Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium
Member
Libraries
http://www.archive.org/details/exchangeratevolaOOcaba
HB31
M415
DEVVHy
•
Massachusetts Institute of Technology
Department of Economics
Working Paper Series
Exchange Rate Volatility and the Credit Channel
Emerging Markets: A Vertical Perspective
Ricardo
J.
Caballero
Arvind Krishnamurthy
Working Paper 04-24
May
Room
1
2,
2004
E52-251
50 Memorial Drive
Cambridge,
This
MA 02142
paper can be downloaded without charge from the
Network Paper Collection at
Social Science Research
http://ssm.com/abstract=555501
in
MASSACHUSETTS INSTtTUT!
OF TECHNQtOqv
'" '-"""".i.N
'"
I
DEC
7 200*»
LIBRARIES
i
Exchange Rate
and the Credit Channel
Volatility
Markets:
Ricardo
J.
A
in
Emerging
Vertical Perspective
Arvind Krishnamurthy*
Caballero
This draft:
May
12,
2004
Abstract
Firms
emerging markets are exposed to severe financial
in
straints, that are
and
frictions
Can monetary
exacerbated by the sudden stop of capital inflows.
policy offset this external credit squeeze?
We
show that although
credit con-
this
may be
the
case during moderate contractions (or in partial equilibrium), the expansionary effect
of
monetary policy vanishes during severe external
The exchange
crises.
rate
jumps
to reduce the dollar value of domestic collateral until equilibrium in domestic financial
markets
consistent with the external constraint.
is
An
expansionary monetary policy
domestic collateral but
in this context raises the value of
it
exacerbates the exchange
rate depreciation (beyond the standard interest parity effect)
aggregate activity. However there
sudden
The
stops.
is
and has
little effect
on
a dynamic linkage between monetary policy and
anticipation of a dogged defense of the exchange rate worsens the
consequences of sudden stops by distorting the private sector incentive to take precautions against these shocks. For similar general equilibrium reasons, dollarization of
liabilities
has limited impact during a sudden stop, but
it
has significant underinsurance
consequences.
Gl
JEL Codes:
EO, E4, E5, FO, F3, F4,
Keywords:
External shocks, segmented capital markets, credit squeeze, monetary
policy, interest parity departures,
exchange rate overshooting,
fear of floating, underin-
surance, capital controls.
MIT
'Respectively:
expanded version
of:
lems," April 2001,
and NBER; Northwestern University. This paper
is
an extensively re-written and
"A Vertical Analysis of Crises and Intervention: Fear of Floating and Ex-ante Prob-
and "A Vertical Analysis of Monetary Policy
in
Emerging Markets," March 2003.
We
thank Francisco Gallego, Andrew Hertzberg, Stavros Panageas, Raghuram Rajan, Roberto Rigobon and
Andres Velasco
for their
comments on these
earlier versions, as well as
seminar participants at Bank of
Canada, Brown, Chicago, Columbia, IMF, MIT, Federal Reserve Bank of
and the
ME-NBER
group.
a-krishnamurthy@nwu.edu.
Caballero thanks the
NSF
New
for financial support.
York, the World Bank,
E-mails:
caball@mit.edu,
Introduction
1
Emerging markets can
What
("sudden stops").
capital inflows suddenly reverse
the right monetary policy response to this event? Answering
is
important question requires understanding the mechanisms through which monetary
this
policy can affect sudden stops
is
when
suffer severe contractions
and
their consequences.
The main contribution
of this paper
to re-think the workings of the credit channel in emerging market economies,
propose an alternative view that
shifts the focus
and
to
from credit-channel and aggregate demand
to insurance considerations.
Much
of the recent literature
on emerging markets
crises highlights
the limited financial
development of these economies and the severe credit squeeze experienced by
during
From
crises.
this structure,
local firms
two opposing arguments are commonly made regarding
optimal monetary policy. Extrapolating from developed economy credit channel analysis,
some advocate an expansionary monetary policy
to offset the effect of the credit squeeze
during downturns. While, others advocate a contractionary monetary policy and dogged
defense of the exchange rate during crises. Proponents of the latter view do not disagree
upon the
centrality of the credit channel
mechanism, but argue that a depreciating exchange
rate will have dramatic effects in the credit channel mechanism, through a deterioration in
The argument
borrowers' balance sheets.
the
economy
The
is
and when
dollarized
is
that this effect
inflation-credibility
starting point of our analysis
A
firm
may
most
limited.
likely to
happen when
1
the observation from Caballero and Krishnamurthy
is
(2002) that credit constraints in emerging markets
well as the firm level.
is
is
have limited
may
exist
collateral,
both at the country
level as
and therefore be constrained
borrowing from either domestic or foreign lenders. Or, the country as a whole
in
may have
limited international collateral, and therefore domestic firms are constrained in borrowing
from foreign
investors.
2
Extrapolating from developed economy analysis misses this dis-
tinction because in developed economies the credit constraint plausibly exists only at the
The
firm level.
distinction
important
is
through the
for thinking
because a sudden stop created by the reversal of capital inflows
effects of
is
monetary policy
predominantly about a
binding international collateral constraint.
We
argue that during mild
constraints
1
immaterial.
al (2000),
The importance
the distinction between domestic and international
Analysis of this case can be conducted in the standard credit
For both sides of the argument
Cespedes et
2
is
crises,
Gertler et
al
see, e.g.,
Furman and
(2001), Christiano et
Stiglitz (1998), Fischer (1999),
al.
of international collateral constraints for
sovereign debt literature
(see, for
Aghion
et al (2000),
and Calvo and Reinhart (2003).
emerging markets was first identified
(2004),
example, Eaton and Gersovitz, 1981, or Bulow and Rogoff, 1989).
in the
Much
channel framework.
of the recent debate on monetary policy during crises implicitly
assumes this to be the relevant scenario. As
is
stressed in the credit channel literature (e.g,
Bernanke, Gertler, and Gilchrist, 1989, and Kiyotaki and Moore, 1997), increasing the net
worth of borrowers
is
There
reinforces the conclusions in the literature:
optimal monetary policy
liability dollarization
The
Our
the optimal response to a negative shock.
and
recipe of using
is
is
some ambiguity regarding whether the
to raise or lower interest rates.
It
depends on the extent of
inflation credibility of the central bank.
monetary policy to increase the net worth of borrowers
the credit constraint exists at the country
The reason
level.
is
is
a foreign investor, monetary policy
international collateral of the country.
markets'
3
We
have important general equilibrium
But
liquidity.
in the relative price of
is
it
does not alter the
the relevant case for emerging
on output during a severe
little effect
effects
on asset
in general equilibrium the
prices.
crises,
they
In partial equilibrium,
it is
by increasing
their
fall in
during the
expansion just translates into a decline
domestic liquidity vis-a-vis international
in international liquidity (i.e.
it
liquidity,
not in an expansion
does not loosen the international constraint).
We show
that
the price of domestic liquidity implies a temporary exchange rate depreciation
crisis.
In fact, the depreciation of the exchange rate
due to the standard
as
argue that this
because
true that a monetary expansion raises firms' borrowing capacity
domestic
the
ineffective
Thus when the marginal
crises.
Although monetary expansions have
still
is
when
fails
that monetary policy only
regulates the borrowing of a domestic agent from a domestic lender.
lender
analysis of this case
interest parity
some have argued
that are to blame;
beyond the depreciation
mechanism. The depreciation does weaken balance sheets
in the context the credit channel.
The exchange
is
rate depreciation
is
But
it is
not firm-level balance sheets
just the general equilibrium conterpart
of the tight international constraint.
Our
conclusions regarding the effects of monetary policy are independent of the dol-
larization of balance sheets.
dollarized balance sheets
again,
is
Monetary policy
undo the
is
impotent during
While
whether or not
beneficial effects of lowering interest rates.
that domestic dollarization just tightens constraints at the firm level,
alter the aggregate international collateral of the country.
3
crises
in
The
it
reason,
does not
4
very different terms, Dornbusch (2001) also expresses his uneasiness about the assumption that
emerging economies with access to monetary policy can use
it
to boost activity. For example, in criticizing
the standard view, he writes: "The loss of the lender of last resort [argument]
is
intriguing. This
argument
is
based on the assumption that the central bank - rather than the treasury or the world capital market -
is
the appropriate lender..."
4
Dollarization of external liabilities has
later in this
paper and, more thoroughly,
some
in
additional external insurance implications which
Caballero and Krishnamurthy (2003a). This
is
we
discuss
not an issue
for
The second main theme
monetary
of the paper concerns the insurance effects of
policy.
Despite the apparent impotence of monetary policy during sudden stops, expectations re-
garding a particular monetary rule have important effects on the private sector's ex-ante
The
decisions.
anticipation of
monetary policy actions during
crises affect agent's expec-
tations of the relative price of domestic to international liquidity, which in turn affects the
private sector's insurance decisions against sudden stops.
dogged defense of the exchange
rate,
then
it
then
if it
expects an expansionary monetary
expects a sharper decline in the relative price of domestic liquidity.
it
incentive to hoard international over domestic liquidity
than in the former
when
the private sector expects a
also expects a smaller decline of the value
of domestic to international liquidity. Conversely,
policy,
If
is
more pronounced
in the latter
Importantly, hoarding more international liquidity
case.
The
is
desirable
there are negative pecuniary externalities in the use of international liquidity, a fea-
ture that
is
inherent to economies with underdeveloped financial markets (see Caballero
and Krishnamurthy 2001).
Prom this perspective, our argument connects with the literature on the virtues of flexible
versus fixed exchange rate systems.
5
In our model, a flexible exchange rate system coupled
with a countercyclical monetary policy, provides better insurance incentives to the private
6
sector than a policy of defending the exchange rate during sudden stops.
monetary policy and
its
More
generally,
constraints have (unintended) consequences for the private sector's
insurance decisions with respect to sudden stops.
Policies that exacerbate the (partial
equilibrium) domestic credit squeeze during sudden stops, or constraints that limit the
central bank's ability to relax domestic financial constraints, lead to ex-ante imprudent
private sector actions.
In Section 2
we present a model
between domestic and international
of market segmentation in which
liquidity. Section 3 describes
etary policy in the segmented financial market. In particular,
of the exchange rate to
monetary
policy,
and the
role of the
we draw a
distinction
the consequences of
we show
mon-
the over-sensitivity
exchange rate in endogenously
aligning the extent of domestic credit squeeze with the limited international liquidity. In
Section 4
we turn
implications.
We
to the ex-ante consequences of
show that the strategy
monetary policy and
of tightening
private insurance
its
monetary policy during
the private return to taking preventive measures and hence discourages
we allow for state contingent debt.
Edwards and Levy-Yeyati (2003) for evidence supporting the advantages
crises lowers
it.
the analysis of this paper since
5
See, e.g.,
of a flexible exchange
rate in absorbing external shocks.
6
Our mechanism
is
distinct (or in addition to) the standard free insurance
perspective of crises (see,
e.g.,
Dooley 2000).
We
argument
in
the moral hazard
return to this comparison after developing our model.
In section 5
we
revisit the dollarization of liabilities
We
of our dual-liquidity model.
in this literature
argument, but now in the context
show that the balance-sheet
effect typically
emphasized
has no aggregate consequences in general equilibrium during the
However, and following our argument above, there
is
crisis.
a dynamic cost. Dollarization lowers
the expected return on hoarding international liquidity and hence reduces the private sector incentives to take adequate precautions against
sudden stops. In section 6 we highlight
that not having the credibility or willingness to conduct a countercyclical monetary policy
means that an insurance mechanism against
ternative measures that induce the private sector to carry
Examples
crisis states.
liquidity requirements,
may be
sures
costly,
Thus, one should look for
crises is lost.
more international
al-
liquidity into
of these measures include taxation of capital inflows, international
and large
While enacting these mea-
sterilizations of capital inflows.
they should be seen as yet another cost of having
the ability to
lost
use monetary policy in an environment of recurrent external crises.
Section 7 shows that our main conclusions are robust to the relaxation of some of the
main
stylized assumptions of our model, such as the presence of a "diagonal" supply of
funds or an alternative model of money. Section 8 concludes and
is
followed by a technical
appendix.
A
2
We
Model
study a three date economy
fully flexible
At date
at date
if
of External Crises
1, u>
=
0, 1,2)
Date
with a single (tradeable) good.
is
a
planning period. Agents make investment and financing decision at this date.
an external
1
(t
G
{b, g},
crisis
may
occur.
We
assume that there are two
which occur with probabilities {n,
the 6-state realizes. At date 2 agents repay
all
1
—
n}.
The
states of the world
crisis
can occur only
debts and consume.
Private Sector
2.1
There
is
a unit measure of domestic firms.
Building a plant of size k at date
c"
>
of
Ak
0.
goods
Each has access to a production technology.
requires a firm to invest c(k)
The output from production depends on the shocks
output goods at date
may be
produced.
A
2,
but
more
if
at date
1.
c(.)
>
There
detailed description
at date 0.
their investment.
is
0, c'
is
negative shocks hit firms then only ak (a
Domestic firms have no resources
row from foreigners to finance
— with
a
>
and
maximum
< A)
output
given below.
They must import
Each firm
neur/manager who has risk-neutral preferences over date
2
is
capital goods
and bor-
run by a domestic entrepre-
consumption of the single good.
Thus, financing and investment decisions are taken to maximize expected plant profits at
date
2.
Domestic firms face
affects
significant financial constraints
investment decisions.
collateral, in
their financial net
worth
We assume that firms are endowed with w units of international
the form of receivables arriving at date
these receivables have collateral value to foreigners
We
and hence
2.
Our
(e.g.,
central assumption
contingent on the state at date
1.
We
all
financing
we assume that the debt
relax this in Section 5
external debt, since the problem of dollar debt
is
that only
prime exports).
disregard explicit equilibrium default and assume that
collateralized debt contracts. Moreover,
is
essentially
is
done via
fully
made
fully
contracts are
when
one of
discussing dollarized
that are too
liabilities
inflexible.
At date
0,
when
firms sign debt contracts with foreigners, the contracted repayments of
/" must not exceed w. Foreigners lend against
ij$
and
i\
We
from period
also
to
1,
and
1
to
2,
1 at
the rate
respectively.
assume that domestic agents have some "domestic"
trade to other domestic agents. First,
and
this collateral at dates
assets
which they can only
we assume that the minimum date
output of ak
2
can be pledged as collateral to other domestic agents. Second, we assume that agents are
endowed with M_ units of domestic money
(see below) that
they can
sell
to other agents.
This modelling captures the distinction between domestic borrowing capacity and
ternational borrowing capacity.
We
in-
think of the domestic assets as "peso" assets, while
the international collateral are "dollar" assets. Thus, changes in the exchange rate have an
effect
2.2
A
on the dollar value of domestic assets but not that of the international
assets.
Credit constraints during crises
crisis, is
Formally,
rises in
a time
when
we model the
there
is
a shortage of resources available to finance investments.
external crisis by assuming that the
demand
the 6-state, while holding the international collateral fixed.
realistic)
for external resources
An alternative
(and more
modelling of a shortage of external resources would be to assume that international
collateral contracts during the crisis, while
demand
for external resources
remains
fixed.
The
implications of these two modelling strategies are similar for our purposes, but our approach
reduces the
to
number
of assumptions
and equations required. In
particular,
add an assumption of incomplete international insurance markets.
In the g-state there are no shocks and investment
7
is
unaffected.
See Caballero and Krishnamurthy (2001) for the alternative approach.
we do not need
7
Let us
See Broner et
now turn
al.
to
(2003) for
evidence that some of these international collateral shocks are due to shocks to specialist investors (in our
framework, this
collateral).
is
as
if
suddenly foreign lenders decide to reduce the share of
w
that
is
part of international
defining the shock in the 6-state,
The
and explaining how
affects investments.
it
plants of one-half of the firms receive a shock at date
plant from
A
The shock only
to a.
arrives in the 6-state, but
<
1)
goods, to give date 2 output
=
A(9)k
=
so that A(l)
We
A.
(a
+ 9A)k <
idiosyncratic in that each
The productivity
firm receives the shock with probability 1/2.
reinvesting 9k (9
is
decline can be offset
by
of,
A = A - a,
where
Ak,
that lowers output per
1
assume that the return on reinvestment exceeds the international
interest rate:
A-l>i*v
This means that firms
occurs
if
will
borrow as much as possible to finance reinvestment.
firms are financially constrained at date
1,
so that 9
<
1.
We make
A
crisis
parametric
assumptions to ensure that this occurs in equilibrium only in the 6-state (see the appendix)
A
firm that receives a production shock
the firm
w — f.
termed distressed. To cope with the shock,
borrows directly from foreigners against
first
After
is
this, it
must turn
for
its
net international collateral of
funds to the domestic firms that did not receive a shock
(termed intact).
As noted above, domestic agents accept the money held by
firms as payment.
They
also accept
ak of the output from a firm's plants as
any domestic debt. Thus the domestic
date 2 goods that can be pledged to domestic agents)
ak
where
e<i
collateral for
entering date
(real) liquidity of firms
distressed
1
(i.e.,
the
is,
+ M/e-2,
(1)
the date 2 nominal exchange rate (also the price level).
is
Intact firms have no output at date
are to finance the distressed firms.
1 either,
At the
so they
must borrow from foreigners
interest rate of if
,
if
they
they can borrow up to f^f
from foreigners.
Since at date
1
firms, in aggregate,
can borrow from foreigners up to
*•
we
refer to
2.3
The
wn
to inject
&(•
<
2>
as the international liquidity of the country during the crisis.
Central
central
=
Bank
bank has
M units of money outstanding
(M — M) more money
"helicopter" drops to firms,
into firms.
and redeemed
We
at date 0.
assume that
At date
this
money
at date 2 with taxes collected
1, it
is
may
choose
distributed as
by the government.
We do not provide a detailed description of the government's tax powers and its tax base.
Instead we assume that the government collects T goods via non-distortionary taxation,
and that these taxes do not come from firms. We think of these taxes as resources from a
consumer sector that has a date 2 endowment of y c
monetary policy
We
transfers resources
assume that the central bank
money
so that
when we
>
T.
When
it is
effective, expansionary-
from consumers to the corporate
is
sector.
8
credible in maintaining the date 2 price level at one,
We
injections do not lead to inflation.
relax this assumption in Section 6
discuss inflation credibility. Thus, for now, the date 2 exchange rate
Our model
of the
them
is
model. Despite
this,
rate. For this reason
as "over-responsive."
from the movement
7,
"peso" interest rate
=
1.
frictions,
frictions
is
and
zero in our
unaffected by monetary policy. Therefore the usual channel (through the
interest parity condition)
Section
The domestic
simplifies the exposition.
environment and
e2
monetary transmission mechanism involves no monetary
because our qualitative conclusions do not depend on the presence of these
ignoring
is
by which monetary policy
we show that
we
refer to
Our
in our
affects the
exchange rate
model monetary policy
is
still affects
absent in our
the exchange
the exchange rate movements induced by monetary policy
be interpreted as in addition to those that arise
results should
in the peso-rate
and
its effect
we sketch a more standard model
in
through the interest-parity condition. In
which money provides a transaction
service,
thus confirming that our results are robust to this simplification.
Monetary
3
Policy,
Exchange Rate Determination, and the
Credit Channel
In this section
we examine the
we take the date
equilibrium in the
crisis state at
date.l.
Thus, for now,
investment decision of k and the financing decision of / as given.
study the case where credit constraints are binding in this
We
crisis-state, illustrating that the
power of monetary policy to relax these constraints changes from mild (horizontal) to severe
(vertical) crises.
All of the discussion in this section takes place after the
1
and the central bank has injected
3.1
Mild and Severe Crises (Date
At date
1,
distressed firms need to
total net worth,
8
(M — M
See, e.g.,
measured
borrow
in dollars, is
)
bad state has
pesos in response.
1)
in order to finance their
ak +
realized at date
M + wn
.
Clearly as
Woodford (1990) and Holmstron and Tirole (1998)
for
M
production shock. Their
rises so
does net worth,
a similar assumption.
DateO
Datel
—I
Date 2
H
1
Shocks, coe{b,g}
Borrowing
Monetary
Investment, c(k)
Output
M-
Injection,
Repayment
M_
Consumption
Production: Time-to-build
(INTACT)
k
R
k
A
e
k
=
R
(e
Prob=1/2
(DISTRESSED)
Need
additional investment of
6
r
k import goods
Figure
at least for a given e±.
The key question
on the severity of the
e\
depreciates with an expansionary
the latter
is
severe, a
monetary expansion
ineffective in raising
output and leads to an exchange rate over-reaction.
despite
target credibility
full inflation
domestic collateral
(i.e.
{e-i
1),
as
it
reflects
the
fall in
This happens
the real value of
1.
Distressed firms can borrow
wn
directly
from
Thereafter they must turn to other domestic agents to raise resources.
Recall that intact firms can also borrow
are willing to accept domestic
by ak of
=
is
not a monetary phenomenon).
Let us consider equilibrium at date
foreign investors.
k
argue that the answer to this question depends
If
crisis.
)
Time Line
1:
whether
is
We
monetary policy and by how much.
critically
k +
collateral,
money
as
wn
from foreign investors. Moreover, since they
payment as well as purchase debt claims backed
they can also finance the production shock of the distressed firms.
There are two regions of
interest.
If,
1
_
-w n >
lak +
2
i
M
+ tj
(3)
demand from
then intact firms have sufficient access to foreign funds to satisfy
firms.
firms
We refer to this case
The other
as the Horizontal region, because the supply of funds
distressed
from intact
margin.
elastic at the
is
all
when,
case, is
1
_
lak + M
-w < -2
2 1 + z*'
,
while
wn <
That
k/2.
distressed firms.
We
intact firms have insufficient resources to satisfy
is,
demand from
because the supply of funds
refer to this case as the Vertical region
is
inelastic at the margin.
3.2
Horizontal region (mild crises):
The standard
Intact firms have portfolios that include domestic loans, money,
Indifference requires that
to 1
from
+ if,
this indifference condition.
the exchange rate
ei
This equation
is
well.
collateral.
The exchange
Since the return on holding
is,
= l + tj.
(4)
the interest parity condition given that e2
special transaction service (so the peso-interest-rate
is
and international
be the interest rate on the domestic loans as
i\
rate can also be determined
money must be equal
credit channel
is
=
1
zero).
and that money provides no
Note that the exchange rate
unresponsive to monetary injections.
Total reinvestment
is
determined by the total resources of individual distressed
ei
where the superscript
is
w
+ z*
H denotes the horizontal equilibrium.
the economy has not used
economy
1
in a crisis:
all
of
The
international liquidity, while
its
firms',
distressed firms are not able to
meet
first
inequality shows that
9H <
1
indicates that the
their production shock fully
because of their binding financial constraints.
As we mentioned above, we
of loans
is
refer to this as the horizontal region
not affected by their quantity.
borrowing at the given interest rate
We
can see the
effects of
though a credit-channel a
money
(4)
i\,
monetary
la
In principle, a distressed firm could continue
as long as its
injections
this policy
monetary expansion
collateral.
It is
is
own
financial constraint
Bernanke and Gertler (1995). From
has no
effect
is
from the preceding expressions.
increases the resources of the distressed firms
we note that
because the price
(5)
and increases
on exchange rates
we note
relaxed.
It
operates
that injecting
their investment.
From
in the horizontal region, so the
a powerful mechanism to raise the dollar value of distressed firms'
possible that reintroducing the standard interest parity effect (which
we
have suppressed) will work against this conclusion. But as we show next, in the Vertical
region
we reach
the unambiguous conclusion that the exchange rate effect will neutralize
monetary expansions.
The
3.3
Exchange Rate
Vertical region (severe crises): Over-responsive
In this region, the international supply of funds faced by emerging market economies during
external crises
is
vertical
price inelastic). Since all investment at date 1
(i.e.
financed by foreigners, the stock of international liquidity
is
is
eventually
what determines investment
in
this region:
9
where the superscript
not appear at
all
V
vk
= 2w n
(6)
,
stands for vertical equilibrium. Note that domestic collateral does
in this expression.
Consider the exchange rate next. Since intact firms borrow up to their
maximum capac-
from foreigners and hold portfolios only composed of domestic money and loans against
ity
domestic collateral,
it
must be that the return on holding domestic money exceeds
i\.
That
is,
ei
and the date
exchange rate
1
is
Indifference between holding
> 1-Hi,
depreciated.
money and domestic
loans implies that the (dollar) interest
rate on loans against domestic collateral also rises above
by
alized
of
if,
w
are
made
at rate
i\,
In equilibrium, loans collater-
made
while those collateralized by ak are
Figure
1
= ei- 1 > i\.
represents the equilibrium determination of e\.
The
vertical axis is the price,
while the horizontal axis measures domestic reinvestment. For e\
e\,
elastically
n
,
at a higher rate
where,
if
w
On
i\.
supply their international liquidity to distressed firms.
intact firms
run out of international
liquidity,
= 1 + i\,
intact firms
However, at the point
and the supply curve turns
the other side of the domestic financial market, distressed firms
demand
international liquidity as they can obtain as long as the exchange rate
is
less
as
vertical.
much
than or equal
to A. However, they are constrained in their domestic borrowing by their holdings of
and domestic
point,
and
collateral.
effective
The demand
demand
horizontal region (panel a)
is
M + ak
and one
.
for international liquidity turns
Figure
1
downward
money
at
some
represents two cases: one equilibrium in the
in the vertical case (panel b).
10
as
(a) horizontal
(b) vertical
loans
Figure
This figure illustrates
liquidity
ment
is
scarce.
Note
between
1
+ i\
how
Equilibrium
e\ rises
also that e\
for distressed firms.
strictly
2:
1
+ i\
in the vertical region
is
a large interval for
reinvestment.
This
is
From
international
demand within which
e\ lies
this case,
M + ak >
1
w"
+ iJ.
(7)
can see that monetary injections have a very different
in the horizontal region.
when
can never exceed A, the marginal product of reinvest-
ei
We
the domestic loans market
in
above
However, there
and A. In
loans
(6),
we note
that injecting
effect in
the vertical than
money has no
effect
because the economy has a shortage of international
on date
liquidity,
1
and
reallocating domestic liquidity has no real (aggregate) effect.
As
before, the net
worth of distressed firms
w" +
ak
is,
+M
ei
That
is,
the credit channel
is still
termines reinvestment at date
injections. Referring
for-one
1.
active
But the exchange rate now
back to equation
by a depreciation
in the
and the net worth of distressed firms
(7),
exchange
we
see that
is
what
de-
over-responsive to monetary
monetary
rate. Essentially, the
is
injections are offset one-
exchange rate depreciates to
ensure that the dollar net worth remains at a level consistent with the availability of exter-
11
nal funds. This role of the exchange rate, captured in (7),
is
what
behind the monetary
is
ineffectiveness result.
Discussion:
3.4
The
credit channel in
emerging markets
Panels (a) and (b) in Figure 2 summarize the differences between the horizontal and vertical
views. In their
downward sloping segments, demands
are equal to (ak
+ M)/e\.
In the horizontal case, the increase in domestic net worth of distressed firms caused by
expanding monetary policy
region, the
same
raises date 1 investment, leaving e\ unaffected. In the vertical
increase has
no
effect
on equilibrium investment, and only
(a) horizontal
raises e\.
(b) vertical
"A
\
\
I
\
\
's
N
\V
N
x
'
>
loans
Figure
We
there
is
3:
loans
w"
Domestic
view the horizontal region as relevant
liquidity
for
expansion
developed economies;
no distinction between domestic and international
liquidity. It
w
is
may
plentiful
and
also apply for
emerging economies experiencing moderate contractions. The vertical region, on the other
hand, seems to be a better description of the environment faced by emerging markets during
severe external crises.
There are two points to emphasize from
mechanisms overturn the
region (for example, debts
region,
this analysis.
result that injecting
may be
money
dollarized, as
we reach the unambiguous conclusion
12
we
is
First,
it
may be
that other
expansionary in the horizontal
discuss shortly).
that injecting
money
is
But
in the vertical
fully offset
by the
depreciation in the exchange rate.
This leads to a second point.
If
a researcher looks at the net worth of credit constrained
monetary policy
firms during a crisis, he will always conclude that the reason for
tiveness
ineffec-
the deterioration in net worth triggered by the exchange rate depreciation
is
particular, he will
blame dollarization of
liabilities
(more on this below). But this
The main
the right conclusion in the vertical region.
credit constraint
behind the
an aggregate constraint not a microeconomic one; the exchange rate depreciation
is
—
is
in
not
crisis is
simply
the equilibrium response to such restriction.
Underinsurance and Monetary Policy
4
Of course
is
all
our claims so far are from the perspective of date
already given. But what
crisis
on date
decisions?
We show that
crisis at
date
1,
is
We
1 in
turn to this discussion next.
is
expected to contract money in the event of a vertical
actually causes the private sector to alter date
at date 1 with a smaller
w
.
when w n
the impact of anticipated monetary policy in the event of a
a central bank that
n
the 6-state,
By
supporting the date
the private sector to over-borrow at date
1
decisions so as to arrive
currency, the central
and under insure against the
0,
crisis.
bank causes
Conversely,
expansionary monetary policy gives the private sector incentives to insure against the
and reduce date
borrowing.
Private Sector Date
4.1
At date
much
0,
Decisions and
how much
the private sector decides
real investment to undertake.
€
{b, g}.
2,
contracts specify an
f", contingent on the date
Since the funds raised from this loan are used in date
budget constraint
foreign investors are risk neutral, the date
In the
e\
to borrow from foreign investors
The borrowing
a domestic firm and a repayment at date
u>
crisis
<?-state at
date
1, all
firms
make
Ak +
(w
+M
amount loaned
-
f
9
)
investment, and since
is,
+ M.
w-f
13
b
\
A
make
to
(aggregate) state
profits of,
In the 6-state, one-half of the firms are distressed and they
ak
1
and how
profits of,
while the other half are intact and
make
profits of,
Ak+(w-f
b
eb
+ M.
)—^
+
i\
1
Combining these expressions leads to the following problem
for a firm at
date
0,
PRIV:
maxfcjgjb
(1
- n){Ak +
+ a^k+(A + e b )^+(l +
+7rl^A
f 9 ,f
s.t.
b
c(k)<
Our
w- f 9 + M)
^)M
<w
{1+i .j{1+i . )
{*f
b
+
{l-*)f°)-
technical assumptions (see the appendix) guarantee that f 9
The former
holds as long as increasing investment in k at date
investment in international markets.
And, f b <
absorb the production shocks at date
1 is
toward investment at date
0.
It is
w
is
= w
more
as long as saving
more valuable than using
all
and f b
w.
profitable than
some resources
to
of those resources
apparent from the private program that e\
equilibrium price that influences the date
<
is
the only
Let us study this connection more
decision.
closely.
Since f 9
k.
At date
=
0,
w,
we simply need
to consider the tradeoff between increasing
b
f and reducing
building a marginally larger plant increases (expected) date 2 profits by,
(l-^A + TriU + a^J.
(9)
Building this larger plant requires the firm to raise an additional c'(k) at date
probability of a crisis
is
w, so in order to raise
an additional
c'(k) at
date
b
0,
f must
0.
The
rise by,
c'{k){\+il){\+i\)
^
IT
The
cost to the firm of raising
production shock. The
fall in
f
b
is
that there are fewer resources to absorb the date
expected profits due to having fewer resources
1
is,
A + e b \c'(k)(l + i* )(l + i*
2{i + q)J
1)
tt
which can be simplified
to:
m(i+i*Q )(^±£\.
The optimal
(io)
choice of k by the private sector equalizes (9) and (10).
parative statics, and the conclusion that follows from them: (1)
14
The
We
note two com-
benefit of building a
larger plant size
decreasing in
is
For both reasons, k
is
e\; (2)
The marginal
decreasing with respect to e
The investment/financing
decision at date
is
exchange
On
rate.
that there
The reason
reason
why
A
matters
e\
that
is
more depreciated
and more incentive to decrease
to
affected
output
by the value of the
meaning
,
international liquidity in the 6-state.
is less
ment/financing decisions.
This
/.
changes the terms of this invest-
it
e\ gives firms less incentive to increase
the reason that k
is
Consumption Maximizing
We now
is
for this is that the
the other hand, to build the plant, the firm has to raise f b
Intuitively, the
4.2
increasing in e\.
really to create a (real) "peso" plant at
and hence
collateral
is
.
the cost of having less "dollar" assets in the 6-state.
from k can only be used as domestic
cost of investing
b 9
is
decreasing with respect to
k
e\.
e\
ask what level of the exchange rate, e b
,
is
maximize (expected) aggregate consumption.
required in order for the date
We
show that
for e\
<
decisions
A, the private
sector generally over-borrows and over-invests relative to this benchmark.
Consider
first
the date
gate consumption.
e\
we
PRXV.
from
If
decisions of the private sector, (k, f 9 f
b
,
),
that maximize aggre-
program we simply substitute out the exchange rate of
To
arrive at this
we
substitute the market clearing condition for e\ from (7) into
PRXV
find that,
AGG:
max fcJ9>/(
(1
s.t.
f
,
9
- n)(Ak + w-P)+tt (^=£a +
J <w
b
c(k)<
benefit of increasing plant size
side,
(nf + (l-n)f9).
r)
< A,
this benefit
is
strictly lower
than the private sector's computation.
9
It is
A, the cost
not
e\,
per
se,
lies strictly
We
ij
the cost
+ i* )A.
above the private sector's computation.
that matters in this tradeoff, but ^-
exchange rate) times one plus the
obscured.
On
costs,
c'(k)(l
<
PRIV. The
very different here than in
- n)A + 7T- {A + a)
borrowing more to build this plant
For e\
is
is,
(1
For e b
^
TITI
b
tradeoff between increasing k versus f
The
4±°fc)
(the peso-rate).
Since
relax these assumptions later in the paper.
15
(i.e.
the expected future depreciation of the
we have
set e2
=
1
and
ij
=
0,
this point
is
We
conclude that,
if
< A,
e\
then the private sector's date
decisions involve over-
borrowing and overinvesting.
Intuitively, the cause of this date
over-investment
is
that the private sector undervalues
international liquidity in the 6-state. At the aggregate level, resources in the 6-state always
A
generate a return of
not having date
from date
1
to date
On
2.
resources as proportional to e\.
1
the other hand, firms see the cost of
As long
as e\
< A,
the private sector's
investment choices do not lead to maximization of date 2 expected aggregate consumption.
The
4.3
Effect of Anticipated
The previous
Monetary Policy
result highlights a basic feature of the environment. Relative to the
maximizing outcome, the private sector
Thus we have a
investing.
tempts to
it
benchmark to evaluate
clear
stabilize e\
.
More
exchange
policies that affect the
bank can
generally, since the central
affect e\
place at date
then k
1,
if
will
the private sector expects an increase in
be reduced at date
with
A
can reduce this over-borrowing problem through monetary policy.
argument establishes that
rise.
always biased towards over-borrowing and over-
In particular, the private sector's biases are exaggerated by a central bank that at-
rate.
M,
is
consumption-
M
its
choice of
straightforward
if
a
crisis
takes
and expected date 2 consumption
will
10
Dollarization of Liabilities
5
The
credit channel literature in emerging markets has highlighted the role played
by
dollar-
ization of domestic liabilities in magnifying the contraction. Indeed, even in our framework,
a depreciated exchange rate deteriorates balance sheets by lowering the value of domestic
But our framework
assets relative to liabilities.
ature:
qualifies the
While dollarization may have the balance-sheet
during mild crises (horizontal region),
At the firm
it
is
and the depreciated exchange rate
relevant constraint
is
The argument
That
clearing at date
This
is
is
1 it
is,
macroeconomic constraint
11
by contradiction.
must
Suppose that increasing
also cause e\ to rise. However, since k
can be distributional
with more dollarized
Firms are
worsen balance sheets. But the
in a vertical crisis, the country faces a
a contradiction. Therefore
n There
will
liabilities:
a shortage of aggregate international liquidity, not a firm-level balance
not a microeconomic one.
10
effect highlighted in that literature
one of dollarized
credit constrained,
sheet problem.
liter-
does not during severe crises (vertical region).
seem that the problem
level, it will
main conclusion from the
liabilities
A:
is
effects
i.e.
is
causes k to
rise.
Then from market
a decreasing function of e\, k must
fall.
M and aggregate consumption increasing M.
from those that don't receive M to those that do, or from those
decreasing in
-
M
is
in
to those with less liabilities - but these are likely to be subordinate to the
16
However, for the insurance reasons we discuss in the previous section, dollarization
not innocuous.
and
generally lead to private sector underinsurance. Let as discuss these
It will
dates
1
5.1
Dollarization during severe crises:
issues in turn.
we
In order to discuss domestic dollarization,
lend to firms at date
at date 1, given
0.
some
is
given
A
fallacy of aggregation
d
it is
Suppose that firms
pre-existing debt.
who
introduce a set of domestic consumers
For the purpose of this subsection,
with some dollar-denominated debt, b
investment
is
sufficient to start the analysis
arrive at the date 1 crisis-state
with domestic consumers. Thus, a distressed firm's
,
as,
ei
The
literature emphasizes the fact that
of local assets
ak and
M while
from (11) a depreciation reduces the dolar value
unchanged the value of dollarized
leaves
it
Moreover, since a monetary expansion depreciates the exchange rate
it
liabilities
may worsen
e\b
d
.
balance
sheets and, perversely, turn contractionary.
But, as we argue in Section
the main binding constraint
is
considerations of the effect of
important
the lack of aggregate international liquidity. Thus
M
for distributional issues
not for the aggregate.
analysis does not consider that under severe crisis
3, this
on
and hence on the domestic
ei,
(from more to
The exchange
rate reacts to
the aggregate level, distressed firms' investment
6k
new
'
(12)
.
~
in pesos,
+M
w + bd
n
W:
bv
ak
+
expressions,
we can
"
p
,
liabilities.
This ex-
when domestic
liabilities
M -W
wn
1
^From these two
M:
ak
stands for the equilibrium exchange rate with dollarized
denominated
At
limited external supply.
change rate expression needs to be contrasted with expression, e 1
are
to cause a
expression for the exchange rate as a function of
1
e1
and consumers) but
is still:
= 2w n
M_
where the
these
credit squeeze, are
monetary policy precisely
demand with the
credit squeeze that equilibrates investment
Replacing (12) into (11), yields a
less dollarized firms
all
consider the effect of monetary expansions on the
exchange rate in economies with different degree of dollarization:
<
tfW = t^tt* <7^ = Wei
aggregate constraint.
17
(is)
The
first
exchange
implication of this expression
A
rate.
bank that
central
is
is
that expanding
money always
depreciates the
concerned with the balance sheet position of firms
be inclined to contract money and support the exchange
(as in (11)) will
during a vertical
the important constraint
crisis,
is
But since
rate.
that of (12), the central bank will be
protecting the wrong margin.
The second
ized
economy
implication of (13)
be
will
economy. This
less
is
that during severe crises the exchange rate in a dollar-
responsive to a monetary expansion than that of the non-dollarized
a sort of market based fear of floating, as
is
siderations for a given
monetary
bank may chose to adopt when
cisely that a depreciation
policy,
results
liabilities
are dollarized.
The reason
needed to reduce investment demand to
easily generate the domestic credit squeeze
levels
compatible with the limited availability of international
5.2
Underinsurance
There
is
that
is
more
for this result is pre-
economy, and therefore can
in a dollarized
more
a
from equilibrium con-
and not from any additional caution that the central
more contractionary
is
it
liquidity.
effect of dollarization
subtle effect of dollarization in our framework.
A
central
bank
at date 1
concerned with preserving the balance sheets of firms may, in the vertical region,
choose to support the exchange rate. But boosting e\ causes firms to undervalue insuring
against the 6-state and set f
b
higher. This follows from the underinsurance results of the
previous section. In this sense, the negative effect of central bank support of the exchange
rate
is
that external
liabilities will
on the denomination of
(see Caballero
That
is,
become
less contingent. In
liabilities, this effect will
lead to excessive dollarization of liabilities
and Krishnamurthy, 2003a). 12
while
we challenge the standard view that
factor during severe crises,
negative effects
it
we argue that
dollarization of liabilities
an important factor behind
it is
crises
is
a key
due to the
has on private agents external insurance decisions.
Dollarization of External Liabilities
5.3
Before concluding this section we should point out that there
which
is
that of the dollarization of external
emerging market external debt
is
a model with richer options
denominated
in a country's
is
liabilities.
is
a related but distinct issue
This refers to the fact that most
not denominated in the issuing country's currency.
own
is
supported by Bleakley and Cowan (2002)
exchange rates the match between the denomination of
economies with
flexible
exchange
debt
currency, then the debt will be effectively contingent on
aggregate outcomes. In practice, of course, unlike our theoretical assumptions, there
This implication
If
rates.
18
liabilities
who
find that in
is
very
economies with fixed
and that of revenues
is
weaker that
in
little
or
contingent debt issued by emerging markets.
would borrow abroad
own
in their
The
developing economies could
if
currency, they would effectively obtain insurance
from foreigners against events that depreciate
external financial constraints.
Thus,
their currency, such as the tightening of
"original sin" literature in, e.g.,
Hausmann
highlight reluctance on the supply side of that market, while Caballero
(2003a) describe
demand
and Krishnamurthy
behind the lack of contingency of external debt.
side problems,
In our current analysis, external dollarization simply means that there
in liabilities so that
f
b
is
higher and closer to f 9
international liquidity in the crisis-state. It
liability dollarization is reinforced
liabilities as
is
.
As a
result, the
is less
country
contingency
will
have
less
worth also noting that the problem of external
by the underinsurance
effects of dollarization of
domestic
discussed above.
Adding external
liability dollarization to
and makes the supply
our problem
of international funds at date
substantive changes in our analysis.
We
1
conceptually straightforward
is
effectively diagonal, but
discuss this extension in Section
nothing
7.
Policy Constraints and Policy Options
6
In
et al (2001)
many
rate.
instances, a central
bank
faces constraints that force
Dollarization of domestic liabilities
central
bank determines
problems
is
this
another one.
is
We
is
it
to support the exchange
one such instance (whether the market or the
not important for our argument here). Inflation credibility
begin this section with a discussion of inflation credibility,
and then turn to alternative policy instruments when monetary policy
pro-cyclical or less counter-cyclical than desirable.
costs of losing
monetary policy during severe
latter, as in the
There
is
constrained to be
is
an important difference
crises vis-a-vis
standard Mundell-Fleming context, the cost
in the
doing so in mild ones. In the
is
that the central bank loses
a countercyclical instrument to smooth fluctuations. In the former, on the other hand, the
cost
is
mainly an "insurance" one.
The
central
bank
loses a relatively
cheap instrument
to induce the private sector to take adequate precautions against crises.
framework, the natural response to the
countercyclical policy instrument
the natural response
is
loss of
monetary policy
(e.g., fiscal policy).
is
In the standard
to search for another
In the vertical crisis environment,
to search for an alternative insurance
mechanism.
We
discuss such
alternatives in the second part of this section.
6.1
Limited Inflation Credibility
Limited inflation credibility has two
effects in
our environment. First, a central bank with
a history of high inflation cannot afford the inflationary consequences of an exchange rate
19
depreciation during a
crisis.
This
is
common explanation for the
a
of central banks in emerging economies (see, e.g., Calvo
we have outlined above, the
the reasons
during a vertical
crisis will
thereby worsen the
A
and Reinheart
anticipation of a contractionary
second problem caused by limited inflation credibility
have a limited
crisis will
However a central bank with limited
on
little effect
real
is
effect
regardless of the
on
is
=
monetary policy adopted
1.
all
money balances during a
To an
crisis,
some of
at date
1.
our derivations up to
now we
But suppose that whenever there
expected to pass through into e 2
it is
the date 2 price level)
(i.e.
In a vertical crisis a distressed firm
money
intact firm, the
is
sells
is
,
d
i-i
I
= M/e 2 n+ ak
w
.
M units of money and
less
on lending international
w
n
against
than the ak
liquidity as,
international liquidity
is
just |^
—
Since this return must equal the return to lending against the domestic collateral of ak,
find that the interest parity condition
of
aA:
1.
money with
to an intact firm of purchasing
a
is
an increasing function of
worth \/eo proportionately
of domestic collateral. Thus, let us define the return
The return
with
return from hoarding
domestic collateral to raise funds for salvaging production. The intact firms lend
these assets.
is,
long run inflation (price level) credibility, so that e 2 remains at one
concrete, suppose that e 2
with e2(l)
4r,
That
real balances.
money balances and thereby on the expected
depreciation during a
To be
that a central bank that
inflation credibility will create inflation
international liquidity In order to see this, note that in
have assumed there
monetary policy
crisis.
our environment a central bank does better by increasing real
crisis.
(2003)). However, for
exacerbate the private sector's underinsurance problem and
expands monetary policy in a
in
apparent "fear of floating"
1.
we
is:
e2
For the case where e 2
fc,
is
=
we previously showed that the date
1,
a decreasing function of
e\.
We
level of investment,
argued that supporting the currency at date
1
had
over-investment problem of the private sector.
the consequence of exacerbating the date
Likewise, allowing e\ to depreciate through an expansionary monetary policy, reduced the
over-investment problem.
When
k
is
e2
is
not fixed at one, a similar argument to that of Section 4.1 establishes that
a decreasing function of S1
e2
.
But
since,
'
ei
e2
_ M/e 2 +
wn
20
ak
the important term for policy purposes
An
M/ez).
is
the term involving real
expansionary monetary policy raises real balances only
money
balances
(i.e.
to,
M < M.
—
i,
e2
Thus, a central bank with limited inflation credibility will have to resort to additional (and
possibly very costly) mechanisms to induce adequate private sector insurance decisions.
We
turn to this discussion next.
Insurance Substitutes
6.2
The problem
created by being unable to raise real balances
A—
remains high. Thus the return to hoarding international liquidity until date
(1
+
if),
is
that the social-private spread,
remains undervalued, and private insurance decisions are distorted.
1
Section
4,
As we argued
in
these considerations are unique to the vertical environment, because the insurance
problem only
arises
the aggregate international liquidity constraint binds.
if
There are two obvious ex-ante policy measures that can deal with the underinsurance
problem: taxation of capital inflows during normal times (date
requirements at date
0.
We now
and international liquidity
0),
characterize the relationship between an ex-ante tax and
H
The
first
c'(k
whereas
1
for
AGG
order condition in
AGG )(l +
i*
Q
is,
)A =
(1
-
tt)A
the private sector the condition sets
(9)
(A +
a)
equal to (10) (with e\ replaced by
+ if).
private and consumption-maximizing incentives
Aligning the date
ing a tax/transfer policy. Suppose that the central
is
+ n±
returned to firms in a lump
sum
fashion.
Then
bank
the
levies a
first
is
a matter of choos-
tax t per unit of
k,
which
order condition for the private
sector becomes:
c'(k)(l
+ io)
A+
+i "
l
2
=
(1
- n)A + *\(a + a-^-d
2 y
+ if
i
Choosing r to align the private and central-bank incentives yields that
rium
level of if T
,
is
any equilib-
the optimal tax solves:
r(il)
The tax
for
=
\
increasing in
follow counter-cyclical
[r^r + c'(k AGG )(l + iS)) (A A—
monetary
(1
+ if T ).
policy,
(1
+ ifiT )).
Thus, economies where the central bank cannot
and therefore
21
A — (1 + if
)
remains high,
may need
to rely
on capital controls to correct the underinsurance problem. Note that the same result
The tax
could be achieved via a contingent liquidity requirement.
solution gives the private
sector incentives to choose the efficient k, thus resulting in the efficient
w — fb
.
Alternatively,
the central bank could mandate directly that each firm preserve international liquidity for
the date
1
crisis-state, so that
In practice, taxes
the
come with
efficient level of
their
costs of enforcement, evasion, etc.
own
w—
f
b
is realized.
sets of distortions: deadweight costs of taxation,
However, the important point to recognize
the vertical environment, the cost of losing monetary policy
demand
aggregate
is
is
is
that, in
not being unable to
manage
or the extent of the domestic credit squeeze at date
the underinsurance by the private sector at date
enforce capital controls
may be
Rather, the cost
1.
In this sense, the cost of having to
0.
seen as a direct cost of losing monetary policy.
Robustness: Peso Rates and Diagonal Supply
7
In this section
most
we show that our main conclusions
The absence
stylized features of our model:
are robust to modifications in two of the
monetary
of a
friction
and the absolute
price inelasticity of the supply of funds during crises.
Peso Rates and Interest Parity Condition
7.1
Up
to
now we have
analysis,
simplified our analysis
and therefore have
set the
by removing
we sketch a more standard model
money, in which money
transaction service.
Our substantive
This simplifies
but leaves us with an unusual model of money. Here
1,
of
from the
frictions
domestic peso interest rate to zero.
exchange rate determination at date
13
monetary
all
results
is
special because
it
provides a
remain unchanged, although now we recover
the standard exchange rate effect via the interest parity condition in addition to the excess
sensitivity result
we discussed
Let us return to the
in the previous sections.
full inflation credibility
case
=
[e-i
1)
and
let
us focus on the vertical
region and the bad aggregate state.
At the end
of date
per capita. Each bond
1,
is
the government has
liabilities
redeemed at date 2
for
B
bonds and
M units of money
one unit of money, and the government
credible in ensuring that the price level at date 2
is
of
is
1.
At date
1,
money
is
the only
domestically liquid asset. Neither claims against ak (corporate bonds) nor the government
bonds are
1
liquid.
Thus, at date
1
distressed firms sell
See Diamond and Rajan (2001, 2003) and Lorenzoni (2001)
based frameworks.
22
M units of money to the intact firms
for alternative
models of money
in liquidity-
in
exchange for
wn
units of international liquidity, which
—
Cl
It is instructive to
wn
means that
.
go through the steps that take us to this equation in order to disentangle
money
the mechanisms through which
affects the
exchange rate during severe
crises in this
economy.
We
introduce a date
_
1
market operations. At date
and the
in order to study peso interest rates
_
effects of
open
the aggregate state of the world has been realized, but the
1
identity of agents receiving the shock (distressed or intact) has not. Because of the latter,
agents are
1~.
all identical at
At date 1~ both the bond market and the money market
date
_
1
,
the government has outstanding
B
bonds and
— B)
open market operation to purchase (B
bonds
are
open to
all
agents. Entering
M money. The government does an
for
(M — M)
money.
Let us consider the relative asset returns on bonds, money, and international liquidity
at date 1~
One bond
1~, where i\
yields
is
1/ep
.
if
the agent
money from
date
which means that the net return
money and bonds
l
unit of
money
in
to date 2
also
sold for international liquidity at a
A
at date 2.
(A/e^ +
Thus the expected
As
1).
before,
A>
e 1 ~,
immediately follow that indifference between
.
(14)
)-
exchange market into one unit
in the foreign
One
unit of international liquidity either can
or sold to a distressed firm at date
1
(l
is
^(A +
e!-).
Thus the
1.
The expected
benefit
interest parity condition
+ iP) ei _ = I(A + ei -).
In the standard interest parity condition, the right
offset
0.5
H(£ +1
can be converted
production at date
must be
is
sold at date 1 to finance
requires that,
of the unit of international liquidity
in i\
is
liquidity yields
of international liquidity, at the price of ej-
be used
money can be
Money
positive. It
is
i+i
One
unit of
distressed.
is
Each unit of international
return on holding
holding
One
the peso interest rate.
the liquidity shock
price of
one unit of money at date 2 and costs 1/(1 +if) units of money at date
by a depreciation
hand
(15)
side of (15)
in ei- to keep the left
is,
hand
is
fixed
and a reduction
side constant as well.
But
the depreciation in our vertical environment raises the return to the dollar-lender in the
right
hand
side of (15),
which means that the
by a further depreciation
in e x -
.
The
left
hand
side
must
rise,
and
latter is the excess sensitivity result
23
this
is
achieved
we have discussed
and
it
results
from the relaxation of the domestic credit squeeze caused by the monetary
expansion.
Finally,
combining (14) and (15) we can solve
rate as a function of
monetary
latter
the expression
is
through which money
we
peso interest rate and the exchange
policy:
e i-
The
for the
—
M
—
w
n
started this section with, and
affects the
exchange
Relative to our earlier model,
it is
summarizes the two channels
it
and domestic
rate: interest parity
credit squeeze.
the former (and standard) channel that
neither channel can change the fact that international liquidity
is
fixed at date
is
new. But
1
and hence
output cannot be affected by monetary policy:
c V = (^_+^y + w n A + y c
And
this
for date
decisions,
it is
only the second channel that matters, which
and the model without monetary
7.2
"Diagonal" Supply
Let us
now
inelastic
is,
where
the supply of international funds
s(l)
is
pure vertical model,
consumption
at date
2.
The diagonal supply
the export sector
is
ei
is
This
CV
1
1, let
us consider instead a supply curve
=
s'(-)
1,
>
0.
"diagonal" as opposed to vertical. 14
M =W
TTW
ak
total
1.
now:
1 is
-
in the
=
,
Equilibrium at date
As
to
friction.
supply of funds at date
w n s(ei)
That
common
return to the model without monetary frictions, and continue to assume e2
But rather than an
of the form
is
+
_
increasing in
.
.
s(e 1 ).
M
.
However, consider the expression
for
is,
={^-)k + wn s(e
1
)A +
also captures the idea that depreciating the
an important part of international
collateral,
c
y
,
exchange rate increases exports and,
thereby expands supply. Christiano et
if
al
(2004) offer a related perspective on diagonal supply. In their model, imperfect liquidity substitution stems
from imperfect input-substitution, and from the
The
fact that different inputs are paid in different currencies.
"diagonal" aspect of their model arises from the (limited) possibility of substituting tradables and
nontradables inputs.
24
which
now an
is
increasing
increasing function of e\ through the s(-) function.
M does have a contemporaneous
Thus, from a date
model and the
1
effect
perspective, the diagonal
As
vertical model.
on
Cv
This implies that
.
model has elements of both the horizontal
in the horizontal model, there
money
channel/credit squeeze channel through which expanding
an aggregate demand
is
increases aggregate con-
sumption. As in the vertical model, the money expansion depreciates the exchange rate
beyond the standard
Now,
let
interest parity effect.
us shift back to date
0.
At
this date, the firm
resources and increasing k, the size of the plant.
is
increasing in e\.
date
1,
at
long as ei
A
As
shadow
before, the
higher expected e\ induces a firm to save
which point these resources can always be
< A,
contemplates borrowing some
some
cost of the resources
dollar resources until
lent to return e\
—
1.
Moreover, as
the private sector's decisions will not be consumption maximizing. Thus,
as in the vertical model, there
model. Expanding
M
is
an insurance channel
at date 1 depreciates e±.
The
for
monetary policy
anticipation of such depreciation
and increase
the private sector reduce investment at date
in the diagonal
makes
insurance against the date
its
1
shock.
8
Final
Remarks
The past decade has witnessed
several episodes in which the
sudden reversals of capital
inflows triggers an emerging markets' crisis. These events have placed the risk of "sudden
stops" as a central macroeconomic concern for emerging market economies.
contribution of our paper
is
to illustrate
of firms during a sudden stop
crisis,
how monetary
policy affects financial constraints
and to highlight an insurance
Domestic monetary expansion relaxes individual
The main
effect of
monetary
policy.
but
is
unable to relax
the aggregate financial constraint faced by these economies during crises
-
it is
financial constraints
kind of liquidity for such a purpose. However, while monetary policy
is
the wrong
largely futile once
the sudden stop has taken place, the anticipation of a particular reaction by the authority
is
important
for private sector
insurance decisions.
A
flexible
exchange rate system coupled
with a countercyclical monetary policy, provides better insurance incentives to the private
sector than a policy of defending the exchange rate during sudden stops.
Our insurance mechanism
insurance argument
implicit
is
(see, e.g.,
commitment by a
local
distinct (or in addition to) the standard moral-hazard/free-
Dooley 2000). In the
latter
argument, crises result from an
government or foreign institution to transfer resources to the
imprudent borrower; a fixed exchange rate transfers reserves
crises) to the dollar-borrower or peso-investor. In
25
at
sub-market value (during
our model, there
is
no direct transfer from
The monetary
the government.
contraction implicit in the defense of the exchange rate
lowers the domestic liquidity that borrowers can offer to dollar-lenders during the sudden
stop,
is,
and
in so
doing
lowers the effective return from
it
since domestic asset markets are illiquid
so, it affects
local dollar lending.
and segmented during
affects the allocation of the (dollar) return of
By doing
new
That
monetary policy
crises,
investment between lenders and borrowers.
the relative incentives to be one or the other, and thus the incentive
to hoard international liquidity for crises.
More
for
generally,
monetary policy and
its
constraints have (unintended) consequences
the private sector's insurance decisions with respect to sudden stops.
Policies that
exacerbate the domestic credit squeeze during sudden stops, or constraints that limit the
imprudent private
authorities' ability to relax domestic financial constraints, lead to socially
sector actions.
Our emphasis on
ernment
policies.
dual-liquidity
and insurance has relevance
for evaluation of other gov-
Pro-actively managing international reserves
may
result in benefits in
our framework (see Caballero and Krishnamurthy 2004). Since international reserves are a
form of international liquidity and the private sector carries too
into crises, the central
bank has a
role to play
little
by carrying reserves
international liquidity
in place of the private
sector. Injecting reserves during a crisis relaxes the international financial constraint faced
by the private sector and
but the
those
effect
we
policy
stabilizes the
exchange
rate.
on the private sector's own insurance incentives
discussed in the context of monetary policy. Reserves
become complementary policy
bank ought
tools in this context.
to sterilize the forex intervention,
course, the insurance dimension
we
is
is
beneficial
not, for reasons akin to
management and monetary
At the very
least,
and possibly go beyond that
expansion in order to offset the perverse incentive
Of
Relaxing the constraint
the central
in the
monetary
effect of the reserves injection.
highlight in our analysis
is
not a substitute for
conventional inflation credibility concerns. Quite the contrary, without inflation credibility
the central bank will be unable to
may
during sudden stops, and
ipation of this behavior
is
let
the exchange rate float and expand monetary policy
well be forced into tightening
monetary
very costly in our setting, precisely because
policy.
it
The
antic-
exacerbates the
private sectors' underinsurance problem. In other words, our insurance consideration raises
the value of achieving medium-term inflation credibility.
Our
stylized
model
is
subject to
many
caveats.
One worth mentioning
the lack of true dynamics. In reality, crises build up, going
in
first
in concluding
is
through a horizontal phase
which domestic financial conditions tighten and external borrowing becomes gradually
more expensive, then
falling into
a sharp vertical sudden-stop phase.
26
A
central question
for
policymakers in this context
of the crisis,
when supply
is still
is
how
to conduct
horizontal but there
At
to a binding international liquidity constraint.
will destroy financially
event.
We
If
a concern that events
this stage, tightening
may
But
if
monetary policy
constrained projects but save international liquidity for the vertical
is
credible, then there
commitment
the
we have
is
is little
need to tighten during the horizontal
not credible or feasible, then the appropriate response
to tighten during the early phase in order to protect international liquidity, very
as taxing capital flows at date
constraints on
crises.
in our simplified
Finally, while
much
model when there were
In fact, the costs in terms of the additional
imposed on the domestic private sector
to the costs of the ex-ante
may be
was advisable
monetary policy during
financial distress
lead
the commitment to an aggressive countercyclical monetary policy in
case of a vertical event
is
is
at the early stages
conjecture that this trade-off can be analyzed in terms similar to those
used throughout:
phase.
monetary policy
are, to
a large extent, comparable
measures we already discussed.
our analysis has focused on emerging markets, the underlying structure
a starting point for other applications.
segment financial markets and create
liquidity
Our model
premia on
illustrates
assets.
It
how
a bottleneck
may
shows how monetary
by these bottlenecks. There are many other
policy affects and
is
as liquidity traps
and post-bubble-collapses, where similar ingredients appear worthy of
consideration.
We
affected
are currently exploring these applications.
27
scenarios, such
References
[1]
Aghion, Philippe, Philippe Bacchetta, and Abhijit Banerjee, "Currency Crises and
Monetary Policy with Credit Constraints, mimeo Harvard 2000.
[2]
Bernanke, B. and M. Gertler, "Inside the Black Box: The Credit Channel of Monetary
Policy Transmission," Journal of
[3]
Bernanke, B., M. Gertler, and
tive Business Cycle
Economic Perspectives
S. Gilchrist,
9, Fall,
pp. 27-48, 1995.
"The Financial Accelerator
in a Quantita-
Framework," Handbook of Macroeconomics, Taylor and Woodford,
eds., 1999.
[4]
Bleakley, H.
and K. Cowan., "Corporate Dollar Debt and Depreciations: Much Ado
About Nothing?"
[5]
02-05, Boston
FED,
2002.
"Why do Emerging Mar-
Broner, Fernando, Guido Lorenzoni, and Sergio Schmukler,
kets
[6]
Wp
Borrow Short term?" mimeo, July 2003.
Bulow, Jeremy and Rogoff, Kenneth, "A Constant Recontracting Model of Sovereign
Debt", Journal of Political Economy 97-1, February 1989, pp. 155-78.
[7]
Caballero, R.J. and A. Krishnamurthy, "International and Domestic Collateral Constraints in a
Model
of
Emerging Market
Crises," Journal of
Monetary Economics
48,
513-548, 2001.
[8]
Caballero, R.J. and A. Krishnamurthy, "Excessive Dollar Debt: Underinsurance and
Domestic Financial Underdevelopment," Journal of Finance. April, 55(2), pp. 867-893,
2003.
[9]
Caballero, R.J.
coming
[10]
and A. Krishnamurthy,
in Inflation Targeting, ed.
Calvo, G.A., and
CM.
"Inflation Targeting
and Sudden Stops,"
forth-
by Michael Woodford, 2004.
Reinhart, "Fear of Floating," Quarterly Journal of Economics,
2003.
[11]
Cespedes, L.F., R. Chang, and A. Velasco, "Balance Sheets and Exchange Rate Policy,"
NBER Working
[12]
Paper No. W7840, August 2000.
Christiano, L.J., C. Gust,
and
J.
Roldos, "Monetary Policy in a Financial Crisis,"
forthcoming in Journal of Economic Theory, 2004.
[13]
Diamond,
DW.
and R.G. Rajan,
"Liquidity
U.Chicago mimeo, June 2001.
28
Shortages
and Financial
Crises,"
[14]
Diamond, D.W. and R.G. Rajan, "Money
a Theory of Banking,"
in
NBER Wp #
10070, October 2003.
[15]
Dooley, M., "Responses to Volatile Capital Flows: Controls, Asset-Liability Manage-
ment and Architecture," World Bank Conference on Capital Flows, Financial Crises
and
[16]
[CHANGE FOR
Policies, April 1999.
2000]
Dornbusch, R. "Expectations and Exchange Rate Dynamics," Journal of Political
Economy
84,
December, pp. 1176-76, 1976.
Dornbusch, R. "Fewer Monies, Better Monies,"
[18]
Eaton,
NBER Wp
[17]
tion,"
[19]
J.,
and M. Gersovitz, "Debt with Potential Repudiation: Theory and Estima-
Review of Economic Studies 48, pp. 289-309. 1981.
Edwards,
Furman,
and E. Levy-Yeyati, "Flexible Exchange Rates
S.
NBER Wp
[20]
J.
and
E. Stiglitz,
J.
Gertler, M., S. Gilchrist
Hausmann,
"Economic
Economic
Shock Absorbers,"
Crises:
Activity
2,
Evidence and Insights from East
1998.
and F.M.Natalucci, "External Constraints on Monetary Policy
and the Financial Accelerator,"
[22]
as
#9867, June 2003.
Asia," Brookings Papers of
[21]
#8324, June 2001.
R., U. Panizza
NYU
and E.
working paper, Feb 2001.
Stein,
"Why
do countries
float the
way they
float?"
Journal of Development Economics December 2001.
[23]
Holmstrom, B. and
Political
[24]
Kiyotaki, N. and
(2),
[25]
Economy
"Private and Public Supply of Liquidity," Journal of
106-1, February 1998, pp. 1-40.
J.
Moore
(1997): "Credit Cyles," Journal of Political
Economy, 105
211-248.
Lorenzoni, G., "Interest Rate Stabilization and Monetary Control:
Mimeo, Princeton
[26]
J. Tirole,
A
Reconciliation."
University, 2001.
Woodford, M., "Public Debt as Private Liquidity," American Economic Review, Papers
and Proceedings
80,
May
1990, pp. 382-88.
29
A
Appendix
A.l
The
Financing Assumptions
financial frictions of the
Assumption
model are embodied
in the following
1 (International Collateral)
Foreigners lend to domestic firms only against the backing of w.
M
two assumptions:
Domestic agents lend against w,
and ak.
Assumption
A domestic
2
(Domestic Collateral)
lender can only be sure that a firm will produce ak units of goods at date
production based on physical reinvestment at date
One
last
assumption
is
1 is
Any
excess
neither observable nor verifiable.
insurance arrangments that transfer resources
required to rule out date
from distressed firms to intact
2.
firms.
Assumption 3 (Non-observability of Production Shock)
The production shock
The
at date 1 is idiosyncratic.
identity of firms receiving the shock is private
information.
The mechanism
design problem associated with these financing and informational constraints
corresponds to the one in
AGG.
There
is
also a
around the private information constraint, but
A. 2
banking arrangement that in principle
this
very
is
may
get
fragile.
Technical Assumptions
Consider next the technical assumptions on parameters that we have used. The program
in
AGG
is,
AGG:
max /SJ
fc
(1
- n)(Ak + w -
f9
J <w
6
,
s.t.
c(k)<
First,
we
require that
w =
f
f
s
)
+
n
(f=£ A +
^i-fc)
b
9
{1+l .J(1+ll) (irf
in this
b
+ (l-n)P).
program, or that the return to investing domestically
exceeds that of investing abroad:
Assumption 4 (High Investment Return)
<
2
"
*) A
+
^^'
( (1
+
W
+ .1) )
(1
+ ?5)(1 + %l)
-
Second, we require that the solution features some insurance against the 6-state, so that f b
30
<
w.
Assumption
5
(High Return to Insuring)
<(iHk)) ,1+ « 4 ^
These
We
last
1
-''',+
A
two assumption can be jointly met by choosing
require that equilibrium, with no central
bank
'x-
large
enough
intervention, places us in the vertical region,
or,
1
The
first
+ i$ < e x < A.
PRIV
order condition for the program in
+ il)^1 =
c'(fc)(l
Denote the solution to
this equation as
and the smallest value when
=
e\
- *)A + i\
(1
Then the
fc"(ei).
A. Using
is,
U + a^J
largest value of k
is
.
attained
when
e\
=
l+i{,
knowledge as well as the market clearing condition
this
leads to:
Assumption
6 (Equilibrium in Vertical Region)
A
tt(M + ak{il))
w-(l+i' ){l + il)c{k(il))
7r(Mafc(A
-
1
1))
>
w-(l + i${l + il)c(k{A-l))
Finally,
we have
implicitly
+ h*
1
assumed that the maximum reinvestment constraint does not bind
in the vertical equilibrium:
w-f
k
2
This can be rewritten
1
w-
c(k)(l
+ iJ
t(1
+
%)(!
_
+ ak)
ir(M
w- (1 + i* ){l + q)c(k)
to:
Assumption 7 (Reinvestment constraint does not bind
1
A. 3
We
+ q)
+ *I)
as,
61
This leads
b
in
V)
+ ij
Dollarization of Domestic Liabilities
sketch an extension to the model of Section 7 in order to address dollarization of
mentioned
in Section 6.2, this is
one of the primary reasons that policy-makers give
to lower interest rates during a crisis.
31
for
liabilities.
As
being unable
Suppose that firm have debts of
D
dollars that have to be settled at date l
do not
.
These debts are
owed
to domestic consumers, so that they
Then
the total peso net worth of agents at date 1~, before any open market operations,
affect
the international liquidity of the country.
NW= YM_+M + ^-De
Now suppose
l
-
that the government does an open market operation where
money. Since this
is
transaction
is
done at market
The open market operation has two
?/,
effects.
J1
prices, the net
First
it
is,
(16)
it
purchases bonds with
worth remains as
in (16).
lowers i\ and thereby raises -s^-
Date Due
Lib-26-67
3 9080 02618 0155