effective domestic collateral.
The
interaction can also occur in the reverse direction.
A
typical
example
is
the large
Asia contracted during the years preceding the
limited domestic collateral
and
amount
crises.
them with too
little
of foreign liabilities that firms in
We show that
the interaction between
the international collateral constraint, in a
causes firms to undervalue international collateral. That
that leave
puzzling aspect of recent
taken by the private sector that leave the economy more vulnerable to
crises are actions
adverse shocks.
A
is,
dynamic
setting,
firms systematically take actions
international collateral during crises, exacerbating the effects
of adverse shocks.
The economics behind
this
under-insurance result arises from the observation that lim-
wedge between the
ited domestic collateral creates a
domestic firm and the external return that
domestic).
In a crisis the
economy
is
requires an ex-ante decision to retain
tracting less foreign debt ex-ante.
is
internal return
on investment
promised to an outside investor
short of international collateral.
some
However
if
(i.e.
for a
another
Abating the
crisis
international collateral - for example by con-
there
is
a wedge between internal and external
returns on investment, the private valuation of this international collateral will be less than
its social value.
The
usefulness of our analysis
two-fold. First, while
is
of the financial factors illustrated by the model,
framework
in
which to
clarify these
among
many
we do provide a
fairly
some
simple and unified
in
the analysis of ex-ante precautionary
the private sector, and to guide policy actions.
Relation to the literature.
arise in
are not unique in noting
mechanisms. Second, the model uncovers the dynamic
undervaluation result, which should be helpful
questions - both
we
As we mentioned above, microeconomic borrowing constraints
parts of the literature on financial constraints in macroeconomics, and aggre-
gate collateral constraints arise most explictly in the sovereign debt literature. Since in the
latter literature default
and
its
costs are mostly
modeled
as a country-wide
phenomenon,
the question immediately arises on whether the domestic private sector internalizes the
fect of their actions
ef-
on the likelihood of these events. Bardhan (1967), Harberger (1985)
and Aizenman (1989) advocated taxing capital flows on the grounds that individual agents
do not pay
for
the increase in a country's average cost of capital brought about by their
ternational indebtedness. While our result on collateral undervaluation
is
in-
related to theirs,
our externality stems from imperfections in the domestic financial system rather than from
the country's
monopsony power
The modeling approach we
in
international financial markets.
follow owes to
Diamond and Dybvig's
(
1983) canonical model
and particularly to Holmstrom and Tirole (1998)'s departure from
of liquidity
in
As
the context of firms.
in that paper,
firms in our
economy
this
model
receive a production
shock that they must cope with by mortgaging collateral in order to borrow additional
funds.
The
investors.
latter
Analogously, firms in the model of Holmstrom-Tirole can borrow funds either
from other firms
the
first
can be obtained from other domestic firms or directly from international
in the corporate sector or directly
from savers.
Intermediation renders
transaction frictionless in Holmstrom-Tirole and their macroeconomic analysis
focuses on the implementation and efficiency of the second transaction. In contrast, both
transactions encounter frictions in our model.
Our underprovision
of collateral results are analogous to the free rider problems and
underprovision of liquidity studied in the banking literature (see Jacklin 1987, Bhattacharya
and Gale 1987, Allen and Gale 1997). However, the source
inefficiency in our
model
is
of the externality behind this
the endogenous result of borrowing constraints, while theirs
the exogenous divergent ex-post valuation of liquidity by consumers.
The
literature
on
is
crises
has focused on models of bank runs and their key sequential service constraint (Diamond
and Dybvig 1983. Goldfajn and Valdes 1998, and Chang and Velasco 1998). While we do
believe that
bank
failures
and runs can aggravate
model points out that the basic features of
agent or institution
is
committed to supply
we emphasize boosts and
crises
as
crises,
and
liquidity.
we
discuss in section
their prelude arise even
Moreover, the
fire
sale
5,
our
when no
mechanism
feeds from the likelihood of banking crises themselves.
Finally
our asset pricing implications arise from market segmentation. Allen and Gale (1994) and
Merton (1987) have noted the impact of segmentation on
Outline.
asset prices.
Section 2 follows with a description of the model and the different regions
that arise from the existence of two types of financial constraints. Section 3 focuses on the
"wasted
collateral'' region,
where the domestic-microeconomic constraints are binding but
not the international-aggregate constraint.
Section 4
is
the core of the paper, as
it
adds
the international financial constraint and shows the dynamic feedback from the domestic
to the international constraint.
the reverse feedback, that
is
from the international financial constraint to the domestic
constraint. Section 6 concludes,
2
A
Section 5 introduces banks and presents an example of
and
is
followed by an appendix.
Time-to-build model with limited international and do-
mestic collateral
2.1
Time.
Preferences, Technology and Financial Constraints
The world
and borrowing
lasts three periods, t
decisions.
Adverse shocks
—
0,1,2. In date
may
affect
0,
agents
make
production at date
real investment
1.
Agents must
H
At date
again making borrowing choices to cope with these shocks.
and consumption takes
realized, debts are repaid,
DateO
2,
place.
Date2
Date
—
+
Repayment
Consumption
Shocks
Borrowing
Investment
Budget Constraint:
Date
d
Investment Constraint:
returns are
all
<
f
A.'
= d
c(k)
w
f
f
Production: Time-to-build
-
(INTACT)
k
R
k
Prob = \/T
(DISTRESSED?
Need
additional investment of
k +
r
(i)
:
The economy has
Agents and heterogeneity.
two types of agents:
1
k
9
= R
k import goods
Figure
A
(0
)
k
Time Line
a single consumption good.
There are
a continuum of unit measure of domestic entrepreneurs-consumers
(henceforth, domestics) with linear preferences over date 2 consumption of a single (tradeable) good,
also value
and
(ii)
foreign financiers (henceforth, foreigners) with large
consumption of the single good
at the final date only.
storage technologies so the international gross interest rate
Production.
all
of their
Domestics have no goods at date
endowment
is
in the
is
and date
1.
time-to-build aspect. Investments are
2.
made
at dates
Creating k units of capital requires a date
The
function c(k)
is
assumed
Agents have
and
1,
frictionless
one.
As we
shall describe shortly,
form of "international collateral" goods
to produce, domestics must borrow and import goods from
endowments who
at date 2. In order
foreigners.
Production has a
and output
is
realized at date
investment of c(k) units of imported goods.
to be strictly increasing, convex
and
positive, to guarantee
an
interior solution.
We
capture the normal churn of the economy, with
with a simple Bernoulli process. At date
of further investment
1,
its
implied domestic heterogeneity,
half of the firms
and go on to produce
Rk
(randomly selected) are spared
units of goods at date 2 ("intact firms").
.
The
rest ('''distressed firms") experience a productivity fall
reinvesting a fraction 9
<
imported good,
of k, in units of the
1
tor < R. This can be
in
offset
by
order to realize output
at date 2 of:
(r
+ 6A)k <
Rk,
where,
A = R-r.
Financing and Collateral
In order to explore the financial problems that concern us
we must
introduce borrowing constraints on domestic firms. Before doing so, however,
with a subtle
date
1
issue.
we
deal
Despite the linearity of preferences, insurance-demand arises as soon as
There are
financial constraints are introduced.
contracts to consider.
The
one
first
principle sign contracts at date
0,
two classes of insurance
essentially
with respect to idiosyncratic shocks. Firms could
is
in
contingent on the realization of types, whereby resources
were always transferred from intact to distressed firms. In reality these contracts are seldom
perfect due to asymmetric information,
its
and
in
our model we capture this imperfection in
extreme form:
Assumption
date 1
is
1
(Non-observability of Production Shock) The production shock
idiosyncratic.
The
identity of firms receiving the shock
But private information does not
is
rule out signing insurance contracts with respect to
aggregate shocks. However, introducing these contracts would
dampen
gate shocks, but would not undo any of our main qualitative results.
we only analyze uncontingent debt
We
endowment
is all
1
the effects of aggre-
Thus, for simplicity,
contracts.
assume that each domestic agent
that are international collateral -
at
private information.
e.g.,
is
endowed
at date
with some date 2 goods
the present value of export sector receivables. This
Thus we make
that can be pledged to foreigners to secure debt repayments.
the extreme assumption that none of the output from domestic production
is
acceptable
collateral to foreigners.
Denote the amount of international
states of the world at date
occurs with probability
the
I
h
Tt
1, uj
=
1
e
—
{/,/?}.
7r.
collateral for a domestic as
State
Across
all
/
Xiw.
occurs with probability
domestic agents, there
There are two
ir,
while state h
is less
collateral in
state than the h state,
<
X'
f
w<
See Caballero and Krishnamurthy (1999. 2000b)
for
\)w.
an analysis of contingency. Also, see the former for
the slightly more general case (but identical in equilibrium) where foreigners hold domestic collateral from
date
to
1
XJw
Since only
amount. The
is
international collateral, foreigners will only extend financing
effect of
and therefore
straint in the i-state
Assumption
the aggregate shock
to tighten the international
is
and international
2 (Foreign debt
Foreign lenders require international collateral
repayments are assumed
denote foreign debt contracted at dates
ci" *
do,f
and
< ^f w =
promised repayments.
1
(in state ui), respectively.
1.
These
Let do j and
Then,
minjA^u;},
—
"S.
X'iw
interest rate
is
pinned down at one by foreigners, as we
do
f.
domestic interest rate (the only equilibrium price in our model)
later the
1.
f
d\
While the international
all
uncontingent on the aggregate shock at date
to be
at date
collateral)
back
to
to this
borrowing con-
and refinancing options
curtail production
up
may
will see
rise
above
This interest rate applies to assets which are valued by domestics but not by
this value.
foreigners. This
asymmetric valuation
is
due to our assumption that foreigners cannot
seize
any of the goods from domestic production, while domestics can. The parameter A^ controls
how much
of this production
is
domestic collateral and represents the degree of development
of domestic financial markets.
Assumption 3 (Domestic debt and
A
collateral)
domestic lender can only be sure that a firm will produce X^rk units of goods at date
Any
Let.
excess production based on reinvestment at date
L^ >
date
1
1
must
all
u>,
then domestic lending at
satisfy,
<
Xd^jj
+ (XfW -
<
d0J - dij),
domestic firms are identical at date
Our assumption plays a
at date 0.
neither observable nor verifiable.
denote the domestic (gross) interest rate in state
<K4
Since
1 is
2.
0,
<
Xd
1.
the domestic debt market
role only at date 1,
when
intact firms
is
not accessed
become domestic
lenders to the distressed firms.
Finally
date
we need
to
make a
technical assumption on returns to date
1
reinvestment versus
reinvestment, in order to ensure that firms do not saturate their budget constraint
at date 0,
A>
R+r
:
>
1.
l
2d{X f w)
Our assumption on
preted in a
rise in
the
number
the tightening of the international collateral constraint in the /-state can be inter-
of ways.
premium that
source of the shock.
This
is
a
terms of trade shock,
a rise in the international interest rate, or a
foreign investors require for lending to emerging markets.
We
are agnostic on the
Optimization and Equilibrium
2.2
The problem can be
Date
1
solved by backward induction.
Date
problem.
and foreign debt of
of k
decisions result in firms arriving at date
At date
do./-
a firm that escapes the shock
is
The balance
(I).
w
of rk units of domestic collateral and
of d
a firm that receives a shock
1,
intact
1
with installed capital
is
distressed (S), while
sheet of a domestic firm has assets
units of international collateral, and foreign debt
./.
Suppose that date
choices of
(A;,
distressed firm in raising funds at date
units
is
at date
1
have been made.
do./)
to alleviate
Consider the problem of a
production shock. Saving production
its
an investment that requires one unit of imported goods, and returns A units of goods
2.
A
Since the return on this investment of
always exceeds the international interest
rate of one, the firm will turn to the international debt market for financing.
maximum, Xiw —
do./,
from
this market.
1
in the
Xjiv
still
reinvestment.
its
As long
economy and pledge
as
A>
domestic debt market. The
L" the
can
raise at
then the firm will have to turn
have excess international
collateral, a
domestic collateral and import
is
able to aggregate the
this to foreigners to raise resources for
firm will borrow as
maximum amount
that can be raised, after exhausting foreign resources,
We
— do,/
this "credit chain," the distressed firm
international collateral of the
date
>
borrow from the intact firm against
Through
additional goods.
k
Since intact firms
to other sources for finance.
distressed firm can
If,
It
much
as
it
can (or needs)
of funds (in units of import goods)
is
shall refer to this inequality as the domestic collateral constraint.
We
can then write the problem of a distressed firm
(PI)
V?
=mMjj^
\» w
s.t.
(i)
dfj
f
{iii)
(ro)
than
liabilities) as
reflects that
date
(z)
1
An
in
and
(r
+ Ae)k-do,f-dfj-L»d» d
+ d ,/ + d^4 < Xfw + ±f&
df f + d0>/ < XJw
u
9 'k = df + 5?
mf
0" < 1.
(ii)
Constraints
+
as,
are balance sheet constraints (net marketable assets greater
(ii)
perceived by domestic and foreign lenders, respectively. Constraint
new investment must be
fully
taking on debts of df ^ and d^
intact firm at date
1
.
rf
(iii)
paid with the resources received by the firm at
Constraint (iv)
has only one decision:
is
purely technological.
how much
finance will
the distressed firm. Suppose that the firm borrows import goods against
its
it
extend to
international
collateral
and lends
x\_ d of these
goods
domestic market at the interest rate of
in the
Lu
.
Then,
=max^ d
V?
(P2)
X?kt
d
s.t.
Date
At date
problem.
in either the
h or the
I
1
can expect to find
itself as
aggregate state of the world. Thus
is,
(P3)
max
fc
.
do/
Z^h.n^U^ + Kn
s.t.
d,Qf
< XfW
c(k)
=
where i^ represents the probability that state
Market clearing
Equilibrium.
+ ^f<XfW
.f
a firm looking forward to date
and
either distressed or intact,
the decision at date
0,
+ Rk + L»a% d - x% d
u
d j.
occurs.
the domestic debt market at date
in
1
(capital letters
denote aggregate quantities) requires that the aggregate amount of domestic debt taken on
by distressed firms
is
fully
funded by intact firms:
u \,d
n
a l,a
Therefore, market clearing,
Dt d = Xl d
(2)
,
determines the gross interest rate If.
An
u}
(9
equilibrium of this
i
,d% ,d% d ,x ?id )
u>s V< h
f
},
economy
and date 1 decisions, (fc,c?o./) and
w
wG
{'.'»>.
respectively, and prices (L )
Decisions are solutions to
consists of date
the firms' problems (PI), (P2), and (P3) given prices. At these prices, the market clearing
condition (2) holds.
2.3
Regions
With
the basic setup behind us,
of the linearity in preferences
we turn
Let us begin with the case that
1,
as stated in (PI). Since
rates, the firm will
investors.
1.
Because
and returns to reinvestment, the decision problems end up
yielding corner solutions, and equilibrium
at date
to characterizing equilibrium at date
into one of four regions.
A > L > 1. Consider the problem of a distressed firm
A exceeds both the international and domestic interest
borrow as much as
The maximum amount
falls
w
it
can (or needs) from the domestic and international
of funds that the firm can raise
\ fVJ-d0t f
+
—.
is,
Since
L? >
1,
—w from foreign investors and lend
intact firms will import goods of Xj
Then equilibrium
to the distressed firms.
these
requires that,
Solving, yields
T"
This
is
the
first case.
excess collateral
Thus
is
In this scenario,
it is
f-i\
-d
XfW
(3)
/
easy to see that since
transferred to the distressed firms, there
total reinvestment
is
all
of the intact firms
full collateral
aggregation.
is,
ff^k
The second
Xdrk
-
~
=
- d 0J )
u
2{\ jw
case follow closely from the above expressions. Note that
possible that this equation requires that
ff^
>
it is
algebraically
Economically this means that there
1.
is
sufficient collateral for all distressed production units to be restructured, and therefore
w
firms will choose 6^ = 1. Since there is excess supply in the domestic debt market, L = 1
(equal to the international interest rate).
The
less
difference between these last
than
its
international collateral.
two cases
We
that in the second, the
is
economy
reinvests
can define the international collateral constraint
as,
6
If
k<2{Xfw-d
(4)
.f).
the constraint binds the economy must be aggregating
and therefore
The
is,
UJ
Z>'
>
final case is
suppose that 9"
1.
When
when L^
<
1,
all
the constraint does not bind,
is
equal to one, because A^
is
it
of
its
international collateral,
must be that
low and actually
Lu =
1.
(1) binds.
That
but
A^rfc
< XjW —
do j.
In this case domestic distressed firms lack sufficient collateral to compensate intact firms
for their international collateral
constrained
demand
and the domestic
the international spread,
s*i,
is
let
us define
= L W -1.
us define the domestic spread, s^ as,
s%
This
Before describing the figure,
as
sj
let
Collateral
depresses the equilibrium interest rate.
Figure 2 illustrates each of these scenarios.
Likewise,
interest rate falls to one.
= A - V.
the difference between the marginal product of investment for a distressed firm and
the interest rate that
it
pays to
its
lenders.
It is
the wedge between internal and external
returns that typically arises in models with credit constraints.
1
i
1
L
\
L= A
\
L=
\
>
\
A
\
\
\
\
N
V
L=
L=
1
1
'
J
=
Panel (a)
1
Regions
:
and
I
Panel (b)
II
:
Regions
III
and IV
Figure 2
The supply curve
It is
both panels
in
is
the supply of international collateral in the economy.
horizontal at the international interest of
and then
collateral constraint binds,
while that on the right
is
firm.
X }. The demand curve
Demand
constraint begins to bind.
for
funds
upto the point that the international
1
turns vertical.
it
1
First, the horizontal part of the curve traces
a distressed
L —
falls
The amount
for
The panel on
the
left
represents A^,
funds can be in one of two situations.
out the marginal product of reinvestment for
below
this curve
when the domestic
collateral
of domestic collateral for a distressed firm
is,
A d rfc
Thus
as
U°
starts to
fall
the collateral
loosens. This traces the diagonal
The demand curve
in
worth more and the
collateral constraint gradually
downwards.
both panels
more constrained demand curve
is
is
illustrates a case of higher
and lower A^. The lower and
a lower A^, representing less
domestic financial market
development.
Proposition
Proposition
1
1
summarizes the main
(Regions).
results
up
The economy can
to this point:
be in
one of four regions
pending on which of the two collateral constraints, (1) and
10
(4). bind.
at date
1.
de-
•
Region I occurs when both the international collateral constraint (4) and the domestic
collateral constraint (1) are slack. In this case, both international
and
are zero
•
there
economy
downsized. s%
is
>
0,sf
=
<
0,6"
IV
=
and domestic spreads
1.
slack but the domestic constraint
some
occurs
are binding.
when
some
projects being
1.
The economy aggregates
aggregate resources and
Region
0,6
U
fails to aggregate all of its collateral resulting in
straint is slack. International spreads
all
jump
of
to
its
A
—
binding but the domestic con-
is
1,
while domestic spreads remain
however there
collateral,
=
1
projects are downsized, s ^
both the international constraint
sj
0,
and
>
is insufficient
0,6^
<
1.
the domestic constraint
Both, international spreads and domestic spreads are positive and
projects are downsized, s^
>
0,Sy
>
<
0,6^
some
1.
Discussion of main assumptions
2.4
We
s^
Region III occurs when the international constraint
at zero.
=
International spreads remain zero, however domestic spreads are positive
binding.
as the
•
=
Region II occurs when the international constraint
is
•
is
full project completion. s%
have assumed a
friction that prevents a
Investment at date
to his output from another domestic.
fully
up
produces rk goods at date
2,
domestic entrepreneur from borrowing
and depending on the realization of shocks and reinvestment, an additional A/c of output.
In a well functioning domestic financial
market -
if
i.e.
there were no frictions in borrowing
from another domestic - both the rk and the AA; of output could be sold to obtain funds
investment. In assuming that only Xjrk
for
is
observable and verifiable,
claims sold to another domestic to this amount.
is
not unusual.
It is
most similar to the limited
3
The modeling
(1997) assume that output
Our X d rk
is
is
akin to their collateral.
restricted
of this borrowing constraint
collateral constraint of Kiyotaki
(1997), or the limited enforcement constraint of
Moore
we have
Kehoe and Levine
(1993).
and Moore
Kiyotaki and
not verifiable, but that physical collateral (land)
Holmstrom and
Tirole (1998) develop a
is.
model with an
ex-ante unobserved effort choice where a fraction of output has to be paid as a rent to the
entrepreneur so that he exerts
it
effort.
While our model does not have ex-ante moral hazard,
shares the feature that a fraction of output
(i.e.
Afc) can never be promised to a lender.
Indeed, any model with borrowing constraints will share the feature that
future output
3
Purely from
is
a
case Aj
=
modeling standpoint,
In Caballero
1
will
fraction of
unpledgeable.
developed domestic markets. In a
collateral.
some
it
is
worth pointing out that Xd
perfect, capital
=
1
does not correspond to
market, one would expect that
and Krishnamurthy (1999) we model domestic
correspond to the perfect capital market case.
11
all
of
Rk would
collateral as \d(r
+
#A)fc.
fully
count as
In this
While the modeling of the domestic borrowing constraint
ature on credit constraints, the
and foreigners
is
received. In
in
standard
fairly
all,
crisis,
many
there
used
is
is
the
liter-
the borrowing constraint between domestics
it
can from
empirical support for this assumption. For example, Mexico,
its oil
revenues as collateral to back the liquidity package that
instances the bias
is
directly justified
it
by mandates on foreign institutional
investors (e.g. limits on real estate investments, see Blommestein, 1997).
the assumption
in
unusual, rk cannot be borrowed against from foreigners, while
domestics. First of
during the 94-95
asymmetry
is
most similar to that of the sovereign debt
literature.
4 5
Theoretically,
'
For example Eaton
and Gersovitz (1981), Bulow and Rogoff (1989) build models of sovereign debt renegotiation
around the question of what international lenders can threaten sovereign countries with
the event of a default. In this literature, debt
is
typically taken to be
occurs
when
some
is
in
limited by international collateral, which
fraction of exports. In Atkeson
and Rios-Rull (1997) a
becomes tighter
this international collateral constraint
crisis
On
for a country.
the
other hand, the sovereign debt literature typically just imposes international collateral as an
aggregate constraint.
We
individual agents in the
take a microeconomic perspective by assuming that
economy who can trade
Wasted international
3
The only one
of the regions in the previous section
We shall
I.
Constraints are assumed to arise in the
in region II, while in the next section,
Of course there are important exceptions
I
we
regions of most interest. In this section
(1997) for systematic
held by
foreigners.
where no credit constraints were binding
henceforth assume that in the h state, the economy finds
region
4
among themselves and with
it
is
collateral
is
I.
w
we
state.
shall
There are
(as
assume that the
we
I
shall
itself in
make
region
clear)
state puts the
two
economy
shall discuss region IV.
(e.g.
"too big to
fail"
utilities),
but
see, e.g..
Kang and
Stulz
Japanese evidence showing that, for small firms, those that are export oriented are
favored by foreign investors.
See, e.g.
Blommestein (1997).
assets considered highly illiquid or exposed to
by foreign institutional
investors.
exchange
for a discussion of
risk are generally
how
real estate
and other
avoided (sometimes by mandate)
Interestingly, the very few exceptions to the sovereignty principle, by
which the rating on debt issued by a country's corporate sector
is
bounded from above by that country's
government debt rating, are for companies which belong to the export sector.
°The September 1998 report on the Asian crisis by the World Bank, describes firms that borrow
foreign currency as "predominantly large exporting firms with ties to foreign companies,
better adjusted to the
crisis..."
(box
4.3,
page 62). The 1997 Industrial Survey
in
in
and they have
Thailand reflected that
of those firms that borrow in foreign currency, 88 percent export, have an average of 818 employees, a debt-
equity ratio of 3.12, a relatively high capacity utilization and optimism during the
respectively.
The same
statistics for firms that
do not borrow
in foreign
crisis.
70 and 37 percent
currency (75 percent of firms) are:
46 percent, 139 employees, 2.36, 61 percent and 19 percent. See Dollar and Hallward-Dreimeier (199S).
12
Micro-economic credit constraints
3.1
economy
In region II the
fails
constraint on a domestic firm
to aggregate
is,
d(j
Interestingly, there
eral.
is
essentially
Adding one unit
The borrowing
international collateral.
its
+ ttt d <
+ Xjw -
X d rk
d 0J
(5)
.
no difference between domestic and international
of domestic collateral to a distressed firm's balance sheet
equivalent to adding one unit of dollar collateral. Alternatively, even
collat-
would be
foreigners accepted
if
the A^rA; as collateral for international loans, the outcome would be the same. Thus, the
aggregate collateral constraint does not play any role in this economy.
To us
this case represents the situation of credit constraints that
developed economies. At the micro-economic
to borrow to meet their investment needs. There
is
is
most common
in
that have poor collateral are unable
level, firms
a wedge
(s^
>
external returns. Bernanke and Gertler (1989) and Kiyotaki and
0)
between internal and
Moore
(1997) are models
of this environment.
One
if
to
could also interpret the wasted collateral region as a "credit crunch." For example
a banking sector was responsible for evaluating the collateral of distressed firms in order
make
loans to these firms, and this banking sector was compromised so that loans were
cut back, again the economy would
up with
Sd
>
fail
to aggregate
Holmstrom and Tirole (1997)
0.
of their credit crunch
There are two
model
of
its
discuss this case.
interest rates
loans are fully collateralized, thus they are default free.
rate of return on fully collateralized lending -
is
(i.e.
A^
is
have already noted the wedge
also define a credit spread as the default
collateral. If the face value
is
focusing on fully collateralized loans
we
is
our purpose
bad
—
(or
in
one. If
-^
is
1.
we were
effect that
in the
/
is
all
the
to allow risky
the interest rate on these
This could also be interpreted
lending of s^
= A—
L.
We
could
to distressed firms with poor
set at Afc, but the loan
is
collateralized
This term also increases as Ad
isolate the effect that lending to firms
not as good) business, while ignoring the realistic
Comparing across the
same, because
return to a variant
and spreads. In our setting
premium on lending
on a defaultable loan
only by Xdi'k, then the credit spread
collateral
We shall
falls),
loans would rise to reflect the growing chance of default.
We
and we would end
Riskless interest rates - that
L, which
loans with default, as domestic collateral worsens
as rising spreads.
collateral
in section 5.
ways to think about
different
all
— but
falls.
By
with poor
less central for
poor collateral induces default premia.
/;
and
/
states,
state, despite the
lack of collateral of the investing firms
one can see that
riskless interest rates are the
marginal product of investment being high, the
means that
13
effective investment
demand
is
low.
Low
investment demand, low interest rates, and high interest rate spreads are
of a developed
recession (e.g. Japan).
problem and Efficiency
Date
3.2
economy
common symptoms
Let us now briefly characterize the efficiency aspects of the decentralized equilibrium.
Among
other things, this will help us understanding more fully the precise nature of the
pecuniary externality described in section
In both, h and
an intact firm's date 2 profits
states,
I
4.
V? = Rk + Xfw-d
In the h state, the distressed firm achieves
V* =
On
the other hand, in the
I
-
l
These
last four
project completion so that date 2 profits are,
Rk-k + \ hf w-d0J
=
l
Vs rk + (A
j.
.
state, since the collateral constraints bind,
9 k
This gives us that profits
full
are,
+ \ fiv — do,/.
l
X^rk
are,
l
l)6 k
+
X\iv
- d 0J = rk + X d rk{A -
problem
expressions can be combined in the date
max MOi/
(P4)
£ we{
,, fc}
7r
+ A(X f w l
1)
do,/).
for a firm:
w V'
V h = X f w - d 0J + Rk - fc/2
V = ^(X f iv - d0J + ^k + Xd rk^=±
l
s.t.
l
1
)
Since c(k)
is
strictly
c(k)
=
do./
< XfW
d
,/
convex and the objective
is
linear, (P4) defines a
unique optimum.
Proposition 2 (Financial market development and "Underinvestment")
A
decrease in A^ has the following effects on equilibrium:
vestment
Proof:
at date 1 falls (8 falls), (ii)
(i)
Return to the
demand curve from
left
k, firms
reduce investment in
at date
panel of figure
2,
Fixing date
decisions, ink.
and run the experiment of
shifting the
the high to the low A^. Alternatively, see the debt constraint of
See the appendix for the algebra.
with
Firms
(i)
The
idea
is
that since A^ enters only
recognize that domestic investment
therefore reduce date
investment.
14
is
V
1
,
less collateralizable in
(5).
(ii)
multiplicatively
the /-state and
The underinvestment
result
Kiyotaki and Moore (1997).
the
is
It
is
collateral constraints are imposed.
same
Bernanke and Gertler (1989)
as that in
underinvestment relative to a
The
inefficiency
is
first
or
best in which no
that distressed firms
who have
a
high marginal product investment (A) do not have the resources to take advantage of this
investment. Its also easy to see that welfare must be increasing in A^.
The reason we used quotations
relative to the first best, they are
these
same
firms
beyond the distressed
We
in the proposition
still
constrained
is
that while choices are inefficient
efficient.
A
central planner, subject to
no mechanism to persuade foreigners and intact
collateral constraints, also has
firms' collateral.
consider a central planner
who makes
choices to
maximize the equally weighted sum
of utilities of the domestic agents, subject to the constraints
imposed by assumptions
1
—
3.
Since domestics are ex-ante identical, this objective amounts to maximizing the utility of
any domestic firm
at date 0.
The only complication
is
imposing the collateral constraints
on the central planner.
Definition 1 (Constrained Optimality)
Let the solutions of (P4) be (k,doj), and
The competitive equilibrium
is
let
the welfare corresponding to these choices beu.
constrained Pareto optimal
if a
planner who directly makes
choices o/(£;,do./), subject to the international collateral constraint of doj
date
but lets debt markets clear competitively at date
By
1,
cannot raise welfare aboveu.
allowing markets to clear competitively at date
international collateral constraints at date
choose (k,dQj) subject to the date
1.
< Awe.
Thus the
1,
we impose the domestic and
central planner's
program
is
only to
international collateral constraint.
Proposition 3 (Constrained Optimality)
The competitive equilibrium
Proof:
By
inspection.
solutions, (k,doj), are constrained Pareto optimal.
Note
first
that prices (L
w
)
do not appear anywhere
in (P4).
Then
note that the central planner's program must be identical to (P4).
4
Aggregate Collateral Constraint and Underprovision
In the wasted collateral region, the fact that the
collateral has
market
is
that collateral
is
The only
emerging economies. Indeed
cost of having an undeveloped domestic
not aggregated properly, resulting in lower investment
and output. However, limited international
in
international
no consequence since the constraints on domestic and foreign borrowing are
substitutable with no effect on outcomes.
financial
economy has limited aggregate
collateral
crises in these
15
seems
to
be an important constraint
economies are times when interest rates
rise
rather than
fall
- a
symptom
constraint (see proposition
that our
model associates with a binding aggregate
1).
we show that the domestic
In this section
collateral
credit constraint
is
even more costly
aggregate international collateral constraint also binds. In particular,
we show
if
the
that there
is
a dynamic effect whereby agents undervalue international collateral, overborrow at date
0,
so that at date
1
they have
a constrained efficient
less international collateral (relative to
outcome). Effectively, a low A^ interacts and tightens the aggregate collateral constraint so
that the negative shock causes investment and output to
assume that the h state leaves the economy
constraints bind, while the
constraints bind.
I
is
at date
1.
= Rk + \)w -
while the distressed firm has date 2 profits
V's
state, the
premium on
I
h
\f
domestic interest rate of l)
IV where both
collateral
in region
d 0J
and distressed firms are as
in the
.
= Rk + L
l
w-d0J
rises
L >
their lending to distressed firms as
{X
l
f
.
above one. Thus intact firms earn a
1.
Profits are,
iu-d0J ).
For the distressed firm, the cost of the higher interest rates
less,
collateral
of,
= Rk-k +
l
Vl
worth
where neither of the
the intact firm profits,
1
I
economy
state leaves the
V/
In the
in region
In the h state, the profits for intact
previous section. That
is
much
too
International collateral undervaluation
4.1
We
fall
lowering reinvestment.
V;'
=
The
that their domestic collateral
profits are,
+ (A -
rk
is
+
l
l)9 k
(X
l
f
w - d 0J
).
where,
e
i
k=
^± +XfW -do,f.
l
(6)
Li
Since a firm can be distressed or intact with equal probability,
profits in the
I
we can
write the expected
state as,
V
1
=
—^k + X'fW-doj
(7)
1
+l(A-l)0 k+ -(L -l)(\ f w-d 0J
l
l
The
by date
first line
of the
RHS
in (7) reflects
l
)
the expected profits, in the
investment net of foreign debt contracted at date
16
0.
/
state, generated
In particular, this line
corresponds to the expected wealth assuming that
to
The
fail.
value of saving these production units
probability of one-half, the firm
distressed
is
units, directly earning the internal return of
half the firm
distressed production units are allowed
all
is
intact in
which case
A
and
against
its collateral.
difference
return of
L
l
between the internal return of A
for the intact firm
see this, let us construct
for a central planner.
two
is
own production
With
probability one-
international collateral and lends
its
to the distressed firm, thereby earning the domestic interest rate of
The
L
l
.
for the distressed firm,
and the external
the source of international collateral undervaluation.
different versions of
To construct the
former,
V
1
one
;
for individual firms
we simply
yields,
V
To construct the
=
1
^
substitute into (7) the expression
+ i(A + L<)(A<x,-do./) +
central planner's version, which
We know
must be equal to the
l
=
we
Substituting this expression into (7)
2{X
l
f
w-d
V
1
*
,
we
substitute instead the
that in aggregate, reinvestment
j).
find,
V * = R±J-k + A(\ f w-d
1
difference
This
total international collateral of the economy. Thus,
9 k
The
(6).
^(A-L')
we denote by
equilibrium collateral constraint on reinvestment.
To
and the other
reinvestment implied by the microeconomic domestic collateral constraint,
for
With
line.
its
borrows to save
it
borrows against
it
captured in the next
is
l
j)
between these expressions highlights the undervaluation
V-V" =
At a given equilibrium,
V
1
and
V
1
*
effect:
i(^-*^)
must be equal. But
it is
(8)
apparent that individuals and
the central planner value a marginal unit of international collateral and domestic collateral
quite differently. In particular:
• Individual firms value
domestic collateral more than the central planner.
than the central planner.
• Individual firms value international collateral less
•
The divergence
with declines
rises
6
Xd
It is
falls.
in these valuations are
in A^.
proportional to the domestic spread,
sd
,
which
6
easy to show that the term in parentheses
in (8)
does not
Starting with a local argument, take do./ (and hence
fc)
(fully) offset the
as given
change
in the
spread as
and reduce A^; then L must
fall
proportionally with X d (see (3)) so their ratio remains constant and so does the expression in parentheses.
Now.
rise
in
only
equilibrium do,/ rises so the expression in parentheses declines with a
if
the distortion increases, the
rise in
the spread
parentheses.
17
must be
fall in
Xd-
But since
larger than the decline in the
do,/ will
term
in
At the aggregate
is
level,
the central planner only values international collateral.
because international collateral
and
at date
1
is
microeconomic
that can be used to attract foreign investment,
all
the binding constraint
This
However
the inflow of foreign investment.
is
at the
both domestic collateral and international collateral can be used to
level,
secure financing. This
is
the basic tension between the individual and the central planner's
problem.
A
wedge between the
positive domestic spread heightens this tension by placing a
and private valuation
A
return of
at date
at date
On
a return of one.
L
l
.
The
unit of international collateral generates a social
while a unit of domestic collateral generates a social return of one
At the individual
1.
return of
1,
A
of collateral.
social
level, for
an intact firm, a unit of domestic collateral generates
the other hand, a unit of international collateral only generates a
difference
l
between social and private valuation of
sd
—A—L
l
results in
undervaluation of international collateral. Domestic underdevelopment prevents collateral
demanders
from transferring the
(distressed firms)
collateral providers (intact firms), since the
As a
result, at date 0, firms will
lenders at date
marginal value of collateral to the
demanders simply have
insufficient collateral.
undervalue international collateral since
they will not share in the
1,
full
full
collateral.
they are the
surplus of reinvestment.
Underdeveloped financial markets (low values of A^) result
and the systematic mis-valuation of
if
7
The
in
high domestic spreads
we have described
externality
is
not
of the traditional Bardhan-Harberger type, where individual borrowers do not internalize
the fact that the country faces an upward sloping supply of foreign loan. Indeed, in our
setup the economy does face a very steep international funds supply, for the country
rationed after some point, but the externality arises only
when domestic
financial
is
markets
are underdeveloped as well.
Date
4.2
Problem
Let us now state this result on the undervaluation of international collateral and
pendence on the underdevelopment of domestic
first
write
down the problem
central planner's
who
chooses
(fc,d ,/)
The program
'
program and
for a firm at
At the individual firm
a return of
A.
On
same way
At the individual firm
collateral, resulting in too
shall
a central planner
1.
The
the necessary counterpart of undervaluation of international
the other hand, a unit of domestic collateral
excess return of zero.
i.e.
level, for a distressed firm, a unit of international collateral still
generates output of A, providing a return of
domestic
as in the last section,
follows closely from the previous derivations.
date
is
We
de-
and then write down the
markets clear competitively at date
Overvaluation of domestic collateral
collateral.
markets more formally.
for the decentralized equilibrium,
in the
lets
financial
its
s'd
>
0.
At the
is
discounted at the interest rate of L' and
social level,
level, this mis-pricing
much domestic
domestic collateral generates an
causes firms to overvalue the Ajrfc of
collateral relative to international collateral.
18
generates
expected value to being in the high state
Vh =
Then
V
using the
1
that
we
-
X'fiv
+ Rk -
do,/
fc/2
just derived, yields the firm's program:
maxuo,
(P5)
is,
E^{i.h}^V^
V h = X f w - d0J + Rk- fc/2
y' = ^±r/c + I (A + L')(A/W' c(fc) = d
,/
l
s.t.
<
do,/
To
arrive at the central planner's
(P6)
+
*,/)
^(A - L
l
)
-^/^
program we use
EW
max fcdo/
€{Z,fc}
V
7t
instead,
" V,,
and note
also that
Vh = V
*
h
'
= A^tu - d j + Pfc - fc/2
y<* = Bp-k + A{\
fW - do, f
c(k) = d ./
V"
s.t.
1
*
l
)
<
do,/
A/tu
Proposition 4 (Constrained Optimality)
Let (fc*,d5/)
These solutions are the constrained Pareto optimal date
solutions to (P6).
fee
choices of the economy,
(a)
(b)
(k*,d,Qf) are independent of the strength of domestic financial links, Ad-
Proof:
Part
(a), as before.
Part
apparent from inspection of program (P6). Since
(b) is
none of the constraints depend on the value of
program
A^, the solution to the
will
not
either.
The second
part of this proposition
planner's choices are invariant to
function of A^
we know
moreover the distortion
is
particularly interesting because
However
A<j.
since the domestic spread
.
(fc,
will
program (P5) in
do./), are constrained
ment by perturbing
says the central
is
a decreasing
that the decentralized equilibrium will be a function of
be decreasing
the case
h
when L = 1,L >
Pareto inefficient.
these decisions to (fc',d
A<f
and
in A^.
Proposition 5 (Undervaluation of International Collateral)
tions to the
it
A
1
and
l
sd
>
Let
0.
(fc,
Then
do,/) be solu-
the decisions,
central planner can effect a Pareto improve-
*),
where d
,
<
do./
(hence
k'
<
k).
Proof: see appendix
The proof
Note that
follows closely from the logic laid out in the previous section.
the only difference between (P5) and (P6)
domestic spread,
s ff
,
equals zero, there
is
is
in the expressions for
V
1
and
V
.
When
the
no difference between the central planner's and the
decentralized solution. In this case, the market equilibrium
19
is
constrained Pareto
efficient.
However, as A^
some point
at
falls,
must.be that
it
the social and private value of collateral, resulting
>
sd
and there
The next
the inefficiency.
in
a divergence in
is
proposition
formally states the dependence of welfare on A^.
Proposition 6 (Financial market development and Welfare)
equilibrium, welfare is increasing in A^
to
In the decentralized
In the central planning solution, welfare
.
invariant
is
Arf.
Proof: see appendix
Discussion
4.3
Unlike region
in region IV,
II,
interest parity condition breaks
XtW
lending capacity of
into a
fall in
(L
little
—
1
>
and are therefore unable to arbitrage
more notation, the
exchange rate
l
in region II, while s d
III, this
>
l
sd
>
maximum
this interest
mapped
All of these are aspects of
and the international
l
is
binding, s d
=
is
may seem
enough
in
is
is
we
the case considered
/ sovereign-constraint
no externality.
counterintuitive that the pecuniary externality arises because
the
slack,
Indeed, though
0.
binding. This
Atkeson-Rios Rull (1996) and other representative-agent
models, where there
is
region corresponds to the case where domestic financial
markets are perfect while the international constraint
e.g.
collateral
the international constraint
0,
in region III while the international constraint
have not focused on region
rise
8
The
1.
could also be
rise in interest rates
when
externality arises only
Thus
constraint binds.
It
the Z-state at date
because foreigners reach their
0)
asset prices or a depreciation of the
The pecuniary
in,
rise in
emerging markets.
crises in
and
down
l
to domestic firms,
With a
rate differential.
domestic interest rates
-
crisis
l
i.e.
sd
>
and the domestic
0,
L
l
does not
interest rate does not reflect the
marginal product of investment. But lending spreads are typically higher
than
in
normal times - so that the s d implication seems borne out. Part of the reason the result
is
in crises
l
counterintuitive
is
that in our model
occurs through the price mechanism.
collateral constraint
is
surplus transfer between distressed and intact firms
all
The
distortion in this transfer
measured by the domestic spread. In
reality,
due to the domestic
some
of this limited
transfer occurs via non-price mechanisms as well - quantity rationing, default, extortion, etc.
8
The
equity value of domestic firms at date
rise in L"'
one
may
causes this to
proceed to do
long as the asset
demand
is
is
fall.
so.
1
is
assets
Our model does not
We
could introduce
also domestic collateral -
determined by domestics - the
a
minus
liabilities,
which
is
-£zr
+
tf w -
define an exchange rate, but their arc two
nominal asset at date
1
and price
this asset.
do./-
The
ways that
Then
as
which seems reasonable since we normally think that money
rise in
Alternatively, one could introduce non-tradeable
V° would
also cause a
and tradeable goods
foreigners did not accept non-tradeable goods as collateral,
exchange depreciation.
20
we would
at
fall in
the nominal exchange rate.
both dates
1
and
2.
and
as long as
arrive at a similar result on the real
L
l
As a
a catchall for both price and non-price transfers.
is
also exist, then
L
result,
if
non-price mechanisms
understate the price response, and will not correspond perfectly to
will
the interest rate.
In the wasted collateral region the only effect of low \ d
while in region IV, the only effect of low A^
would be
tional collateral than
is
point. It
IV would be blended. At
region II and region
and the undervaluation
The
result
elastic
all
both the
points,
Can the
what
to eliminate the inefficiency? If not,
We deal with the second
We
in
which case
collateral aggregation
would be present.
existence of the externality begs two questions.
among themselves
less interna-
up to XfW and completely
model a smoother supply curve,
possible to
is
with
because of a modeling abstraction.
have taken the supply curve of foreign funds to be perfectly
beyond that
1
date 2 output. However,
effects lead to lower
the sharp separation between these two effects
inelastic
poor collateral aggregation,
that firms arrive at date
is
Both
efficient.
is
is
private sector contract
optimal government policy?
question in Caballero and Krishnamurthy (2000a), and shall return
briefly to policy issues in the conclusion.
As
far as the first question, there could
be mechanisms around the
previous section, insurance contracts signed at date
But we
firms would work.
rule these out
to transfer
all
As
inefficiency.
in the
liquidity to distressed
on grounds that the type of firm at date
1
is
unobservable.
Another way to try and solve the problem
all
firms got together
the
state
/
However
would be
this
is
in
set at
L = A. Then
— cio./) on
region IV,
A
last
''fix"
interest rate
date
the price at date
incentives
it
must be that
is like
A^r/c
some
< A(X jiu
l
-ofo./).
1.
on any loan contract
their loans, while borrowers only
to agree ex-ante to transfer
- but this
to
Imagine that
at date
would be aligned with
arrangement would be budget infeasible at date
receive A(AyU>
and
and agreed that the
is
1
since lenders
The only way around
in
efficiency.
must
have A^r/c of domestic
1
in total
collateral,
this
problem
of the domestic collateral from intact to distressed firms
the insurance that
we have
mechanism worth discussing
ruled because of unobservability of types.
a credit line
is
arrangement akin to Diamond and
Dybvig (1983)'s deposit contracts. Let us denote the privately optimal amount of international collateral to be held in the
to be
,
and
that drew
its
C of international
credit line at date
would receive
while
state as
C* where we know that C* >
collateral
2
I
if
AC* >
A^r/c.
Then,
if
C=
C Now
.
AVu;
— do./, and
suppose that
collateral to the bank,
1
all
the socially efficient
firms gave A d rfc of domestic
with the arrangement that any firm
would receive C* and any firm that drew
A^rfc
> C*
it
is
amount
its line at
optimal for intact firms to wait
X^rk distressed firms would draw down C* at date
I.
till
date
date
2,
Since in region IV,
A>^>1,
choosing C*
>
C
makes the arrangement
is
21
budget
feasible,
and can improve
efficiency.
The problem with
fragile
—
the arrangement
is
that
it is
informationally intensive at date
same reason that Diamond-Dybvig
for the
is
Thus
if
is
L* which
result firms in the
Alternatively,
if
the firm
is
if
they are
L* to anyone delivering international collateral. As a
banking arrangement draw down their credit
distressed firms, which they are
is
L >
more
banking arrangement, a firm
has the incentive to opt out of banking and invest on their own. In this case,
distressed they offer a return of
First
is less
the private sector was choosing C, their incentives would be even
distorted. Second, holding fixed every other firm joining the
firm.
and very
fragile (see Jacklin (1987)).
note that the effective interest rate offered by the bank that uses C*
than L.
1
intact,
happy
it
simply
offers to lend at
The only way
to accept.
line
and lend to the rogue
the bank rate of L* to
to rule out this deviation
to enforce that firms can contract only with other firms in the banking arrangement - a
possibility that
Each
seems
unlikely.
of the above solutions rely on
some form
While
of insurance across firms.
in the
above paragraphs we have appealed to theoretical arguments to rule out such insurance,
the short answer to the question of does the private sector contract away the externality
that empirically there
which there
is
is
is
possibly partial insurance, but our model investigates the case in
none.
5
Disinter mediation
The
results of the last
two sections can be summarized
[loosely] as, in
the wasted collateral
region outcomes depend separately on domestic and international collateral;
when
the ag-
gregate collateral constraint binds as well, both forms of collateral interact, so that outcomes
depend on international
this section
we
collateral,
which
effectively a function of
is
domestic collateral. In
discuss the converse case in which the shock to A/ feeds into the available
domestic collateral
—
i.e.
the latter
is
effectively
a function of international
collateral.
Collateral Aggregation
5.1
Distressed firms access the international collateral of intact firms by borrowing from these
firms against their domestic collateral.
is
that this collateral aggregation
is
The
done
implicit assumption in the preceding analysis
in
a perfectly functioning public debt market.
Distressed firms issue debt claims in the domestic market secured by domestic collateral.
Intact firms issue debt claims to foreign investors secured by international collateral.
They
then use the proceeds from this issue to purchase the debt claims of the distressed firms.
In emerging markets, liquid domestic debt markets are the exception rather than the
rule.
Invariably, the financial
system
is
bank based -
for
good reasons: lack of investor
protection and the necessity of monitoring to alleviate asymmetric information places banks
at the center of the financial system.
Banks
issue claims to foreign investors against balance
22
sheets that are partially backed by international collateral - both of the distressed and the
intact firms.
These funds are then channeled to the distressed
firms. Implicitly, the intact
firm extends credit to the distressed firms against their domestic collateral.
But
immutable debt markets of the previous
unlike the
experience sharp
falls,
prices
and
real activity.
real activity
and
is
This endogenous disintermediation process and
asset prices
the early 80s.
crisis,
crisis.
In this section
system.
we sketch
There
is
a
Intact firms
at its worst
Mexico during the tequila
model whereby
all
no domestic debt market.
from banks. To attract these funds, banks
rate.
And
crisis,
it is
become the
our model -
in
plays
reminiscent
and Chile during
depositors.
aggregation at date
1
is
done via the banking
Distressed firms in need of funds take loans
offer deposits at
They mortgage
a market determined interest
their international collateral to
By
these transactions, the banking system serves to aggregate collateral. This
banks
it
Financial Bottlenecks
foreign investors and then deposit these funds in the banking system.
of
in asset
feedback into
its
9
The Banking System: Amplification and
5.2
distress, the
conventional wisdom well. Arguably, some form of
fits
of Indonesia in the recent Asian
crisis of
asset prices
weakened, triggering further declines
a role in virtually every emerging market's external
the debt
As banks enter
banks themselves run into trouble.
economy's ability to aggregate collateral
when
sections,
we do not provide an
explicit
microfoundation
for
intermediating
is
the only role
the existence of
intermediation.
Consider a representative bank
Balance Sheets.
measure).
We
shall
assume that entering date
in
1,
a competitive banking sector (of unit
the bank has assets consisting of loans
to firms backed by domestic collateral, and no liabilities (for simplicity).
10
market interest rate on loans of L", the value of capital (entering date
in units of dollars)
in the
bank
1,
Then
given the
is,
w
*=£•
The
net worth (assets minus
to reflect the additional
liabilities)
bank loan
liability:
X1 W
°See Gelos and Werner (1999)
liberalization period.
Among
for
a
of distressed and intact firms need to be modified
rk — A
+ ~jp~ ~
d °J
study of lending practices by Mexican banks during the post-
other things, they document the significant role played by banks in lending to
manufacturing firms without access to foreign funds, and their reliance on domestic
collateral.
They argue
that this mix contributed significantly to the collapse of the banking system during the 94-95 crisis.
'"The purpose of this section is to highlight, the potential date 1 bottleneck brought about by the deterioration in banks' balance sheets.
There arc several interesting date
issues
discussion but which would lengthen this section substantially. For example, one
sign contingent contracts with borrowers
and depositors
23
which merit an extensive
may
ask
why banks do
not
Without further assumptions, the preceding case of banks
Capital Constraints.
to that of public debt markets with Ad equal to one -
now
the banking system
i.e.
is
a
is
identical
We
veil.
introduce capital constraints that limit the capacity of the banking system to aggregate
collateral.
At date
1,
a bank takes on debt from intact firms of d\ d and makes loans of x\ d to the
distressed firms -
on
and
loans,
in units of face value of date 2 goods. Let
all
Ld
both of these are made at the single market interest
goods at date
2).
We
is (x'j
If
,
—
d
assume that capital constraints
represent the interest rate
no other constraint is present,
u and the expected date
rate of L
represent the deposits' interest rates.
2 repayment to the bank from this operation
Lw
d\ d )L? (in units of face value of
restrict the size of this operation. In
particular,
#>a(A + 4 d (u;)),
< a <
where
value of x\ d
there
is
1 is
n
(w).
(10)
the capital that banks must hold against making loans with date 2 face
constraint binds,
If this
Lx
L but
equal to
is still
Lrj falls to one since
an excess supply of deposits but a scarcity of loanable funds.
Taking as given the market interest rate (on loans) of L"', the problem of a bank at date
1
is,
W
<%>a(A
# > a(A + xx\bldd (u))
(i)
dbQf
4
<f
+
[L"
(L-
objective of the
bank
reflects profits
w
of L and taking deposits of d\
d
on banks. Constraint
must be
make
all
from making loans oi x\
(ii) reflects
relegate
it
Banks
to the appendix.
while in equilibrium L u
will
behave exactly as
binding, the problem
is
(L£ - l)d blA (u)
—
LD
the date
d
(u>)
at the interest rate
Constraint
.
1
{%) is
simply the
resource constraint: funds
loans.
Since the formal analysis of equilibrium
we
W
(w) at the interest rate of
capital constraint
raised to
-
>0
(a)
The
l)x>» -
-
A-
^
s.t.
MA M
nrnx
max
<dMi<>)
\
is
very similar to that of the previous sections,
will lend as
when demand
much
for funds
in the previous sections.
is
as they can
less
whenever L u
than loanable funds.
In fact, as long as constraint
{%)
entirely analogous to that of the previous section with Ad
=
>
1,
Firms
is
1.
not
To
keep matters simple, we shall proceed under the assumption that the domestic collateral
constraint does not bind - this
11
is
the case that
is
derived in the appendix, and
is
the case
In practice, capital adequacy requirements are based on both assets as well as liabilities. For example.
BIS standards assign capital requirements to
sorted
(i.e.
common
stock, preferred stock)
different assets held
and weighted
bank and hence ensure that the capital requirement
is
to
by banks. The
24
the bank are
met. Capital requirements can be justified from
principles on the basis of moral hazard within the banking sector.
example.
liabilities of
determine the amount of capital held by the
first
See Holmstrom and Tirole (1997), for
'
that
is
regions,
Panel
reflected in the figure.
(i)
and
(ii),
There are two
(a) in figure 3 depicts this scenario.
which are distinguished by whether or not the international
collateral
constraint binds.
1
A
f
*
L= A
It
\
L= A
\
\
\
\
L=
1
J
L=
1
A
1
1
1
»
'
'
'
e„=i
Panel
The
demand from
:
new
interesting
the economy can
(a)
fall
Regions
(i)
loans which yields a region
is
high.
then interest rates
Panel (b)
(ii)
(iii)
is
the capital requirement
(iii)
or (iv) depending on
now
how
line in figure 6 (b)
equilibrium: loan
if
(in)
and
(iv)
Equilibrium with Banks
when
Conversely,
that
Regions
3:
demand
loan supply
rise to ration the scarce
ingredient to highlight
:
Figure
The dashed
distressed firms.
low and output
and
cases arise
into regions
e n =i
is
is
(i)
does bind. In this case
tightly
it
binds relative to
corresponds to a supply of
fully satisfied, interest rates are
constrained by capital requirements,
supply (dotted
line,
region (iv)).
The key
the supply curve for loans bends backwards at the point
where the capital requirement binds. This occurs because the value of banks' capital
when
falls
interest rates rise (see equation (9)).
The backward bending nature
differentiates
it
from region IV
of supply plays a significant role in region (iv),
in section 2.
While
in
and
both cases an aggregate collateral
constraint and a microeconomic capital requirement bind, in the banking-intermediated case
there
is
wasted
collateral, in
to the distressed firms.
full
aggregation.
the sense that not
Equilibrium
is
at point
all
excess domestic resources are channeled
A
rather than B, where the latter reflects
Constrained banks become a financial bottleneck, which bring about
.''.
simultaneously the worst of each scenario described in the previous sections: sharp declines
in asset prices coupled with collateral waste.
Multiple Equilibria
5.3
The
rise in
domestic interest rates
in region (iv)
is
not only a
symptom
of the
crisis, it is
also a cause of the crisis.
L= A
Figure
We
4:
Multiple Equilibria with Banks
illustrate this in figure 4.
causes the rise in interest rates.
As we described above, the contraction
The
fall in
in loan
supply
domestic asset prices amplifies the impact of
the crisis by deepening the credit crunch caused by banks' distressed balance sheets. 12 But
it is
easy to see that the feedback between asset prices and feasible intermediation brings
about the possibility of multiple equilibria.
'"'This amplification
prices
is
and theirs
lies in
fall in
is
asset prices,
on the supply side
and Moore (1997)
similar in spirit to Kiyotaki
in
that a
fall in
asset
Moore the
on the demand side - constrained demand reduces the productivity of assets, causing
and further constraining demand. However,
- constrained
and further constraining supply.
in
is
the identity of the agent with the crucial financial constraint. In Kiyotaki and
financial constraint
the
mechanism
reinforced by a tightening financial constraint. However, an import difference between our approach
supply causes interest rates to
In this sense, our
model
Krishnamurthy (1999).
26
is
in
rise,
our model the key financial constraint
is
causing bank capital to be compromised,
closer to the amplification
mechanism studied
Points (A) and (B) in the figure represent two equilibria which are distinguished by low
and high
interest rates, non-binding capital requirements (A)
(C). Point (B)
is
also a possibility in
which
interest rates rise to constrain loan supply to
exactly meet demand. While in both (A) and (B)
The equilibrium
so in (C).
with
<
A(j
1
Final
(C)
is
Low
production units are saved, this
all
is
not
Pareto inferior to that of (A) and (B). More generally,
or a concave but not rectangular
Pareto ranked.
6
in
interest rates, credit crunch
demand
interest rate equilibria are
more
function, equilibria can be strictly
than high rate ones.
efficient
Remarks
This paper has presented a model that highlights financial factors behind crises in emerging
markets. While
we
are not the
first
to point out
financial underdevelopment can lead to wasted
falling asset prices
First,
we
some
of these factors
—in particular that
collateral, as well as to the possibility that
can lead to disintermediation
are able to study the effects of domestic
—
our approach
is
and international
novel on two counts.
financial constraints in
a simple unified framework. Independently, both sets of constraints seem meaningful in the
context of emerging markets, and appear in different forms in the literature.
We
have found
that distinguishing between domestic and international collateral, and understanding
and when they
interact,
is
useful in thinking through crises. Second,
market underdevelopment,
financial
in a
dynamic context,
how
we show that domestic
will lead to the
undervaluation
of international collateral.
We
think that the collateral undervaluation result
through precautioning
to a
number
cessive short
(i.e.
is
particularly helpful in thinking
date 0) questions in emerging markets. Observers have pointed
of vulnerability factors - balance sheet maturity
mismatches caused by ex-
term debt, currency mismatches caused by contracting foreign currency debt,
shortages of public and private reserves, to
name
a few.
Yet there has been
little
said
about why firms make choices that lead to such vulnerabilities. In Caballero and Krish-
namurthy (2000b), we ask the question
of
why
firms,
knowing the
sheet mismatches, choose to contract foreign currency debt.
eign currency debt
that
if
is
We
show that contracting
and choose excessive dollar
liabilities.
13
will
our results points out that vulnerability
fies
11
is
is
undervalue international
Indeed we think that a number of these
vulnerability factors can be restated as ones that reduce international collateral,
The
for-
akin to contracting away international collateral. Thus our result
domestic financial markets are underdeveloped, firms
collateral
possibility of balance
and thus
structural to emerging markets.
results are also helpful in providing policy guidance. First of all the
model
identi-
the primitive behind the incidence of crises as the shortage of international collateral.
We
also connect domestic financial
market underdevelopment to the reluctance of foreign investors to
enter the domestic lending business.
27
Purely from an accounting standpoint, by sorting on their effects on international collateral,
the model helps us to classify and order the different vulnerability factors that others have
More
pointed out.
interestingly, since the collateral undervaluation result
externality, a natural line of inquiry
The
policies that
prescriptions of
(2000a),
emerge are
inflows. Since the
we
Pareto
losses.
in the design of policies to alleviate the externality.
ex-ante ones --
monetary policy during the date
we study the
tation,
all
is
stems from an
i.e.
precautions at date
1 crisis,
fn
0,
rather than
Caballero and Krishnamurthy
effectiveness of sterilization as a policy response to date
model motivates both the need
are able to characterize
when
for policy as well as
policy leads to Pareto gains,
models
its
and when
capital
implemenit
leads to
14
There are a number of related precautioning policy questions that the framework
provide guidance on.
will
For example, the question of which of the private or public agents
should hold international reserves, or the differential effects of domestic reserve requirements
and foreign
framework
liquidity requirements.
will
And, once extended to incorporate nominal
shed light on the role played by financial underdevelopment in the design of
optimal exchange rate arrangements.
1
Our
analysis
is
We
are currently working on these extensions.
guided by the effect of policy on international collateral - does the private sector's
reaction to a policy lead to a gain or loss in international collateral.
when the government bond market
may
lead to a
fall
assets, the
in the
is
illiquid, sterilization policy
A
result
which we find striking
(swapping reserves
for
is
that
government bonds)
spread between corporate lending rates and government bond rates. This causes
the private sector to overborrow from foreign lenders at date
dictate.
28
(I
the opposite of
what optimal
policy would
Appendix
7
Incentive Constraints
7.1
One
possible formulation to justify agents' borrowing constraints
contract between a firm and a lender at date
assume that the
realization
t
t
the support of
]
cost of
c.
Then,
The
contract will be a debt contract (see Gale and Hellwig (1985)).
a face of /;
R >
if
t
f the firm makes the repayment of
lenders pay the audit cost and receive
Assume
t
is
t
t
t
R
expected flows
scaled by Rt this
t
t
t
/]
_ Kobfr <
so that this
c,
Proof of Proposition
First of
all
L =
component
+ fProb{R >
t
f]
+
^^
>f] =
1
_
*<*%
_
is
<
/]
9t
Rt
we have
simply the complement of what the firm
0.
A
at date
Given
1,
this, let
The max problem
maxTr^A^ -
FOG
c(fc)
for this
t)
c
Xfiu will never bind.
1
The
reason
is
that
reinvestment
is
if it
did,
sufficiently
investment, firms would not saturate their budget constraint at
us substitute do./
for
=
the firm at date
+ Rk -
+
<
and since we have assumed that date
problem
K h {-d{k)
Let k{\
2
note that the constraint do j
profitable relative to date
Xd
Lenders receive
itself.
7.2
in
*
the component of firms that can be held by outsiders. In our formulation
suppressed the cost
The
/]
Rt
Rt
date
<
is,
E[R -f\R < f)Pro b{ R <
then
^
f]
of,
t
holds
define,
t
E[Rt - c\R < f}Prob[R t <f}
is
Then
return of Rt.
i.i.d.
the share of each firm that the entrepreneur necessarily holds.
is
This
lender can observe
c.
E[R - f\R > f]Prob[R >
=
Now
the firm defaults and
/; otherwise,
individually financed via a debt contract.
dt
[0,oo).
is
t
contract will specify
that each firm has a continuum of such projects each with
Each project
This
R —
R
at
standard to show that an optimal
fairly
it is
A
of the firm.
t
payment only by paying a
this
Suppose that a firm investing one unit
0.
R where E[R = Rt, and
of R is private information
yields date 2 flows of
date
as follows. Consider a
is
R-
fe/2)
+
c (^) in
the objectives.
is,
7r'(^-±-^/,;
+
^^(A - l) + I(A + l)(A^to -
c(fc)).
is,
1/2)
+ tt'(^±I +
^(A -
be the solution to this equation. Since, c(k)
.
29
is
- |(A +
=
0.
strictly convex, k(Xd)
is
1)
l)c'(k))
increasing
Proof of Proposition 5
7.3
For the decentralized equilibrium,
h h
m>axTr
(\ w - c(fc) + Rk - k/2) + n (^-k + -£j-(A - L') + ^(A + L
k
2
2
2L
l
'f
And
for the central
msx.-n
Since c(k)
sufficient for
is
h
{\)w - c(k)
+ Rk -
+ ^{^--k + A(AU -
k/2)
and the objective
strictly convex,
c{k)).
c{k)).
is
the
linear,
FOC
is
necessary and
an optimum. To compare the decentralized and central planner's solutions, we
ix\-d{k)
R-
+
1/2)
While the central planner's
h
ir
+
FOC
{-c'(k*)
clear that as long as
FOC
V(^ + ^g(A - L
l
)
is,
- i(A + L
l
)c'(k))
=
0.
is,
+ R-
A—L >
0,
+ n (^~^ l
1/2)
k*
<
Ac'(k*))
=
0.
k
Proof of Proposition 6
7.4
First note that the central planner's solution
all
w-
2
can simply compare the FOC's. The decentralized equilibrium
is
l
){\ f
planning problem,
k
It.
l
L
we need
The
FOC
F(k, \ d )
show
to
for
=
is
that /c(Ad)
is
+
R-
1/2)
+ tt'(
L
—i
R L
+
i
)
^(A V)
L -\
-(A + L')c'(k))
ij^-
'
)
-
{A
x drk
_
X
l
t
w - d0J
Then, we wish to show that,
ML
<
<
0.
dk
we can show,
f = -c»(^W^)-fi,v4 +
c<(i,,<0.
Also,
i
Sr -*¥ + '<><°Combining the above
results,
k*.
Then
is,
where,
After some algebra
>
decreasing.
the decentralized problem
n»(-c'(k)
independent of A^, and that k
is
we can conclude that
30
/c(Arf) is
decreasing in Ad-
=
0.
Supporting Formulae for Banking Section
7.5
The
solution to
program (P7)
is
determined by the binding constraints.
Constraint
(i)
requires that loans satisfy,
a
Now
given loans and deposits,
where the
last step follows since
the
amount
is
(
A
L"d
Lf
equal the amount of funds lent to firms
which
Q) =
A + x\ A {u>)
a
% _
L<"
of funds raised from depositors
'^
) Thus,
if
(
'y^
)
must
the capital constraint binds,
decreasing in LF.
Turning next to the decisions of firms, demand
from intact firms follow from the same
for loans
logic as before (see
by distressed firms and deposits
PI and P2). Intact firms extend
deposits to banks of x± .^(u;) (face value of date 2 goods),
Lp =
We
=>
l
x
M (w) < XJw - d
,/.
write the problem for distressed firms assuming that the domestic collateral con-
straint never binds.
Distressed firms take out bank loans of di td (u)) to save production
units.
A>LU
d\.d{u)
'
Z>
A = Lw
=>
^M
=max[fc-d?:/ ,0],
<
m ax[fc„ -
<,,()].
Vis-a-vis the previous analysis, market clearing conditions need to be modified slightly
The aggregate amount
to take into account banks.
,i""n"Mir dcpo
.its
i)i
the ml
in
t
H'..
c.l
fii
must equal the
In ins
Likewise, the total loan supply from banks
dist
of deposits in banks
must equal the new domestic debt issued by
ins
x bhd (uJ )
= Xl d (u;) = D hd (uj) =
31
^
d
Taken together and combined with the capital constraint on banks
j^ - —
A
cf
Xfw - d0J <
Lra
Consider the
first
M (w) = x ia
d
(uj)
a
Lr
When
inequality.
=>
this implies that,
the capital requirement on banks does not bind, banks
aggregate domestic collateral perfectly - they channel
the intact firms to the distressed firms.
all
of the excess debt capacity of
The economy behaves
exactly as
if
A„
=
1
in a
public debt market (section 2.4). Asset prices and investment depend on the international
collateral constraint,
When
k
< 2(\?w
-
'
do..f)
=»
Lu
k
> 2{\^w
-
~
doj)
=>
L"
the capital requirement
depending on how tightly
L"D =
1
it
(i)
-.
=
L£ =
1
=
=
L"D =
A
binds, the
binds relative to
<r
e
economy can
demand from
fall
w
=
<
1
i.
into regions
—
—
32
=>
L"
=
=>
L"
=A
l
or (iv)
distressed firms. In this case,
and:
-(&-(A>-cZ f))<
-A
2
a
hk - {X?w - do,/)) > - A
(iii)
6^
9"
=
=
1
1
References
[1]
Aizenman, Joshua, "Country Risk, Incomplete Information and Taxes on International
Borrowing," Economic Journal
[2]
Allen, Franklin
March
99,
1989, pp. 147-161.
and Douglas Gale, "Limited Market Participation and
Volatility of
Asset Prices," American Economic Review 84(4), September 1994, pages 933-55.
[3]
Allen, Franklin
and Douglas Gale, "Financial Markets, Intermediaries, and Intertem-
poral Smoothing," Journal of Political Economy, 105(3), June 1997, pages 523-46.
[4]
Bardhan, Pranab K., "Optimum Foreign Borrowing,"
Theory of Optimal Economic Growth:
[5]
American Economic Review
Bank
proaches
to
series:
Blommestein, Hans
rities
Markets,"
J.,
March
1989, pp. 14-31.
IADB
WP #
359,
November
New
ap-
Economic Theory and
69- 88.
Reforms and Emerging Secu-
Bulow, Jeremy and Rogoff, Kenneth, "A Constant Recontracting Model of Sovereign
Caballero, Ricardo
J.
Economy
Caballero, Ricardo
J.
97-1,
February 1989, pp. 155-78.
and Krishnamurthy, Arvind, "Emerging Markets
IMF WP,
Caballero, Ricardo
J.
Crises:
An
1999.
and Krishnamurthy, Arvind, "International Liquidity Manage-
mimeo MIT, January
and Krishnamurthy, Arvind, "Dollarization of
derinsurance and Domestic Financial Underdevelopment,"
[12]
J., ed..
1997.
ment: Sterilization Policy in Illiquid Financial Markets,"
[11]
Kenneth
International Symposia in
"Institutional Investors, Pension
Assets Market Perspective,"
[10]
79-1,
Cambridge University Press 1987, pp.
Debt", Journal of Political
[9]
Essays on the
Press, 1967, pp. 117-128.
Policy", Barnett, William A., ed.; Singleton,
monetary economics.
Econometrics
[8]
ed.,
Bhattacharya, Sudipto, and Gale, Douglas M., "Preference Shocks, Liquidity, and
Central
[7]
Karl Shell,
Bernanke, Ben and Mark Gertler, "Agency Costs, Net Worth, and Business Fluctuations,"
[6]
MIT
in
Calvo, Guillermo A., "Balance of
Payment
Crises in
2000a.
Liabilities:
mimeo MIT,
Un-
July 2000b.
Emerging Markets," mimeo, Mary-
land, January 1998.
[13]
Chang, Roberto and Andres Velasco, "Financial Crisis
ical
[14]
Model,"
mimeo Atlanta
in
Emerging Markets:
A
Canon-
Fed, July 1998.
Diamond, Douglas W. and Dybvig, Philip
Liquidity," Journal of Political
Economy
33
H.,
91-3,
"Bank Runs, Deposit Insurance, and
June 1983, pp. 401-19.
[15]
sults
[16]
and Mary Hallward-Driemeier,
Dollar, David,
"Crisis
from the Thailand Industrial Survey," mimeo
Adjustment, and Reform: Re-
May
1998.
Gale, Douglas and Martin Hellwig, "Incentive Compatible
Debt Contracts: The One-
Period Problem" Review of Economic Studies 52(4), October 1985, pp. 647-63.
[17]
Gelos, R. Gaston and Alejandro
straints,
and
Collateral:
WP/99/25, March
[18]
Investment
in
Liberalization, Credit Con-
the Mexican Manufactuing Sector,"
IMF
1999.
Goldfajn, Ilan and Rodrigo O. Valdes, "The
Intermediation,"
[19]
M. Werner, "Financial
IMF mimeo, August
Twin
Crises and the Role of Financial
1998.
"Lessons for Debtor- Country Managers and Policymakers," in
Harberger, Arnold,
J.Cuddington and G.Smith
(eds.),
International Debt and the Developing Countries.
Washington: World Bank, 1985.
[20]
Holmstrom, Bengt and
Tirole, Jean, "Financial Intermediation,
Loanable Funds, and
the Real Sector," Quarterly Journal of Economics 112-3, August 1997,
[21]
Holmstrom, Bengt and
of Political
[22]
Jacklin,
Prescott,
ral trade.
of
[23]
Economy
Charles
Tirole, Jean, "Private
Edward O,
and Public Supply of Liquidity," Journal
106-1, February 1998, pp. 1-40.
"Demand
J.,
pp 663-91.
ed.;
Deposits, Trading Restrictions,
and Risk Sharing,"
Wallace, Neil, ed.. Contractual arrangements for intertempo-
Minnesota Studies
Macroeconomics
in
series, vol. 1
Minneapolis: University
Minnesota Press 1987, pp. 26-47.
Kang, Jun-Koo and Rene M.
Portfolio Equity
Ownership
Stulz,
in
"Why
Is
There
Home
Bias?
An
Anlysis of Foreign
Japan," Journal of Financial Economics 46-1, October
1997, pp. 3-28.
[24]
Kehoe, Timothy, and David Levine, "Debt-constrained Asset Markets," Review of Eco-
nomic Studies
[25]
60-4,
October 1993, pp. 865-888.
Kiyotaki, Nobuhiro and Moore, John, "Credit Cycles,"
Journal of Political Economy,
105-2, April 1997, pp. 211-48.
[26]
Merton, Robert, "A Simple Model of Capital Market Equilibrium with Incomplete
Information," Journal of Finance, 4%-3, July 1987, pp 483-510.
[27]
[28]
World Economic Outlook. IMF,
May
1998.
The rWexk} Bank, East Asm: The Road
J L
6 U H J
to
34
Recovery, September 1998.
Date Due
Lib-26-67
MIT LIBRARIES
3 9080 01917 6152
M
f;
;
fclfc-I
Mlp
:
In'fllJHiHHHIilJH^
'.
IBiBilllimil
:;,i:
i
i
:
I fliHIwIiiiiW
lili
M s
itll
:.
fi
ptlf
MmiliflmwiiiiUtUliiiU,
-
!
-'''- ;;
'..';