Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium Member Libraries
http://www.archive.org/details/internationalliq00caba2
31
OEWEY
[415
Technology
Department of Economics
Working Paper Series
Massachusetts
Institute of
INTERNATIONAL LIQUIDITY ILLUSION:
ON
THE RISKS OF STERILIZATION
Ricardo
J.
Caballero
Arvind Krishnamurthy
Working Paper 01 -06
February 2001
Room
E52-251
50 Memorial Drive
Cambridge, MA 02142
paper can be downloaded without charge from the
Social Science Research Network Paper Collection at
This
http://papers.ssrn.com/paper.taf ?abstract id=260577
Technology
Department of Economics
Working Paper Series
Massachusetts
Institute of
INTERNATIONAL LIQUIDITY ILLUSION:
ON
THE RISKS OF STERILIZATION
Ricardo
J.
Caballero
Arvind Krishnamurthy
Working Paper 01 -06
February 2001
Room
E52-251
50 Memorial Drive
Cambridge, MA 02142
paper can be downloaded without charge from the
Social Science Research Network Paper Collection at
This
http://papers.ssrn.com/paper.taf ?abstract id=260577
Introduction
1
During the booms that precede
limit capital flows
and
sterilize capital inflows
requirements.
large.
These
their expansionary consequences.
In particular, they attempt to
through an open market sale of domestic bonds or increased reserve
sterilized interventions (henceforth, sterilizations)
During the early 1990s
over three quarters of
GDP
emerging economies, policymakers struggle to
crises in
its
in Chile, for example, the
large capital inflow
can be extremely
exchange intervention meant that
— amounting to around seven percent of
per year -- went to international reserves accumulation at the central bank.
The
monetary
sterilization of this intervention increased the ratio of international reserves to
base from 3.5 around 1990 to over 6.0 by 1993. This pattern was repeated in
its
many emerging
economies during the early 1990's, when capital flows to the developing world surged. 1
Despite the fact that sterilization
is
flows, there is
widespread recognition that
range from
ineffectiveness to
its
common
the most
policy response to capital in-
not free of complications.
it is
The concerns
an outright backfiring, where the policy may actually
fuel
term ones, and lead to an expansion rather than the desired
capital inflows, especially short
contraction in aggregate demand.
The widespread and nearly automatic explanation
Fleming (MF)
rate target
these outcomes follows the Mundell-
with reference to a scenario where the authority has some exchange
logic,
and there
is
extensive capital mobility.
point out that this explanation
understand ineffectiveness and
is
its
flawed,
Our
starting point in this paper
and that an alternative framework
(MF) framework (Mundell
is
fixed
and the country has
is
fully
and the
cost of capital
While the
Calvo
et al.
is
full
unchanged by the central bank's
MF
logic
can potentially explain
(1993) p. 146 write:
was the most widely and
countries in our sample."
Uncovered
interest parity holds
determined by the international interest
why
And
and
rate, so that firms'
action.
policy
"Sterilized intervention has
present episode of capital inflows in Latin America."
ization
sector as capital inflows exactly offset the
interest rate remains unchanged.
is
which the exchange
access to international financial markets. In this case,
undone by the private
the domestic interest rate
1
needed to
depends on exchange rate systems and the degree of integration to international
the sterilization
sterilization
to
1962), the effectiveness of
financial markets, with the worst case for sterilization being that in
rate
is
is
extreme form, backfiring.
In the textbook Mundell- Fleming
sterilization
for
may be
ineffective, its
mecha-
been the most popular response to the
so does the World
Bank
(1997), p. 181: "Steril-
intensively used policy response to the arrival of capital inflows
The sample included 22 emerging economies.
among
the
.
nism
is
not consistent with the prevailing facts in emerging economies.
ineffective
because the central bank
does raise domestic peso rates.
framework
—
that
link;
is
MF,
policy
is
unable to tighten monetary conditions and hence to
domestic peso rates. But one of the established facts
raise
MF
is
In
Indeed the problem
that in practice sterilization
is
the next
lies in
between
in the lack of connection
— and
alien to the
rises in this rate
and the
response of asset prices and real investment.
Even more puzzling
in the context of the traditional
model
"backfiring." Since the rise in domestic interest rates triggered
is
by
the related concern of
sterilization
nied by an increase in capital inflows, especially short term ones, policymakers
it is
these flows that fuel
what
is
and exchange rate overvaluation,
1993, Williamson 1995,
result does not follow
if it
is
accompa-
warn that
perceived as an excessive expansion in aggregate
may be
sterilization
counterproductive
Corbo and Hernandez 1996, Massad 1997)
from the standard
MF
logic,
where
3
(e.g.
as
demand
Calvo
et al.
Again, this backfiring
at worst the policy
is
ineffective
cannot affect monetary aggregates.
In this paper
we argue
that the standard analysis omits two aspects of emerging markets
which are central to understanding these unusual outcomes.
First, in
economies where
external crises are a recurrent possibility, dollars are the scarce input into production and
as such the question to ask
is
Thus, we argue that a
firms.
whether policy
full
affects the expected dollar cost of capital for
analysis of sterilization
must look
cost of capital, not just at the current peso interest rate.
financial
at the effective dollar
Second, in economies where
markets are underdeveloped, the outcomes of policies that rely heavily on these
markets can be quite different from their developed economy counterparts.
if
the instruments used in the intervention are
by the
sterilization.
Most importantly,
show that
the domestic dollar cost of capital,
illiquid,
which considers the whole path of expected future
We
rates,
may be
lowered rather than raised
this dollar cost of capital reduction
does not
reflect
a
true increase in the country's international liquidity during the external crisis and reversal,
as
would be the case with a successful
2
For summaries of the evidence
3
In his statement
sterilization,
see, e.g., Agenor (2000) p. 223 and Montiel (1996) p. 212.
on behalf of the Latin American Governors of the Fund at the joint World Bank
annual meeting of the Board of Governors held
return on capital in a
but just a decline in domestic private
booming economy
in
Hong Kong
(1997),
Massad
writes:
if
the
monetary authority safeguards domestic
equilibrium. This, in turn, could provide a further incentive for capital inflows.
The probable outcome
be continued appreciation of the local currency, the resulting risk of widening the current account
and the greater danger that these capital flows
4,
our emphasis).
IMF
"The high rate of
attracts large inflows of external resources... Capital flows stimulate
domestic demand and could push up domestic interest rates
(page
-
will
will
deficit,
be reversed, should some negative external shock occur."
—
liquidity.
The impact
of the latter
which
liquidity illusion"
on
fosters rather
relative asset prices creates a sort of "international
than depresses external indebtedness and investment
in non-tradeables.
There are three central ingredients behind our analysis and
results:
a)
a significant
probability of an external crisis in the near future; b) illiquid or underdeveloped domestic
and
financial markets;
tressed firms, as
is
c)
the sterilization removes domestic liquidity to potentially dis-
when the instruments used
the case
in the sterilization are not fully
liquid.
The
first
ingredient implies that there are times
when the uncovered
interest parity
condition does not hold, while the second one says that during these times distressed firms
may be
unable to fully pledge the value of their projects to potential financiers (those
with some access to international markets).
mestic loans during crises
—or
In this context, the expected return on do-
(henceforth) the dollar cost-of-capital for domestic firms
above the international interest rate but below the marginal product of investment.
lies
Thus
cost-of-capital during crises
uidity (or collateral)
is
determined by the relative availability of domestic
and international
liquidity (or collateral). In other words, while
liqit is
the international financial constraint that determines the spread between domestic marginal product
and international
interest rates,
it is
the domestic financial constraints that
determine whether borrowers or lenders earn the spread generated by the marginal dollar.
The
reversal of a sterilization during a crisis
by either raising the available international
international spread) or by lowering
its
and
it is
is
It is
in this context that the third ingredient
bank removes domestic
backfire.
The former
that of an unsuccessful one.
in sterilization are illiquid, in
A
similar
outcome
is
(i.e.,
when the
by lowering the
larger share
a case of a successful sterilization,
above plays
we analyze
its role:
If
in detail,
the bonds used
for international reserves the central
from the private sector and
arises
(i.e.,
by transferring a
the latter case that
exchanging these bonds
liquidity
lower the effective cost-of-capital
liquidity of the country
domestic liquidity
of the international spread to borrowers).
while the latter
may
may
cause the intervention to
sterilization generates
a significant tax burden
to potentially distressed firms.
In conclusion, since emerging economies are characterized by the illiquidity of their
financial markets
and
their
weak
links
and backfiring must be perceived as
with international financial markets, ineffectiveness
real possibilities that reflect these pervasive structural
problems rather than the particulars of an exchange rate system or monetary mechanism.
—
Other related literature
methodology.
Aside from the international economics references already mentioned and to be mentioned throughout the paper, our paper belongs to a growing literature that studies the
macroeconomic consequences of aggregate
mination of the
latter.
liquidity shortages,
Our modeling approach owes to Diamond and Dybvig's
Holmstrom and
canonical model of liquidity and particularly to
from
model
this
and the endogenous deter-
Our
in the context of firms.
basic
(1983)
Tirole (1998)'s departure
model-economy
is
a simplified version of
that in Caballero and Krishnamurthy (2001), which departs from this literature primarily
in that
it
considers two forms of liquidity, one domestic and one international. In this sense,
our basic model also relates to the recent work by
Diamond and Rajan
(2001),
where banks'
solvency constraints play a role similar to our domestic collateral in determining domestic
interest rates during times of aggregate liquidity shortages.
Given our focus in this paper on the interaction between aggregate liquidity considerations
and the impact of macroeconomic
work
relates
most
closely to that of
While not addressing
through changes
sterilization,
in public debt.
policies relying
on
asset
market interventions, our
Woodford (1990) and Holmstrom and Tirole
both present models
Government
in
which policy has
(1998).
real effects
policies are non-Ricardian because public
debt provides the private sector with liquid assets that they are unable to create for themselves.
sector
Holmstrom and Tirole (1998) show that when there
may have
a shortage of liquid assets. Public bonds
ernment policy has
real effects.
level,
make up
this shortfall
and gov-
In our model, on the other hand, issuing public bonds
does not increase private liquidity and
At a more abstract
are aggregate shocks, the private
may do
just the opposite
nonetheless, the policies
we study
when the
policy backfires.
are related to theirs in that
government policy acts through changing the liquidity of the assets ultimately held by the
private sector.
In section 2
we
lay out our basic model.
are underdeveloped an externality arises
We
show that when domestic
whereby the private sector draws down
tional liquidity too fast relative to the constrained efficient outcome.
show that
in equilibrium asset prices are
financial
More importantly, we
determined by domestic and international liquidity
whether actual and anticipated interventions
in asset
markets by the central bank
supporting the exchange rate) always result in an implicit financial subsidy.
offset
interna-
Section 3 begins our discussion of a sterilization, and asks the question
considerations.
of this section
its
markets
is
(e.g.
The main
by
result
a stark Ricardian benchmark, where the private sector's actions perfectly
those of the central bank. In a well defined sense, and despite the large ex-post impact
bank interventions
of central
in a liquidity-constrained environment,
the bond market
it is
rather than the central bank that determines the ex-ante asset prices that influence firms'
financial decisions. This reference
a useful starting point in the process of identifying the
is
mechanisms that are actually responsible
central
for the potential financial subsidies implicit in
bank intervention.
The main
result of the
paper
presented in section
is
environment of financial underdevelopment, intervention
markets are
illiquid or interventions
firms. Section 5 addresses another
We show that
4,
where we show that
may
backfire
if
an
in
domestic bond
generate sizable tax burdens to potentially distressed
dimension of "backfiring" often stressed
in the literature.
our mechanism also fosters a shift in the maturity structure of capital inflows
toward short term debt, even
term.
4
2
The Basic Setup
Section 6 concludes and
This section sets the stage
for
—or
is
particularly
when
—
the sterilization bonds are long
followed by an extensive appendix.
our discussion of the consequences of sterilized interventions in
an environment with underdeveloped domestic
financial markets.
The model
is
a simplified
version of that in Caballero and Krishnamurthy (2001) and has two objectives: First,
illustrates
how
international
and domestic
role in the
main section
impact on
this price.
interest rate. Second,
it
it
liquidity considerations influence the domestic
cost of capital during crises. This will be the only price in our
its
a
model and
will play a
key
of the paper, as the novel effects of sterilization will arise from
We
will clarify that this price
different
is
than the domestic peso
uncovers a pecuniary externality that justifies policy intervention
to reduce domestic investment during capital inflow booms.
2.1
A
Liquidity-based
The main macroeconomic
Model
of Domestic Interest Rate Determination
event that concerns us here
is
an external
crisis,
as a shortage of external financial resources relative to the country's needs.
at date
1,
and as
this
is
We
The
relatively plentiful
introduce one additional date where precautions
shift in
by, e.g., Montiel
crisis
it is
occurs
followed
but could not be credibly
1.
public sector (next section)
4
The
the result of a binding external financial constraint,
by a date 2 when the country's resources are
pledged at date
to be understood
— against
the
crisis
— by the private sector, and the
can be taken.
Date
is
a
fully flexible
composition of capital inflows during periods of intensive sterilization has been documented
and Reinhart (1999) and Calvo, Leiderman, and Reinhart (1993).
when agents make investment,
period
indexed by
There
t
is
=
decisions.
The
periods are
0, 1, 2.
a unit measure of domestic agents, each endowed with
(tradeable) good, that arrives at date
foreigners (e.g., prime exports),
per period (henceforth
i*
and insurance
financing,
all
who
These date 2 goods have
2.
are willing to lend against
rates are dollar-rates).
it
w
units of the single
full collateral
and
at dates
value to
1 at
a rate
Foreigners play no other role in our
model.
Domestics also have access to a production technology. Building a plant of
requires
them
to invest c(k)
output proportional to the
at date
0,
— with
>
c(.)
>
0, c'
size of the plant (see
below)
and
.
c"
size k at
date
— which yields date
>
2
Since domestics have no resources
they must import the capital goods and borrow from foreigners, doj, to finance
this investment.
The
plant profits at date
financing and investment decisions are taken to maximize expected
To keep matters
2.
simple, domestics have risk neutral preferences
over date 2 consumption of the single good.
Neither the plants nor their expected output are valued as collateral by foreigners.
When
domestics mortgage a part of their international
real investments are undertaken,
collateral in securing foreign funds.
debt contracts, thus, doj
<
For now,
financing
all
is
done via
fully collateralized
w.
Crises
We
thus define a
crisis as
an event
in
which the financing needs of the economy exceed
the aggregate external resources available to
it.
In our simple
economy the financing need
stems from the ongoing maintenance of the productive structure. The plants of one-half
of the firms receive a production shock at date
r.
However, the productivity decline can be
date 2 output
i*.
offset
=
(r
+ 9A)k<
where
Rk,
(9
<
1)
to
goods, to give
A = R-r.
assume that the return on reinvestment exceeds the international
This means that firms borrow as
much
the economy can only borrow up to
w—
interest rate:
A— 1 >
as possible from foreign lenders to finance
reinvestment, and the only limit on reinvestment
is
by reinvesting 6k
R
of,
R(9)k
We
that lowers output per plant from
1
the external financial constraint. Since
is
doj goods at date
1,
the condition for a
crisis
that firms' reinvestment needs be greater than the economy-wide debt capacity (see the
appendix
for restrictions
on primitives),
k
-
>w-d
6
0if
.
Domestic
financial
Both the need
markets
for and,
role of the latter in our
first
policy, will
depend
on the functioning of domestic financial market. Let us turn next to outlining the
crucially
A
most importantly, the impact of central bank
model.
firm that receives a shock
borrows
w—
To cope with the shock, the
said to be distressed.
is
doj directly from
foreigners.
5
After this,
must turn
it
firm
to the domestic
firms that did not receive a shock ("intact firms") for funds. Intact firms have no output
at date
1
either, so
they must borrow from foreigners
w—
can do up to their
firms. This they
Our domestic
system
financial
is
international collateral at times of
firms
any more than foreigners?
installed assets as well.
But
doj of financial
slack.
the institution where the economy aggregates
crisis.
We
they are to finance the distressed
if
But why would
all its
intact firms lend to distressed
assume that domestics value as
collateral the firm's
since a perfectly functioning domestic financial market
is
hardly
a good description of an emerging economy, and this departure has central implications
for
our analysis, we assume that only a fraction of the output from domestic investment can
be pledged to other domestics. Domestics lend to each other using
contracts, but this collateral
is
limited to the
functioning domestic market will have the
minimum output
full
debt
fully collateralized
of rk. In contrast, a perfectly
output of Rk as
collateral. Later on,
we
will
associate the aggregate quantity rk to the development of the domestic financial market.
These two ingredients: heterogeneity
in financing needs
among
domestics, and a differ-
ence between foreigners and domestics in their willingness to extend finance, are necessary
to give a distinct role to the domestic financial market. 6 7
'
Dollar cost of capital for firms/Expected return on domestic loans
The
price
Note that
In a
we are
this
is
interested in
is
the rate on borrowing dollars against domestic collateral.
a dollar-cost interest rate - so in practice,
crisis, this will
the economy would be
reflects
not be sufficient to fully rebuild the plant, since
less
if it
the
sum
of peso interest
did the total financing need of
than the international collateral of distressed firms of
than the financing need assumed during a
6
it
—
^-^-
.
This term
crisis.
See Caballero and Krishnamurthy (2000b) for a model where the asymmetry
in valuation
comes from
the fact that domestics value nontradable goods while foreigners do not. Here, on the other hand,
goods are tradable but there
7
Since
is
is less
an asymmetric valuation of
we do not want insurance markets
to
less
than prime
all final
assets.
undo the ex-post heterogeneity, we assume that idiosyncratic
shocks are non-observable and non-contractible. Moreover, we assume that the coordination-fragile banking
equilibrium
is
not feasible (see Caballero and Krishnamurthy 2001).
rates
and expected exchange rate appreciation. This price
Consider the rate at date
This price
is
rk/2.
both the dollar cost of capital
need of funds, as well as the expected return on loans
for firms in
amount
is
is
1
collateral.
determined by both, the amount of collateral of distressed firms and the
Thus they pledge rk/2 of date
—
domestic lenders.
on a one-period domestic loan against rk units of
of international liquidity of intact ones.
goods of (w
for
dQj)/2.
If
2
domestic collateral
The aggregate
collateral of distressed firms
goods to intact firms
in
not too limited, there
is
exchange
is
for
i*:
1
a scarcity price for
the international liquidity, and the price needed to clear the domestic loan market,
exceed the international interest rate,
date
v\, will
8
TK
.7)
1
1
w - d0J
^
•*
>
1
I
This expression highlights the effect of domestic collateral and international liquidity on
date
A
1 cost if capital.
shortage of the latter means that
former means that the cost of capital will generally be
investment at date
i\>
less
i*\
while a shortage of the
than the marginal product of
(A):
1
if
Figure
I:
<A-1.
Liquidity Based Cost of Capital Determination
f
A-l
i
1
w Af
Figure
investment
8
By
1
illustrates this.
(less cost)
not too limited,
The curve
that rk
r
Domestic loans
in long dashes represents the
for a distressed firm.
we mean
Constrained
demand
Co
,
> (w —
That
do,/)(l
+ i*).
marginal product of
in short dashes represents the access
This supply becomes vertical when the external
to international funds from intact firms.
The
financial constraint binds (our external crisis).
funds by distressed firm.
solid line represents actual
below the long-dashed curve because
It lies
downward
the limited collateral of distressed firm.
It is
domestic collateral of rk
more resources
As long
is
able to fetch
it is
as i\
our analysis
is,
however, that this return will generally be
domestic investment.
When rk
is
amount
assets.
The
must
above
lie
emerging
crisis,
novel observation of
than the marginal product of
less
in
an
low, domestic borrowers have limited collateral so that
external investment in these borrowers returns less than A. In equilibrium, the return
somewhere
of
falls.
This simply means that during times of
market assets must return a premium over international
for
constrained by
sloping because a fixed
as the external financial constraint binds, the dollar cost of capital
the international interest rate.
demand
between the international interest rate and the marginal product (the
lies
social
value) of a unit of international liquidity.
In other words, in the aggregate, a shortage of international collateral yields a spread
between domestic marginal product and international cost of
on the other hand, determines the sharing of
borrowers of a marginal dollar.
to distressed firms (borrowers).
As domestic
That
is,
this spread
Domestic
capital.
between domestic lenders and
collateral falls, a larger share of
in the latter
collateral,
it
example the cost of capital
because the economy has access to more dollars from foreigners but because
it
accrues
falls
not
has
less
domestic collateral to pay for the dollars already at home.
Consider next the two period dollar cost of capital at date
on a loan against rk of date
2 collateral
made
take one unit of international liquidity, invest
at rate if at date
1.
at date 0.
it
A
0.
the dollar-cost rate
is
domestic lender could always
at i* from date
Indifference between this strategy
This
to date
and lending
for
1,
and then lend
it
two periods implies
that,
The date
one period cost of capital
is
therefore just the international interest rate,
i
p -
i*
Since the external financial constraint does not bind at date
the
if
same return
the
as international assets -
economy runs
into a crisis
in the interest parity condition.
i.e.
0,
domestic assets must yield
the interest parity condition holds. At date
and the external
financial constraint binds, there
is
a
1,
shift
Interpreting the domestic cost of capital
Since the negative relation between the domestic cost of capital and the domestic finan-
may appear
cial constraints
as counterintuitive,
it is
important to pause and highlight two
features of this price:
•
As
a matter of interpretation, i\
is
the rate for a domestic taking out a fully secured
domestic loan and receiving dollar proceeds. Literally the loan
firm one dollar at date
at date
It is
2.
1,
and
collateral, as
interest rate
emphasized
at date
goods secured by domestic
made by another
opposed to a domestic currency
in
MF
and other uncovered
collateral
domestic, secured
interest rate (which
is
the
interest parity setups).
it is
the expected return to a domestic lender of making a loan
not the promised interest rate on a defaultable loan. This promised interest
1,
which
there
{\-\-i^)
corresponds to the rate on a fully secured loan, the interest rate has no default
it
risk built in. Rather,
rate,
repay
the interest rate on a dollar loan
by domestic
• Since
will
it
lend the distressed
is:
is
what we
typically observe, rises
when a borrower has
less collateral since
a higher risk of default.
is
In the particular context of our model, the natural risky-rate
A—
1.
rate,
but
it
and thus
it
A
premium
decline in i\
due to a reduction
is
constant and equal to
in domestic collateral does not affect this
raises the risk-compensation (or, loosely, the default risk)
component of it,
reduces the payment for the scarce international liquidity (the liquidity-
falls).
As we show
in the next section,
it
is
the liquidity
premium that
matters for precautionary decisions.
Besides the visual aid in the figure above, the international interest rate
is
low to justify borrowing
sufficiently
henceforth set
2.2
We
it
to zero:
i*
=
—as long as
— plays no significant role in what follows.
We
it
shall
0.
Overborrowing and Incentives to Precaution
turn
now
to
show how the domestic
sector investment
and precautionary
cost of capital at times of crises affect the private
decisions. This
will tie the justification for restrictive policies to the
markets.
Second, and most importantly,
sterilization
may
Our main
starkest case
affect,
results
is
made
it
is
important
for
two reasons.
First, it
underdevelopment of domestic financial
will identify the channels
through which a
perversely or not, these decisions.
do not depend on the presence of aggregate shocks, and indeed the
precisely
when the
crisis is fully
10
predictable but
still
the economy runs
.
We
into trouble.
assume that date
of an external crisis
— that
aggregate conditions are fully anticipated to be those
1
the external financial constraint
is,
is
binding in the aggregate.
See the appendix for parameter conditions to arrive at this scenario.
Let us consider firms' incentives to precaution against the date
how
these incentives are affected by the domestic cost of capital.
an extra unit of k
an intact firm
Rk goods
firm obtains
collateral
for
at date 2.
is
On
composed of two
the cost side,
which would have yielded a return of
it
i\ in
The
and, particularly,
net return on investing
On the gross
pieces.
return side, the
sacrifices c'(fc) units of international
the domestic financial market at date
Thus, at the margin increasing k yields an ex-post return, net of opportunity
1.
fl-cf(fc)(l
tional collateral
is
+ *?).
(2)
undervalued by an intact firm as long as
a unit of international collateral
for
cost, of
follows immediately from (2) that the opportunity cost of sacrificing a unit of interna-
It
collateral, the value of holding
distressed firms at date
firm's
1 is
is
i\
< (A —
Since the
1).
demand
constrained by the distressed firms' limited domestic
on to a unit of international
collateral in order to supply
depressed relative to the socially
efficient
it
to
— and the distressed
— valuation of A.
Now
consider the
same net return
for
margin the firm obtains r unit of goods
a distressed firm.
directly.
1,
multiplied by the private return that each generates
secured by r goods at date
2,
generates proceeds of
Thus the
reinvested at a gross return of A.
cost side,
the gross return
side, at
the
the value of these r goods must be
(its
value as a collateral asset)
—^ goods at date
1,
.
A
loan
which each can be
gross benefit to increasing k
sacrifices c'(k) units of international collateral
it
On
But because these goods can be pledged
as collateral in the domestic loan market at date
A
1 crisis
is
r—rp.
On
the
which yield a private return of
to a distressed firm. Thus, at the margin increasing investment in k yields a net return
of:
It
follows immediately
collateral
as
ij
is
from
that while the cost of sacrificing a unit of international
properly valued by a distressed firm, the domestic investment
< (A —
1),
a distressed firm
selling its overvalued
during an external
is
able to keep
domestic collateral.
crisis
A
is
not.
As long
some of the surplus from reinvestment by
central planner,
on the other hand,
realizes that
only international collateral generates social surplus, and hence
discounts domestic assets at
9
(3)
A
rather than
Since the analysis above takes date
(
1
+ i\)
decisions as given, the expected return on domestic loans splits
11
Although
for different reasons,
both intact and distressed firms overvalue (from a social
point of view) domestic investment relative to
its
cost in terms of the international collateral used.
the average of these two outcomes,
under-precaution for the date
it
The problem
- limited rk constrains
is
we
really
demand
shall
based on
follows that firms will overinvest, overborrow
and
when domestic
arises
financial markets are underde-
so that
A—
1
>
Asset market interventions such as
if.
and
and thereby
effect of sterilization policy
affect real decisions. In
within our model in terms
their feedback into the underlying financial constraints
real decisions.
A
3
is
one of distorted asset prices due to financial constraints
study the
of possible effects on asset prices
and
opportunity
Since the ex- ante decision
sterilization are policies designed to alter asset prices
the next sections,
its
shock. 10
1
The over-borrowing problem only
veloped.
— namely
opportunity cost
Ricardian Benchmark despite Liquidity Pricing
In our model, the only relevant prices are
through changes
by the
and
i$
in these dollar costs of capital.
interest parity condition, thus
we can
if, thus policy
The date
can only have
rate
is
real effects
pinned down to be
i*
further conclude that the effects of policy can
only arise through effects on if. 11
the reinvestment surplus,
A — 1,
between domestic lenders and borrowers but
of the economy. Total reinvestment
(that
is,
the
sum
is fully
of the international collateral available to distressed
-9k
it
does not
affect the real side
determined by the total availability of international collateral
= -(w -
d
,/)
+
-^(w
-d
and
intact firms):
,/),
to imply (in a crisis):
k
This dichotomy between the real and financial side disappears at date
ment and
portfolio decisions.
one. In particular,
It is this financial
10
it
gives
When
a domestic decides to
up some of
decision that
is
its
make a
0,
when domestics make
real investment,
it
makes a
financial
international collateral, w, in exchange for domestic collateral, rk.
affected by
i\.
This opens the door to international liquidity management policies, as we study
namurthy (2000a). See
also
their invest-
also Harberger (1985)
and Aizenman (1989),
for alternative
in
Caballero and Krish-
models of over-borrowing
based on the undervaluation of the country's "monopsony" power in international financial market.
n We
should note at the outset that whether a policymaker has our structure in mind when he decides to
sterilize capital inflows is
not what concerns us here.
He may
well take as reference the standard Mundell-
Fleming framework. Instead, we simply ask the positive question of how does domestic underdevelopment, as
characterized in the previous section, affect the outcome of this intervention. Given this objective, our results
are clearest once the standard distinction between
money and bonds
12
is
removed and analysis
is
refocused
The
in the
MF
framework emphasizes a
on borrowing pesos
different price; namely, the price
domestic market or the peso interest rate. This price plays no direct role in our model
margin firms always need to borrow
since at the
to this
model to determine
dollars. It
is
possible to
neutral-money expanded
this peso interest rate. In its simplest
form, the dollar costs of borrowing
we have focused
on, if
and
add a monetary side
would be related to these
if,
peso interest rates by a standard uncovered interest parity condition: if
where the
latter
split
between the
latter
and peso
interest rate
that in the
it
MF
peso
+
e\/e2,
components, neither
of which would independently have any real effects in our setup. This
but
i
term represents the expected appreciation of the peso. Money quantities
would pin down the
stylized,
=
serves to highlight the very different
mechanism
is,
of course, quite
model versus
at play in our
framework.
Let us begin with the question of whether actual and anticipated interventions in asset
markets by the central bank always have
effects
on private sector actions. Indeed
a good starting place because this seems to be the prevalent view in policy
committing reserves to support an exchange
rate, for
implicit subsidy to external borrowing.
12
considerations at the time of crises,
may appear
intervention
must always have
This implication
is
it
effects
example,
is
Since in our model if
circles,
this
where
automatically seen as an
is
determined by liquidity
that an anticipation of a central bank
on the perceived cost of
not as straightforward as one
may
capital.
think at
first,
however, because
arbitrage actions by the private sector are a powerful force that need to be accounted
The main
result of this section
is
for.
a stark Ricardian benchmark, where the private sector's
actions perfectly offset those of the central bank. This reference
in the process of identifying the
is
is
a useful starting point
mechanisms that are actually responsible
for the effects of
a central bank intervention.
The
3.1
intervention
Let us postpone issues of generality to the next sections and appendix and concentrate
instead on a small central bank intervention in domestic asset markets.
sterilization
on the
we mean
effects arising
a date
from the swap of public bonds
capital inflows. Alternatively,
system. In this case,
sale of dB public
we could have
money supply
given
bonds
Concretely, by
to the private sector in
for international reserves implicit in
money a
specific role but
assumed a
exchange for
a sterilization of
fixed
actions are fully offset by capital flows, eliminating the
exchange rate
monetary (but
not ours) transmission channel as well.
12
In our model
i\
relative to the date 2
can be reinterpreted as the depreciation in the real exchange rate during the
exchange rate (which
is
one, since all goods are alike).
13
crisis,
dB
G international liquidity (reserves), where
is
i*§
the interest rate
This intervention comes with an anticipation of how
We
crisis.
shall
it
on these bonds.
to be reversed during the
is
postpone a thorough discussion of this issue and
for
now
just assume, as
with the private sector, that the central bank cannot discriminate across firms but only
across financial instruments. Interventions
and
open markets.
reversals occur in
The bond market and arbitrage
3.2
Recall that in our setup firms have rk of date 2 domestic collateral that they use as security
to obtain financing. For the sake of exposition,
collateral as corporate
bonds -
in fact, in
let
us refer to the claims issued against this
emerging markets, financing
is
more
likely to take
the form of bank loans, but the precise instrument does not alter our message. Firms can
issue
up to rk domestic corporate bonds.
an active bond market must place constraints on the
Logically,
intervention. Since investors can invest in corporate
these two bonds must be related.
liquidity
firms
and public bonds, the
Suppose that the government
that the private sector purchases at interest rate
at date
13
sell
Denote the
effects of the central
interest rate
on these bonds
Iq
bank
interest rate
issues two-period
on
bonds
with units of international
at date 1 as if
.
At date
1,
distressed
both corporate bonds and [now, one-period] government bonds to obtain funds
The buyers
for restructuring.
reserves). For
in this
market are intact firms and the central bank (using
an intact firm, purchasing one unit of either a corporate or government bond
gives one unit of
consumption at date
2.
Therefore, arbitrage across these instruments will
require that,
i
An immediate
is
G - ip
implication of this restriction
irrelevant for outcomes.
bonds or repurchase
its
portfolios accordingly
At date
1
is
that the
form of
the government can choose to purchase either corporate
government bonds
in the
open market, intact firms
and ensure that the prices on these
14
We
The form
will alter their
securities are identical.
to target purchases toward a particular asset will not boost
assets.
the intervention reversal
its
Choosing
price above that of other
will revisit this issue in section 4.
of reversal
is
irrelevant because of cross-instrument arbitrage conditions.
We
next show that intertemporal arbitrage renders the entire sterilized intervention irrelevant
13
See the conclusion for a discussion of the realism and relevance of our assumption on the maturity of
public bonds.
14
This should serve as a warning to some of the informal "moral hazard" stories that see any asset support
as a cause of moral hazard.
14
.
economy. The point can be made precisely by examining interest rate
for the real side of the
determination at date
can only have a
Since in our setup the only equilibrium price
1.
real effect
by changing
Denote the private sector decisions
the
moment
central
rises
let
this cost of capital for a firm.
an economy without intervention as
in
will
bonds, increasing their domestic collateral to rk +
why
bonds against
now have dB
dB
appendix
(see the
w—
lowering this to
distressed firms sell their
dB/2
doj
—
G
^
.
The buyers
bank has dB/(l
of international liquidity, while the central
post-sterilization (hence, the hat) interest rate
p
H -
TK
Note next that a firm at date
w
_i_
bond
to date
we can then
,
l
sterilization has
ment depends only on
decisions at date
4
Iq
no
and
1 is
i$) of reserves.
zero.
=
/2
)
Thus, the
for
one unit
an rate oiif Alternatively,
.
=
for
it
one period. Arbitrage
or that rate of return on holding the
Thus intertemporal arbitrage
=
if
Iq
requires that:
if-
(5)
rewrite the post-sterilization cost of capital as:
~p
+ %K =
effect
if,
w — doj — /fa
(4)
at date 1 at
same outcome,
»o
(4),
(
1.
d&
can invest these funds internationally at an interest rate of i*
requires that either strategy yield the
corporate
of both the corporate and
can purchase a government bond yielding
of international liquidity, and then sell this
government bond from date
+
sell
ur>
-t-
A
-
t^%
is,
~i
is,
a sovereign risk
public bonds. In addition they
their domestic collateral totalling
Substituting (5) into
for
.
public bonds are intact firms and the central bank. Intact firms have
That
For
of domestic government
public bonds are not accepted by foreigners). Second, they have
less of international liquidity,
1,
(k, do,/)-
which implies that private international indebtness
as given,
by the amount of the intervention. Firms
At date
if, intervention
us assume that the private sector does not re-optimize as a result of the
bank actions and take k
justification for
is
fk
—=
w-d j
.p
l
+ %\
on the cost of capital
for
the firm.
Since real invest-
and these remain unchanged by the intervention, private
are also unchanged.
Sterilization Backfires
Financial underdevelopment in emerging economies extends beyond the private sector. Particularly during crises, illiquidity also affects the secondary
15
market
for public bonds. In this
we show that
section
intervention
may
the central bank sterilizes by transacting in illiquid bonds, the
if
backfire. It does so
may
External borrowing
tic firms.
by generating an implicit financial subsidy to domes-
rise
beyond the intervention, resulting
rather than the desired contraction in domestic activity. This subsidy
that the
economy ends up with
less international liquidity
is
in
an expansion
illusory however, in
during crises and firms experience
greater financial distress.
The essence
4.1
In rough terms, the intuition behind this backfiring result stems directly from the results in
section
There we noted that a reduction
2.
in
and over-borrow
firms to undervalue international liquidity,
illiquid,
to
fall
The
at date
0.
If
fall
inzf causing
,
public bonds are
then in the sterilization the central bank removes a liquid asset (international
and replaces
liquidity)
%i
domestic collateral leads to a
and
is
with an
it
one (public bonds). The
illiquid
loss of liquidity causes
akin to a reduction in domestic collateral.
made
point can be
suppose that there
is
precisely
by focusing on a stark representation of
no secondary market
these bonds can only be redeemed at date
for
government bonds so that once purchased,
On
2.
illiquidity:
the other hand,
let
us continue to assume
that firms can issue corporate bonds with no frictions other than the collateral constraint
ofrfc.
15
The
backfiring result appears counterintuitive, because at
first
glance one would think
that the private sector would compensate by reducing investment and borrowing
thereby making up for the
Fixing if at
decision at date
lost liquidity.
its pre-sterilization level,
0.
The
first
1
This intuition turns out to be misleading.
we know that
step in our argument
firm has less liquid assets, the illiquidity
depend only on aggregate international
less,
is
is
firms will not alter their investment
though each
to recognize that even
only apparent, since investment at dates
liquidity,
and
this has not
changed
and
— some of
it
has just been reallocated to the central bank.
Taking
i\ as given, investors in
on the purchase of
intact has
illiquid
our model adjust to the illiquidity by requiring a premium
bonds rather than by cutting
an opportunity cost of
i\ for
real investment.
A
feature that
market of public bonds.
is
It
is
each unit of international liquidity used to buy
public bonds; a distressed firm, on the other hand, has an opportunity cost of
The only
firm that
essential for our results is that there
needs not be as extreme as
is
some imperfection
we have assumed
(A —
in the
1) for
secondary
here, nor does the (short term)
corporate bond market need to be more liquid than the (secondary) public bond market. See below and the
appendix
for these extensions.
16
same
that
unit of international liquidity. Thus,
However, the key step
domestic liquidity
decline in if.
To
A ~\~ H
= *f +
*o
>
*f
(6)
to note that if will not remain the same, since the loss in
is
when domestic
markets are underdeveloped translates into a
financial
investment decision at k (and the corresponding
see this, fix the date
private external debt associated to this real investment at doj). Distressed firms issue in
total
r
corporate bonds at date
~2
1,
where
T
is
a tax to finance the quasi- fiscal
caused by the government having to offer a liquidity premium on public bonds.
16
deficit
This tax
plays only a secondary role here (see section 4.3 for a converse case), which nonetheless
reinforces our conclusion.
do j
—
17
The supply
-^— )/2 of international liquidity from intact firms and
u
central bank. This gives
market clearing
p
h
l
Since
of international liquidity
dB and T
IK
are strictly positive,
-T1
—
-^%
of reserves
—
from the
~
1-
(7)
-
it
must be the case that if <
the post-sterilization equilibrium k and do,/ must
4.2
exactly as before: (w
of:
-
w
is
That
rise.
is,
if.
follows that in
It
the policy backfires.
Reversing the sterilization through public bonds:
On
the irrelevance
of the reversal's instrument
Does the
for
fact that the reversal
our results?
We make
If so, it
would appear that our
two assumptions on taxation that continue
tions. First, the
goverment balances
at date 2. Second, as
dowments of agents
(i.e.
rk
falls
it
result
in
our
is
directly matter
equivalent to the moral-hazard
spirit of
introducing no extraneous distor-
budget by raising lump-sum (non-distortionary) taxes on
to give the
government the power
we assume that an
to access future
all citizens
endowments that
increase in taxes reduces the promisable future en-
by the amount of taxes). See Woodford (1990) and Holmstrom and Tirole
models where the main
implicitly pledging
its
we do not want
the private sector cannot promise,
(1999) for
was done by supporting corporate bonds
role of
ability to tax.
While
government policy
this feature of
is
to increase the private sectors collateral by
governments seems
realistic it
clouds the issue
we
are trying to highlight.
17
The tax-component
sterilization
of our story relates to one of the few existing explicit models of the perils of
Calvo (1991).
Building on Sargent and Wallace's (1981) unpleasant monetary arithmetics,
he formally shows that, by raising domestic interest rates, the government increases
and creates a
quasi-fiscal deficit that
may jeopardize
protected by the sterilization. This mechanism
inefficiencies rather
is
its
the very stabilization attempt that
different
than monetary commitment problems.
17
from ours
in that
debt-service burden
is
supposedly being
we emphasize
financial
market
caused by ex-post bailouts that has been discussed extensively in the literature. However,
this
is
the wrong lesson to extract. As section 3 showed, the problem
Rather, our main point
per-se.
is
the necessary ingredient
transacts
is
not in the "bailout"
—hence this
that the implicit subsidy problem stems
is
— from the
instrument that the central bank
illiquidity in the
in.
In fact, while our explanation in terms of a reversal targeted to corporate bonds makes
the implicit bailout interpretation obvious,
Hong-Kong, Taiwan and possibly Brazil
it is
actually not needed. In a few cases
— central banks have directly supported traded
monetary expansion and an expansion of bank
To the extent
credit to the private sector.
that the central bank backs this expansion, in fact
is
e.g.
In others, the reversal of sterilization takes the form of a
private claims during crises.
usual case of reversal
—
acquires private assets. However, the
it
bank purchasing back public rather than private
of a central
Next we enrich the model to address
assets.
this scenario.
Let us endow firms with public bonds of Bo that the government finances at date 2 by
lump sum taxes
collecting
domestic collateral of rk
We move
bonds
+ Bo —
To and international liquidity of
it
markets incurs a linear transaction
incurs a transaction cost of aB.
collects proceeds
on
effectively (1
— a)B
analysed in the previous subsection.
The
date
central
1 it
firms to
bank
sterilizes
injects these reserves
rate of if.
Suppose that a
sell their
is
by
a=
dB bonds
selling
1
a firm
and
1,
sells
this
B
government
means a
seller
corresponds to the case of illiquidity
for
small enough compared to
-^%
of international reserves.
At
A
so that
it is
optimal
for distressed
bonds:
—Q
A.
Clearly, intact firms will not liquidate at date
transaction cost. Market clearing at date
,
P _ G _ rk
.
shall not provide a deeper
One
If
back into the market to repurchase public bonds at interest
-<
it.
cost.
With a between
bonds,
1
We
do,/-
18
1
motivate
w—
to a less extreme form of illiquidity by assuming that the sale of government
in secondary
bonds, then
of To. After investment of k, the position of each firm comprises
+
(gg
model of
possible interpretation
is
1
1
since they will not wish to incur the
gives the post-sterilization cost of capital,
+ dB)(l -a)-T -dT _
w - d°J + 145*
this transaction cost but
that
it
it is
clear that there are
many ways
represents the costs and profits of a market maker.
18
to
.
We
can rewrite this expression
(l
+ l?)(w-doj) =
rk
as,
1+i (n ^ 1+i
-a);
+ Bo{l-a)-To + dB^
1 + %
1 + *o V
-
l
';'
I
Consider the term in parentheses on the right hand
government bond and always
the international interest rate
liquidate his position at date
at date
sells it
is
zero,
1, it
A
side.
receives, (1
1
I
dT.
J
i
firm that invests one dollar in a
— ct)r—ty
and the marginal investor
must be
I
I
,
dollars at date
at date
1.
Since
does not always
that,
(l-a)ff&<l.
With
small liquidation costs, this
is
the expression that
tells
us that the intervention leads
to a liquidity loss by firms.
Given
and the
this
(i.e.,
that the term in parentheses on the right hand side of (8)
fact that sterilization causes
1
Thus, even
if
+ ;P <
rk
dT >
0,
we can conclude
is
negative)
that,
+ B (l-a)-T
=1+ f
w-d j
.
the central bank intervenes at date
1
by targeting
its
reserves only at public
bonds, the intervention would backfire. Corporate bonds are supported indirectly by the
decline in domestic liquidity.
The
4.3
It is
relevance of distressed firms' liquidity loss
important to remind the reader that domestic financial underdevelopment
ingredient for backfiring. If rk were large (to highlight a large
rather than a higher productivity at date
product of reinvestment (A) and the
1),
amount
is
a necessary
of domestic collateral
if would be determined by the marginal
caused by the sterilization would be
loss of liquidity
inconsequential
The other necessary
which aspect of
distressed firms
ingredient for backfiring
this illiquidity matters?
when they purchase
is
the illiquidity of public bonds.
The answer
is
that
it is
the loss of liquidity to
these bonds that matters, because
it is
need to liquidate their bonds before maturity. The point of this subsection
the reduction in liquidity of distressed firm
bonds are
may
arise
But
these firms that
is
to
show that
from other channels, even when public
fully liquid.
In particular, a
common concern in many emerging economies stems from possible misuse
of public resources. Suppose, for example, that the reserves gained in sterilization are used
to bail out the failed real estate development of
19
some highly placed
officials.
We show
below
that aside from the standard concerns with such practice, this action triggers a liquidity
mechanism
loss
similar to the one
government runs a
quasi-fiscal deficit
any public assets they hold
burden
we have emphasized with the
illiquid public
which means that distressed firms
will fall in value, or will find that
in the future. Either situation
will find either that
they have an increased tax
akin to a domestic liquidity
is
loss.
=
Let us return to the case where public bonds are fully liquid (a
by
dB bonds
selling
-^u
for
= Bq
of taxes at date
bonds outstanding.
central
If
Bo bonds outstanding, and was expecting
After the intervention, the central bank has
1.
the reserves of
bank
central
of international reserves.
Prior to the intervention, the central bank had
to raise To
but continue
0),
The
to assume that firms have initial holdings of government bonds of Bosterilizes
The
bonds.
/f 6
i-b
had been used to purchase bonds
bank would have had resources of dB
% + Tq
at date
2.
=
Since if
+ dB
Bo
at date
iff
1,
the
in this case,
the intervention would not have affected the budget of the government,
B + dB = T + dB,1+if
i+ir
and the Bo bonds would
Now
still
be repaid with To taxes.
suppose that instead of reversing the transaction in the open market, the central
bank simply
gives the reserves
away
at date 1 to a single agent (a gift).
19
This agent then
takes these reserves and purchases corporate (or public) bonds with them, that he redeems
at date 2.
The government now
loses the revenue
from the repurchase so that
it
has a
deficit
of:
Tq
There are three options
available: the
thereby reducing the value of
causing the debt to
- Bo - dB <
fall in
its
0.
government can finance the
deficit
by
inflating
bonds; the government can default on a fraction of
government can
value; or the
All three of these options have the
same
effect in
its
and
debt
raise taxes to cover the shortfall.
our framework. To illustrate this
effect,
consider the default option.
Since the government has expected taxes of To and issued bonds of
of each
rk
—
19
bond must
To,
be,
Pb = B +43 A
and government bonds
Since this
gift is
distressed firm at date
totalling
Bo
+ dB,
given to an agent of infinitesimal
are infinitesimal as well.
That
is,
size,
1
Bo + dB, the
has domestic collateral of
and international
the direct effect of the
liquidity of
gift
w—
do,/-
on the agents actions
traditional bailout effects are absent in this modeling.
20
price
Thus market
clearing at date
~.
1
gives the post-intervention cost of capital
rk + (Bp + dB)PB P _
=
liquidity.
rises,
each bond
The market
falls in
value,
Again, the cost of capital
fall (or
and we have a
r^
w—
to
()
result akin to a loss of domestic
clearing condition can be rewritten as,
~p
The government
T
—hj + gk
"
As dB
of,
do j
for firms falls
action of giving
+
and
p
-•
77
i+j;
'0
sterilization backfires.
away reserves
at date 1 causes the value of
future tax liabilities to rise) and results in a liquidity loss at date
mechanically the same as that of
0,
its
bonds
so that this
of government bonds.
case
is
5
Another Symptom of Backfiring: Short-term Capital
full illiquidity
In-
flows
The observed shortening
larly interesting
in the maturity of capital flows following sterilization
from our point of view
(see, e.g.,
Leiderman, and Reinhart (1993)). The step
defining short
and long term debt
in
is
particu-
Montiel and Reinhart (1999) and Calvo,
in arriving at this result
from our model
is
in
terms of their insurance features. One can think of
long term debt as short term debt plus rescheduling insurance. Since our primitive result
one of under-insurance, a simple extension of the model presented
is
that agents undervalue the insurance component of long term debt
markets are underdeveloped. The result
is
in section 2
when domestic
shows
financial
akin to our previous result on the undervaluation
of international liquidity.
Consider the following model of debt maturity structure. Suppose that only a fraction
1
— tp,
where
<
i{>
<
taken on against this
1,
1
—
of
w
ip
of international liquidity
is
directly pledgeable to foreigners at date
the rest, tpw, can be seized by foreigners at date
monitoring cost per unit of
A
(1
it
—
of
e
>
at date
domestic firm has two choices. (A)
It
roll this
Diamond
model
is
short term debt.
20
Now
Debt that
is
suppose that
however doing so requires payment of a
2.
can take on one period debt up to the limit of
ip)w at the international interest rate of zero.
can
20
w
2,
is
l.
Then
at date
1, if it
needs the funds,
over and take on additional debt of f/w. However, the interest rate on this
(1991) develops a model of debt maturity structure based on liquidity risk.
related but the maturity structures
depends on aggregate
21
liquidity risk.
The
sketch of our
—
additional debt will be
(B)
It
to compensate the foreign lenders for bearing the monitoring cost.
e
can take on long term debt against the
full
w, in which case the foreign lenders will
always pay the monitoring cost to seize the additional tpw. In this case the interest rate on
the long term debt
of borrowing
\/l
is
it
1 will
1.
Thus, domestics face an upward sloping term structure
where two period debt
Let us suppose that
date
+e—
ij,
<
e
<
—
is
g
more expensive than one period
E
With option A, only
-
down
take on the additional debt and draw
debt.
firms that are distressed at
ipw, since only these firms will find
worthwhile to borrow at the high interest rate of e. Alternatively, with option B,
will
have borrowed against their extra-collateral at date
corresponding international funds at date
1
below
e if
as before,
is
term debt and
is
The
latter option is
is
only
ij,
is
A—
1
>
e.
which could well be
illiquid.
In the scenario outlined above, the equilibrium
service of long
intact firms will sell the
an extra unit of liquidity
that the return to intact firms
domestic financial markets are
and the
to the distressed firms.
clearly socially preferable, since the social value of
The problem,
0,
firms
all
is
one where no firm values the insurance
willing to pay the high interest rate. This
comes at great
cost in terms of the aggregate supply of international liquidity since a central planner
would
on the long term debt.
prefer that firms take
Since a sterilization with illiquid bonds lowers
short versus long term debt even more.
i^, it
As a
distorts private incentives to contract
result,
symptom
another
of backfiring
is
that firms shift their external borrowing towards short term debt and the capital inflow-
composition problem
Final
6
is
worsened. 21
Remarks
Emerging economies have poorly developed
dependence on external resources,
Coupled with their large
financial markets.
this deficiency
is
a source of constant instability. In this
paper we have shown that the same deficiency complicates the outcome of central bank's
interventions. In particular,
when
financial
markets are underdeveloped, sterilizations
may
backfire and increase rather than decrease a country's external vulnerability.
Does
this
whether to
mean
sterilize
highlighted with
21
Of
that central banks should never sterilize? Not necessarily. In deciding
one must weigh the importance of the liquidity mechanism we have
more
traditional ingredients in
its favor. First,
course, in reality the maturity decision, and hence the impact of sterilization, are smoother than in
our stark example. This concluding paragraph has this generalization as
it
the standard money-bonds
must be interpreted around the point
of switch.
22
its
background context, otherwise
—
mechanism of the Mundell-Fleming model, and second, the
elasticity of the international
supply of funds faced by the country during the sterilization period. Having said
former
not a very powerful mechanism
is
if
this,
the
countries have a "fear of floating," as Calvo and
Reinhart (2000) have recently documented for a large group of emerging markets. As for
the latter, by assuming perfect intertemporal arbitrage
of funds
is
we have assumed that the supply
locally very elastic during the sterilization period. This
sterilizations
during times of large capital inflows.
On
seems
small
realistic for
the other hand, this also suggests
that to succeed, sterilizations have to be extremely large at these times so as to break
intertemporal arbitrage and hence decouple date
Where should
backfiring be feared the most?
Economies with an intermediate
level of
to international financial markets,
On
the
cost of capital from future rates. 22
Our framework
offers a
.
simple answer:
domestic financial development, limited integration
and weak
fiscal discipline, are
the most vulnerable.
countries that have very limited financial development can
first characteristic,
never borrow and leverage enough for the international financial constraint to become a
serious issue.
Those that have very developed domestic
financial markets,
are sufficiently liquid so as to approximate the Ricardian
section
On
3.
financial
the second characteristic,
if
the country
on the other hand,
benchmark that we discussed
in
well integrated to international
is
markets then there should be no difference between international and domestic
financial constraints, so that the effects
the third characteristic, lack of
we have
fiscal discipline is
highlighted would be absent. Finally, on
more
and shrink
likely to dissipate reserves
domestic liquidity through the tax burden that this implies.
In addition to these ingredients, our
sterilization
this
be longer
in maturity
mechanism required that the public bonds used
than the period where a
crisis
may
take place.
in
While
does not seem very restrictive since there are always public bonds outstanding at the
time of
crises, it
does say that since the bonds used in sterilization are typically short term,
our mechanism becomes more powerful in the proximity of
crisis.
Importantly, however,
emerging economies are now hard at work trying to lengthen the maturity structure of
23
While there are
public debt, including that of the bonds used in sterilization.
22
23
See this paper's precursor: Caballero and Krishnamurthy (2000a)
See
e.g.
Guidotti (2000)
— who
and the general guidelines on
who
to
many good
writes:
"...public
this
is
one of the architects of Argentina's liquidity management strategy
matter put together by the group of 22 shortly after the recent
debt management and banking regulation
compensate the natural maturity mismatch of the private
in
crises
emerging markets should be designed
sector. ..it also suggests avoiding the use of
short-term debt to sterilize capital inflows...." (p.30).
See also, Agenor (2000), p.
221.
In his
summary
of the policy advise on sterilizations, he writes:
"To
discourage speculative short-term capital flows, medium-term sterilization bonds are generally preferable to
23
reasons to implement this lengthening, our analysis points at a potentially serious drawback
of this strategy,
and suggests that
this
move should be accompanied by attempts
to increase
the liquidity of the secondary market for public bonds.
Beyond the problem
in the design of
of sterilization, this paper raises a series of issues that are relevant
an adequate international
liquidity
exploring this policy problem in more detail.
short-term securites...."
24
management
strategy.
We
are currently
References
The Economics of Adjustment and Growth, Academic Press, 2000.
[1]
Agenor,
[2]
Aizenman, Joshua, "Country Risk, Incomplete Information and Taxes on International
P.,
Borrowing," Economic Journal 99, March 1989, pp. 147-161.
[3]
Allen, Franklin
and Douglas Gale, "Limited Market Participation and
Volatility of
Asset Prices," American Economic Review 84(4), September 1994, pages 933-55.
[4]
Caballero, R.J. and A. Krishnamurthy, "International and Domestic Collateral Constraints in a
Model
of
Emerging Market
Crises," forthcoming in Journal of
Monetary
Economics, 2001.
[5]
Caballero, R.J. and A. Krishnamurthy, "International Liquidity
ization Policy in Illiquid Financial Markets,"
[6]
Steril-
7740, June 2000a.
Caballero, R.J. and A. Krishnamurthy, "Dollarization of Liabilities: Underinsurance
and Domestic Financial Underdevlopment,"
[7]
NBER WP #
Management:
Calvo, G.A., "The Perils of Sterilization,"
NBER WP #
7792, July 2000b.
IMF Staff Papers
38-4,
December
1991, pp.
921-926.
[8]
[9]
CM. Reinhart,
IMF Staff Papers
Calvo, G.A., L. Leiderman,
"Capital Inflows to Latin America:
Role of External Factors,"
40,
Calvo, G.A., and
CM.
March
Reinhart, "Fear of Floating"
The
1993, pp. 108-151.
mimeo
University of Maryland,
2000.
[10]
Corbo, V. and L. Hernandez, "Macroeconomic Adjustment to Capital Inflows: Lessons
from Recent Latin American and East Asian Experience," The World Bank Research
Observer 11-1, February 1996, pp. 61-85.
[11]
Diamond, D.W., "Debt Maturity Structure and Liquidity Risk," Quarterly Journal of
Economics, August 1991.
[12]
Diamond, D.W. and P.H. Dybvig, "Bank Runs, Deposit Insurance, and
Journal of Political
[13]
Economy
Liquidity,"
91, 1983, pp. 401-419.
Diamond, D.W. and R.G. Rajan, "Liquidity Shortages and Banking
mimeo, January 2001.
25
Crises,"
U.Chicago
[14]
Dooley, M., "Responses to Volatile Capital Flows: Controls, Asset-Liability Manage-
ment and Architecture," World Bank Conference on Capital Flows, Financial Crises
and
[15]
Policies, April 1999.
Frankel,
J. A.,
Money
"Sterilization of
Inflows:
Difficult (Calvo) or
Easy (Reisen)?"
Estudios de Economic,, 24(2), December 1997, pp. 263-285.
[16]
Guidotti,
P.E.,
"Toward a Liquidity Management Strategy
for
Emerging Market
Economies," mimeo October 2000.
[17]
Harberger,
A.,
"Lessons
J.Cuddington and G.Smith
Debtor-Country Managers
for
(eds.),
and Policymakers,"
in
International Debt and the Developing Countries.
Washington: World Bank, 1985.
[18]
Holmstrom, B. and
Political
[19]
Economy
J.
Tirole,
"Private and Public Supply of Liquidity," Journal of
106-1, February 1998, pp. 1-40.
Massad, C.A., "Statement by Mr. Carlos Massad, Governor
for Chile,
Representing the
Latin American Governors of the International Monetary Fund at the Annual Meetings
of the
World Bank and the International Monetary Fund," Hong Kong, September
23,
1997.
[20]
Montiel, P. "Policy Responses to Surges in Capital Inflows: Issues and Lessons," in
Private Capital Inflows to Emerging Markets, ed. by Calvo, Goldstein and Hochreiter,
IEE, 1996.
[21]
Montiel, P. and C. Reinhart,
fluence the
"Do Capital Controls and Macroeconomic
Volume and Composition
of Capital Flows?
Policies In-
Evidence from the 1990s,"
mimeo, March 1999.
[22]
Mundell, R.A., "Capital Mobility and Stabilization Policy under Fixed and Flexible
Exchange Rates," Canadian Journal of Economics and
Political Science, 1962, pp. 475-
485.
[23]
Sargent, T.J. and N.Wallace,
serve
[24]
Bank
Williamson,
"Some Unpleasant Monetary Arithmetic," Federal Re-
of Minneapolis Quarterly Review, 5(3), 1981, pp. 1-17.
J.,
"The Management of Capital
January- June 1995.
26
Inflows,"
Pensamiento Iberoamericano,
[25]
World Bank, Private Capital Flows
Integration,
[26]
Developing Countries: The Road
to
Financial
Oxford University Press, 1997.
Woodford, M. "Public Debt as Private Liquidity," American Economic Review, Papers
and Proceedings 80
[27]
to
May
The World Bank, East
1990, pp. 382-88.
Asia: The
Road
to Recovery,
27
September 1998.
A
Appendix
A.l
Let
Detailed programs, definitions and assumptions
=
i*
0.
There are three main assumptions we have made
Assumption
1
in the model:
(Non-observability of Production Shock)
The production shock
The
at date 1 is idiosyncratic.
identity of firms receiving the shock
is
private
information.
Assumption
A
2
(Domestic Borrowing Constraint)
domestic lender can only
be sure that a
firm will produce rk units of goods at date
production based on physical reinvestment at date
Assumption
1 is
Any
2.
excess
neither observable nor verifiable.
3 (Liquidity Bias)
Foreigners lend to domestic firms only against the backing ofw. Domestics lend against both
w
and
rk.
<
debt constraint with respect to foreigners of do./
This gives a date
takes on additional debt with foreigners, the date
doj
1
debt constraint
+ d\j < w
w. At date
The program
for
^
a distressed firm at date
1 is:
= max0 ;dl
f ]dl d
(PI)
V
s
s.t.
rk
l
-
(i)
and
(ii)
An
firm.
for
1
(in)
0k
(iv)
e
in taking
<
f$
i.
1
new investment must be
on debts of d\j and
d\.d-
contribution of
(P2)
Vi
s.t.
**>.%
=
which
max. Xld
is
fully
1
paid with the resources received by
Constraint
how much
has only one decision:
Suppose that the firm accepts claims at date
1
= dXJ +
are balance sheet constraints (net marketable assets greater than liabilities),
intact firm at date
making a date
a
!
(i)
while constraint (Hi) reflects that
the firm at date
j
1 is,
w + R(6)k-doj — dij — di d
dij+do.f<w
J^fp + di,f + d 0J < w + -^
(**)
Constraints
a firm
is:
Since domestics lend against rk the debt constraint for domestic lending at date
di.d
1, if
28
it
extend to the distressed
of x\,d (face value of date 2 goods) in return
financed with
+i p
purely technological.
finance will
w + Rk + x lt a do j + d{ f < w
i
(iv) is
S a hf
new
external debt d\
d
- d[j
,
f
,
.
Then,
Date
At date
problem.
distressed or intact.
a firm looking forward to date
0,
Thus the
decision at date
Market clearing
+ V )/2
(Vs
s.t.
Equilibrium.
itself as either
is,
maxM0|/
(PS)
can expect to find
1
%
do./
<w
c(k)
=
in the domestic debt
d
,/-
market at date
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:
D\.d
=
-^d\.d
Xl.d
=
yX\,d-
Therefore, market clearing,
D hd = X1<d
(10)
,
determines the cost of capital, if
An
equilibrium of this
respectively,
and prices
economy consists of date
if.
24
and date
decisions, (k,doj)
1
and (0,di.f,di
:
d,xi.d),
Decisions are solutions to the firms' problems (PI), (P2), and (P3)
given prices. At these prices, the market clearing condition (10) holds.
Let us
now study equilibrium
in
more
detail.
investment choices of the distressed firm given
Starting from date
(/c,do./).
would choose to save as many of its production units as
it
First, if
can. It
A >
1,
1,
and
consider financing
then the distressed firm
may borrow up
to
its
international
debt capacity,
dij
If
the
amount
= w -d
raised from international investors,
j.
(11)
w—
will
choose to do this as long as
cost of capital.
If
A—
1
>
if, or the return
the firm borrows fully up to
its
di, d
and
raise funds
right
24
hand
is
=
it
will issue
debt totalling,
rk,
(12)
^p
fact that at equilibrium prices,
29
for
shortfall. It
on restructuring exceeds the domestic
.
more than the borrowing need, the firm
Where we have used the
market to make up the
domestic debt capacity,
with which to pay for imported goods of
side of (11)
than the funds needed
do./, is less
restructuring, k, the firm will have to access the domestic debt
As long
is
as the
sum
unconstrained in
j^p =
d\j.
of yrfpits
and the
reinvestment
at date
1
and
all
domestic debt capacity (and perhaps
most
Intact firms can tender at
for
its
than the international debt capacity).
less
their excess international debt capacity of
w—
do.f in return
purchasing domestic debt. They will choose to do this as long as the return on domestic loans
exceeds the international rate, if
Assume
and
for
a
moment
intact firms lend as
which
is
In this case, the firm will borrow less than
production units will be saved.
is
0.
A—1 >
that
much
>
>
if
so that distressed firms borrow as
1
as they can. Then, in total the
A
directed to the distressed firms.
much
economy can import
necessary condition for
as they can,
w—
do./ goods,
production units to be saved
all
that,
-<w-d j.
We
(13)
shall refer to this constraint as the international liquidity constraint.
(13) binds, all
production units are saved. Since there
relative to domestic
demand
for funds, there
is
is
When
neither (12) nor
excess supply of funds from intact firms
no international
liquidity
premium, and if
is
equal
to the international interest rate (zero).
The
other extreme case
is
when both
(12)
and
(13) bind.
Equilibrium in the domestic debt
market requires that,
Since (12) binds, distressed firms borrow fully
purchase this debt with
all
rk
p
w-
1
market
is,
their debt capacity.
As
(13) binds, intact firms
of their excess funds. Solving for if, yields
.
That
up to
in this case the if is
-
1
>
0.
(14)
j
above the international interest rate
One
for scarce international liquidity.
transferable domestic resources
d
owned by
in order to clear
the domestic
half times the numerator in (14) corresponds to the
distressed firms.
Define the index of domestic illiquidity as the difference between the marginal profit of saving a
distressed production unit
and the domestic
sd
Equilibrium at date
of the two (domestic
on a
scenario,
1
liquidity constrained
= A - if -
.
When
(13) binds, this
classified
liquidity constraints are binding.
liquidity constraints are binding.
level, firms
according to which
In the text,
At the aggregate
with respect to foreigners; at the individual
30
is,
1,
can place the economy in one of four regions,
and international)
where both
interest rate of if
level,
we focused
the economy
is
are liquidity constrained
with respect to other domestics since they are selling
investment
real
is
is
constrained; domestic spreads are positive; and the domestic cost of capital of if
above the international interest
policy questions
of their domestic liquidity in aggregation;
all
we
rate.
This
is
the most interesting configuration for the prevention-
address.
Technical Assumption 1 (Conditions for Crisis)
Assume
that:
W^!+c(W^±W
A
+ a;
i
V
v
The assumption guarantees
on a case where both
(12)
and
1
+
//
1,
liquidity of the
A—
1
When
Liquidity.
V
/
>
if
V
<
and 6
+
1
cfc-'
i?
2A
This ensures focus
1.
is
both liquidity constraints are binding at
constrained inefficient.
the private sector to decrease (fc,d
The welfare gain from
economy.
>
+
(13) bind.
the decentralized equilibrium
improvement by forcing
V
that in equilibrium
Underprovision of International
date
<l lc-fi±^
2A
A
./),
central planner can effect a Pareto
thereby increasing the international
index of domestic
this intervention rises with the
illiquidity, Sd-
and
Proof: First let us rewrite (P3), substituting in the value function from (PI)
choice of (k, do./) result in date 2 resources (net of any contracted debt)
(w-4/)A +
the firm
is
distressed. This
is
invested at the project return of A.
invested at A.
If
the firm
The rk
program
(P4)
for 1
+
sum
~r li
direcly pledged to foreigners,
of domestic liquidity
-d
(R
s.t.
weighted
,
f )l
+ if +
for
is
sold at 1
and the proceeds
+if and
,
the proceeds
Rk.
for
+ r Tf^)k + (A + l+i^){w-d0J
w>
doj
c(k)
=
a central planner
d
who
directly chooses (k, do./) to
of utilities of agents in this economy.
a central planner
{PS)
To do
this
we simply
an expression that
{R + r)K +
s.t.
w>
c(K)
31
do./
= D 0J
2A(W-D
maximize the equally
substitute the expression
is,
maxKtDof
)
j
if, (14), into the objective of (P4), arriving at
program
of,
is,
max Mo/
Consider the program
is
date
intact, date 2 resources are,
is
(w
Thus the date
— do./)
because (w
A
-^pA
J-
if
(P2).
,f)
is
free of prices.
The
The
solutions to (P5) are the constrained efficient decisions of the economy.
between the programs (P4) and (P5)
of (P4)
we
is
The only
difference
in the objective. Subtracting the objective in (P5) from that
arrive at,
sJ-JL^K-iW-Doj]
At a given equilibrium,
this
term must be
zero.
But
apparent that individuals and the central
it is
planner value a marginal unit of international liquidity and domestic liquidity quite
Moreover, this misvaluation
The
first
is
differently.
directly proportional to Sd, the domestic illiquidity index.
order condition of (P5) gives,
c(A) =
-2zT'
while that of (P4) yields,
R + TT^l
c'(fc)
A. 2
A + 1 + if
Sovereign risk as justification for liquidity bias
In assumption
3,
we
stated that foreigners only hold claims on w. Claims on rk (corporate bonds)
and claims on domestic government bonds were assumed
one possible justification
Sovereign risk
is
Assumption
4 (Sovereign Risk)
The government suspends
•
to be part of domestic collateral only.
for this assumption.
convertibility at date 1
and imposes two
restrictions
on exchange:
Foreigners are prevented from transporting any date 2 goods from within the economy across
the border.
•
Any
international liquidity holdings of the government/central bank are released to the private
sector only on presentation of an invoice for imported goods.
At date
owed
1
the government will suspend convertibility and repudiate any date 2 claims directly
to foreigners. Foreigners holding claims directly
by suspension,
on export sector revenues
since they can seize these receivables directly.
on a domestic firm
a domestic agent
or a domestic
who
However a foreigner holding a claim
government bond has only one
has some international liquidity and
(w) are unaffected
exit.
choice.
That
He must
is
sell
these claims to
domestic claims are
inter-
nationally liquid only to the extent that they can be exchanged for the private sector's international
liquidity at date
1.
Liquidity Bias and Foreigner's Short Horizons.
to
domestic claims.
Any domestic claim
acquired at date
international liquidity are free of this short horizon.
32
Foreigners have short horizons with respect
will
always be sold at date
1.
Claims on
—
Proof: Suppose that the private sector holds
w — do.f =
0.
0,
the government controls
the government suspends convertibility at date
for
an imported good to
has
—
We
rK
domestic claims, and has
show that Bf =
of the international liquidity in the economy.
all
only releases reserves
1, it
when shown an
investment plans. The government from oale of B
fulfill
in equilibrium.
+ Bf
invoice
bonds at date
L of reserves.
r
Case
government bonds,
Suppose foreigners hold Bf government bonds.
When w — d$j =
If
B
Suppose that the foreigner goes to the central bank with Bf bonds and an invoice
I:
Then the government
jrfp import goods.
releases jrju reserves to the foreigner that
pay the cost of the import good. The foreigner must
Bf domestic
claims.
repudiated at date
2.
may be
This argument
Bf =
uses to
to the domestic distressed firm for
sell this
However, these claims have no value to foreigners since they
Thus,
it
for
will surely
be
0.
construed as unrealistic on grounds that foreigners could take the im-
ported goods and liquidate them outside the country for international
to deal with these issues. If there
is
any liquidation cost
liquidity,
but
its fairly
in this transaction, foreigners
easy
would prefer
not to hold domestic claims, since they must always bear this cost, while domestics never bear the
cost.
An
alternative objection
that firms could over-invoice for imported goods as they do rou-
is
That
tinely during periods of capital-controls.
is
they claim higher prices than actual ones
for their
goods, thereby getting their hands on more valuable international reserves. Suppose that a firm can
get
away with over-invoicing by a multiple of
claim receives
Mll \
-c\
import goods, and
be sold to a distressed firm,
to a foreigner to redeem
maximum
^^
J
for
some
reserves.
M>
M qT*<Si
A
1.
international reserves.
jf domestic claims.
The
Thus
the discount that foreigners
enough bonds so that iff
=
if
M^lpB{B + Bf)
Bf bonds can
sold
receive at
sell their
bonds
at.
-1
In equilibrium, foreigners will hold only
This means that a fraction
.
now be
let,
Bf
is
The import goods can
international reserves can
of the foreigner's bonds. Foreigners selling
lJ
This
firm that tenders one unit of domestic
promised away to foreigners by the private sector at date
0.
M
^
1
of international reserves can be
Over-invoicing creates a leak in the
system.
Case
II:
Suppose the foreigner goes to a domestic investor and
liquidity. Since
doj
=
_5 a
and then receives jt^u
import goods, hence the domestic
Finally
let
us
Bf bonds
for international
w, domestics have none, and can only get international liquidity by tendering
bonds to the central bank. Suppose a domestic tenders
goods of
offers
show that only
will
reserves.
be
left
B
bonds to the central bank, shows import
However these reserves
will exactly
pay
for the
with no liquidity to offer foreigners.
distressed firms will sell
tional reserves.
33
bonds to the central bank
for its interna-
Case
III:
Either distressed or intact firms can tender
rK + B — T
to the government for
1+j e/
of reserves. If intact firms tender, they onsell the imported goods to distressed firms in exchange
for
some
receives
is
of
rK + B — T
of distressed firms. Suppose an intact firm tenders one domestic claim,
it
Thus
it
*a import goods, which
indifferent
it sells
to the distressed firm for
between tendering and not. Assume that
Distressed firms receive
all
imported goods
it
1
+if domestic
claims.
does not.
totalling,
rK + B-T
1+ifThis
is
exactly equal to
firms tender, this
is
1-M J
which
is all
of the government's reserves.
an equilibrium.
34
Thus when only
distressed
5645 035
Date Due
Lib-26-67
MIT LIBRARIES
3 9080 02246 0775