Description
Developing countries fall into international financial crises for a variety of reasons, including fiscal profligacy, exchange rate mismanagement, international financial shocks, financial liberalization, and weaknesses in the domestic banking sector.
Alternative Approaches to Financial Crises
in Emerging Markets
Jeffrey D. Sachs
Abstract
Developing countries fall into international financial
crises for a variety of reasons, including fiscal profligacy,
exchange rate mismanagement, international financial shocks,
financial liberalization, and weaknesses in the domestic banking
sector. Market expectations may play an independent role in a
financial crisis, by triggering a self-fulfilling financial
panic. International public policy should be aimed first and
foremost at avoiding financial crises, but must also be prepared
to ameliorate financial crises after they begin.
Despite ample experience with financial crises in the past
decade, there are still serious differences of opinion with
regard to best means of their avoidance, and their proper
management once they occur. These difference relate to the
appropriate roles of exchange rate policy, banking policy, fiscal
policy, and the international institutions. The purpose of this
paper is to review the lessons of the past decade, in order to
draw some policy conclusions for developing country governments
and for international institutions such as the IMF, World Bank,
and the Bank for International Settlements.
Jeffrey D. Sachs is Director of the Harvard Institute for
International Development (HIID) and Galen L. Stone Professor of
International Trade, Harvard University. The paper was prepared
at background for discussion during meetings in Basel, SZ,
December 9-10, 1995.
1
Alternative Approaches to Financial Crises
in Emerging Markets
Jeffrey D. Sachs
I. Avoiding international financial crises
At the risk of gross analytical oversimplification (and in
lieu of formal models), I will focus on three main types of
international financial crises that plague emerging market
economies:
(1) fiscal crises: the Government abruptly loses the ability to
role over foreign debts and to attract new foreign loans,
possibly forcing the government into rescheduling or default of
its obligations;
(2) exchange crises: market participants abruptly shift their
demands from domestic-currency assets to foreign-currency assets,
depleting the foreign exchange reserves of the central bank in
the context of a pegged exchange rate system;
(3) banking crises: commercial banks abruptly lose the ability to
role over market instruments (CDS) or to meet a sudden withdrawal
of funds from sight deposits, thereby throwing the banks into
illiquidity and possibly insolvency.
While these three types of crises are logically distinct, and in
some cases come in a “pure” form, they often arrive in
combination, since common underlying shocks or market
expectations are likely to operate simultaneously in the market
for government bonds, the foreign exchange market, and the market
for bank assets.
For each type of financial crisis, we can find cases of
four distinct triggering mechanisms. The first is an exogenous
shock to markets that causes market agents to reassess the
ability of the government, the central bank, or the commercial
banks, to meet various intertemporal commitments. For example,
the collapse of the term of trade of an oil-rich country may call
into question the debt-servicing capacity of the government,
thereby reducing its creditworthiness. Or a terms-of-trade
collapse might lower domestic prices, in turn undermining the
solvency of the banking system, and thereby provoking a banking
crisis. Frequently, a rise in world interest rates undermines
the ability of the government or the banks to obtain further
loans.
The second possible cause of financial crisis is an
inadvertent policy shock, in which a policy reform in one market
2
triggers an adverse market reaction in another part of the
economy. For example, money market liberalization might cause a
rise of money market interest rates which in turn leads to
disintermediation of the commercial banks, and to a banking
crisis. Alternatively, if the government was borrowing at low
domestic interest rates, internationalization of financial
markets might cause a hardening of terms for government
borrowing, leading to a fiscal crisis.
The third possible cause of crisis is the exhaustion of
borrowing limits, whether by the government, the central bank, or
the commercial banks. All market borrowers have borrowing limits
determined by their intertemporal solvency and institutional
constraints (e.g. prudential limits on borrowing that govern the
borrowers and the creditors). In some cases, these limits
produce “boom-bust” cycles, in which the borrower incurs debts
very quickly up to the borrowing limit, and then suddenly is
unable to borrow further beyond that point. This may lead to a
crisis of the borrower itself (if the arrival at the borrowing
limit is not properly anticipated), or to a crisis among other
borrowers in the economy, if the sudden cutoff of funding leads
to a sharp shift in market conditions.
Suppose, for example, that after the end of capital
controls, a country’s commercial banks borrow very heavily in
international capital markets, until their borrowing limit is
reached. Initially the borrowing will lead to a boom in domestic
spending, as the banks intermediate a large flow of foreign
loans. This initial boom will also tend to produce a real
exchange rate appreciation. Once the banks are fully borrowed up,
the inflows stop; the domestic credit markets tighten; domestic
demand declines; and their is almost surely the need for an
exchange rate depreciation, to reverse the preceding
appreciation. This kind of boom-bust cycle in bank borrowing
abroad contributed to the onset of financial crises in Argentina
and Mexico in the past year.
The fourth possible cause of crisis is a self-fulfilling
panic. In almost all financial markets, asset prices may be
subject to multiple “rational” equilibria, including highly
adverse equilibria such as government bond crises, foreign
exchange crises, or banking crises. These types of panics are
well known and have been discussed and debated for nearly two
centuries, even if there remain important differences of opinion
about the likelihood and frequency of self-fulfilling panics.
Consider three cases of panic, one for each of our paradigmatic
financial crises.
A self-fulfilling fiscal crisis emerges when the markets
“rationally” expect the default of an illiquid, but solvent,
government borrower (see Sachs, 1984, and Sachs, Tornell and
3
Velasco, 1994). Suppose that the government has a solvency limit
of $100 billion (equal to the discounted value of future taxation
that can be devoted to debt servicing), and a current debt of $50
billion in short-term liabilities. Suppose also that the maximum
debt-servicing capacity in any one year is far below $50 billion.
Individual creditors will be happy to roll over their claims on
the government, and even to increase their exposure, if they are
confident that other creditors will continue to lend. If a panic
begins, however, in which each creditor believes that the other
creditors will withdraw their short-term claims, then the
individual creditor will also call in his claims. The government
will find itself unable to borrow, and may be pushed into default
since it is unable to come up with $50 billion in short-term
revenues.
A self-fulfilling foreign exchange crisis can similarly
arise (see Obstfeld, 1995). Suppose that the government is
defending a mildly overvalued exchange rate, trading off higher
unemployment for lower inflation in the short term. An exchange
rate attack begins, in which market participants believe that the
Central Bank will choose to devalue in order to reduce short-term
unemployment. In order to defend the exchange rate, the Central
Bank would have to raise interest rates sharply, which would do
grave damage to the banking system and the real economy.
Therefore, the Central Bank decides to forego a defense of the
currency, and resets the rate at a more competitive level in line
with the sudden shift in market expectations. In this way, there
may be two rational market expectations: no devaluation (and no
panic), or panic followed by devaluation.
A self-fulfilling banking panic is the most familiar of
these cases (see Dybvig and Diamond, 1983). A bank’s business is
to engage in maturity transformation, taking short-term
liabilities and long-term assets. This maturity transformation
exposes banks to the possibility of a bank run. Individual
depositors begin a panicked withdrawal of funds on the belief
that other depositors are also withdrawing their funds. Each
depositor knows that the bank lacks the funds in the short-run to
meet all demands for withdrawals. In this way, an otherwise
solvent bank can be pushed into insolvency.
Based on theory and recent experience, can we draw certain
policy prescriptions about how to avoid financial crises, other
than the obvious ones of fiscal discipline, realistic exchange
rate policy, banking supervision, and the like? I think the
answer is yes, with the proviso that a full treatment of the
issues here requires a vastly more detailed analysis and
argumentation. There are, nonetheless, several general lessons
that can be gleaned from recent cases.
4
1. Strictly pegged exchange rates raise the risks of financial
crises, and therefore should be used only in very specific
circumstances.
Consider the various sources of financial crises. Adverse
external shocks, e.g. a fall in the terms of trade, a rise in
world interest rates, a shift in world demand against a country’s
exports, tend to require a real depreciation that is extremely
difficult to achieve through internal deflation (as opposed to
currency depreciation). Similarly, the exhaustion of external
borrowing limits (e.g. when a government has accumulated a heavy
load of foreign debt) typically requires the shift from current
account deficit to current account surplus, which in turn
requires a real exchange rate depreciation.
Markets know that pegged exchange regimes are fragile in the
face of adverse external shocks. Market participants are thus
are likely to speculate against pegged rates in the face of such
shocks. Either the defense of the pegged rate will prove very
costly in the face of such speculation (as in Argentina in 1995),
or more likely, the market pressures will lead to the abandonment
of the pegged rate (as in several ERM countries in 1992, or
Mexico in 1994). See Obstfeld and Rogoff, 1995, for similar
conclusions.
Perhaps less obviously, self-fulfilling panics are also much
easier to handle, or can be obviated entirely, by floating
exchange rate regimes compared with pegged rate regimes.
Consider the case in which the government owes debts of 50
billion in local currency (e.g. pesos), with a solvency limit of
100 billion pesos. We noted earlier that a self-fulfilling panic
can ensue, even though the government is not at its solvency
limit, if each creditor believes that other creditors will stop
lending, thereby pushing the government into default. In a
floating exchange rate regime, the Central Bank can stop the
panic by standing behind the government, promising to offer
credits to the government so that the government can avoid
default on its market debts. The market knowledge that the
Central Bank stands ready to provide credit to the government to
prevent a default is enough to rule out a default as a rational
equilibrium. Under fixed rates, by contrast, the Central Bank
would be less able or willing to act as a lender of last resort,
unless, of course, the Central Bank has 50 billion pesos worth of
foreign exchange, or an international line of credit of that
amount.
For the same reason, a bank panic is much more likely in a
fixed exchange rate regime. In the event of a banking panic, the
Central Bank can act as a lender of last resort in a floating
regime, but not under a pegged rate regime without jeopardizing
the exchange rate peg itself. Wigmore (1987) has argued that the
5
failure of the Fed to protect the U.S. banking system in the
winter of 1932-3 was the result of the Fed’s fears that lender-
of-last-resort credits to the banks would undermine the link of
the U.S. dollar to gold. In this way, the commitment to the gold
standard opened the U.S. to a self-fulfilling banking panic.
Similarly, in 1995, Argentina’s strict currency board
arrangements opened it to the threat of a banking crisis, since
the Argentine Central Bank was unable to act as a lender of last
resort.
What, then, are the appropriate circumstances for a pegged
exchange rate regime for an emerging market. The first, and very
rare, case is a true optimal currency union with one or more
countries. In this case, a common central bank can act as lender
of last resort for the whole union. Note that many cases of
“permanent” exchange rate pegs, such as the CFA Franc in West
Africa, or the dollar peg in Liberia, were not even remotely
appropriate as optimal currency areas, and therefore produced
situations of prolonged currency overvaluation.
A second exceptional case for pegging may be an extremely
open and diversified economy with an extremely flexible labor
market. Such an economy can adjust to external shocks through
internal deflation if necessary, rather than depreciation. Also,
by virtue of diversification, such an economy may be less likely
to be hit by serious external shocks than an economy highly
specialized in a few export goods, particularly natural
resources. Hong Kong and Estonia, which both use currency boards,
are possible examples. Even in these cases, however, there is
still the need to maintain a lender of last resort mechanism. It
is rarely recognized, but extremely pertinent, that Hong Kong has
relied on the Bank of China to serve as a lender of last resort
in the event of banking crises.
A third exceptional case for pegging, at least in a
temporary manner, arises in the wake of hyperinflation or in the
case of the introduction of a new national currency, as in the
successor states of the Soviet Union and Yugoslavia. In both
cases, the economy tends to be under-monetized, with extremely
low ratios of M to GDP. A temporary peg of the exchange rate
serves several purposes: it increases confidence in a fragile
currency; it stops a self-fulfilling flight from the new (or
newly stabilized) currency; and most importantly, it provides an
automatic mechanism for remonetization of the economy -- through
the balance of payments. Thus, after Estonia pegged the Kroon to
the deutschemark, Estonia ran a significant balance of payments
surplus, as Estonia households and firms repatriated foreign
exchange in order to build up money balances in the new currency.
Foreign reserves and the money supply rose by more than 10
percent of GDP in the first two years.
It is at least arguable that Argentina presents a unique case.
1
After 40 years of chronic high inflation, punctuated by episodes of
hyperinflation, the economic team in 1991 judged that only the straightjacket
of a currency board could break the historical cycle of monetary disarray and
the virtual complete absence of credibility of stabilization policies. The
currency-board arrangements have succeeded remarkably in reducing Argentina’s
inflation to low single-digit rates, indeed making Argentina one of the lowest
inflation countries in the world. Moreover, economically policy has become
directed to vitally needed “real” reforms: of labor markets, public
administration, social security, tax policy, public ownership, and other
areas.
Thus, the undoubted costs of Argentina’s monetary straightjacket,
including the banking panic of 1995 and Argentina’s high labor costs in
international terms, must be balanced against the historic break in its deeply
disfunctional monetary policies over four decades. No other country failed
for so long to find other institutional bases (e.g. central bank independence,
pegged but adjustable exchange rates, etc.) to limit inflationary pressures.
For that reason, few if any other countries should take Argentina’s gamble on
a strict currency board arrangement. The long-term consequences of
Argentina’s own bet on the currency board system are still to be seen.
6
History has shown that the stabilization of a
hyperinflation, and the introduction of a new currency, tend to
be less costly in terms of lost output, with the use of a pegged
rate rather than a floating rate (see Sachs, 1995, for further
discussion of this point). The floating rate deprives the
economy of the mechanism for automatic remonetization, and the
currency also seems to suffer from a lack of confidence relative
to a pegged rate. At the same time, however, it is very important
to switch to from the pegged rate to a more flexible exchange
rate arrangement (a float, crawling band, or some other flexible
arrangement) as soon as the remonetization is substantially
accomplished. If the pegged rate is maintained too long, then
the economy is subjected to the problems noted earlier:
inflexibility in the face of shocks (as in Mexico, 1994); and
vulnerability to panics (as in Argentina, 1995, in the aftermath
of the Mexican crisis).
1
2. The Central Bank should discourage the dollarization of
accounts in the domestic banking system. Similarly, the
government should avoid the dollarization of its own short-term
debts.
In many developing countries, particularly those recovering
from high inflation, central banks often encourage re-
intermediation of the banking system by allowing dollar-indexed
assets and liabilities. Even when these assets and liabilities
are roughly balanced in overall totals (as is often required),
there is usually maturity transformation, so that the dollar
liabilities are short-term while the assets are longer-term. The
main problem in this case is that the Central Bank is less able
to act as a lender of last resort, even under floating exchange
7
rates, since it must provide foreign exchange (dollars) in the
event of a precipitous withdrawal of dollar balances. In many
cases of dollar balances, therefore, dollar withdrawals are
frozen when a bank run ensues; eventually, the dollar-denominated
balances are then converted into domestic currency with some
degree of confiscation.
Similar logic militates against the reliance of governments
on short-term dollar denominated debts, such as the Mexican
Government’s issuance of tesobonos which played such havoc with
Mexico in late 1994. Mexican Government indebtedness was not
extraordinarily high in late 1994 (perhaps 35 percent of GDP),
but it was dollar denominated and very short term. Since dollar
debts were much higher than dollar foreign exchange reserves at
the end of 1994, a market panic could and did develop. Investors
in Mexican government debts refused to roll over the outstanding
tesobonos and Mexico was pushed to the brink of default. Because
the liabilities were dollar denominated, the Banco de Mexico was
unable to stand behind the government’s obligations via a line of
domestic credit to the government. There was, contrarily, much
less, if any, risk of default on the government’s peso-
denominated liabilities. (See Sachs, Tornell, and Velasco for
further details on the Mexican case).
3. As a corollary to (1) and (2), currency board arrangements
should be avoided, except in the case of very small, very open
economies. Even then, the Central Bank should maintain some
scope for lender-of-last resort actions, and should arrange
international lines of credit in advance in order to be able to
respond to a future banking crisis.
“Permanently” fixed exchange rates have almost always
provoked a serious crisis -- whether in West Africa, Liberia,
Panama, or more recently in Argentina (where the economy is now
suffering from 20 percent unemployment, with no respite in
sight). They are an invitation to chronic overvaluation, banking
crises, and often both together. Moreover, despite the claims
made by Hanke and Schuler, 1994, that currency board arrangements
run smoothly despite the lack of a lender-of-last resort
facility, even Hong Kong has had to rely on the Bank of China, in
conjunction with the Hong Kong and Shanghai Banking Corporation,
to act as a lender of last resort in the event of the failure of
Ka Wah Bank in 1995 (see Jao, 1992, p. 58 for details). And, of
course, in 1995, Argentina required an emergency international
loan package to help stave off an intense banking crisis (though
see the discussion in footnote 2).
8
4. The liberalization of controls on capital inflows should be
managed gradually, to avoid a boom-bust cycle.
Many financial crises in the past 10 years were directly
related to financial market liberalization, especially the
elimination of controls on international financial movements.
Crises linked to financial market liberalization include
Argentina, Mexico, Venezuela in Latin America, and Israel, Sweden
and Norway in Europe. Each case has important distinctive
features, but all exhibited a boom-bust cycle, in which a sharp
but temporary wave of capital inflow accompanied a pegged
exchange rate. When the inflow subsided, the exchange rate
needed to be devalued. The subsequent devaluation was often
delayed until after a serious macroeconomic crisis had
transpired.
The exact nature of the boom-bust cycle in capital inflows
is hard to characterize precisely (Are these cycles rational? Are
they based on myopic euphoria, which then bursts). In any case,
the basic contours of the cycle are clear. Once capital controls
are lifted, domestic banks and other financial intermediaries are
able to tap the international capital markets. A sizeable stock
of foreign debt is rapidly accumulated, until eventually the
banks and other intermediaries hit a borrowing limit, determined
by their capital base and institutional rules. When the limit is
hit, the borrowing slows dramatically.
This repeated experience probably justifies the decisions of
countries such as Chile and Korea to liberalize gradually, so
that domestic financial institutions can not suddenly take on
foreign debt. The borrowing limit is reached more gradually, so
that the boom-bust cycle is greatly damped if not eliminated.
The real exchange rate does not appreciate as much as in the case
of rapid liberalization, and the subsequent devaluation needed
after the slowdown in capital inflows is much less dramatic. A
reasonable sequencing of the liberalization of capital inflows
would probably allow immediately for free movements of foreign
direct investment (including repatriation of capital and
profits), followed in sequence by liberalization of portfolio
equity investment, long-term borrowing by non-financial
enterprises, short-term borrowing by non-financial enterprises,
and then finally, by short-term commercial bank borrowing.
5. The BIS prudential standards for commercial banks are usually
not sufficient for developing countries. More stringent capital-
asset ratios should be encouraged, as well as rigorous minimum
capital levels for banks.
Financial crises are stoked by undercapitalized banks, which
are more vulnerable to self-fulfilling panics and more prone to
excessively risky borrowing. (Raising the risk of the bank’s
While it is far beyond the scope of this paper, I should note that in
2
many developing countries a separate regulatory regime should be instituted
for micro-finance institutions, which are specialized in providing small,
uncollateralized loans to the poorest parts of the population (e.g. Grameen
Bank in Bangladesh).
9
portfolio effectively transfers wealth from the deposit insurance
fund, usually provided by the government, to the bank’s
shareholders). Most emerging markets lack effective institutions
for bank supervision, so that the book value of bank capital
often greatly overstates the amount of shareholder equity (e.g.
bad loans may not be written down expeditiously). Moreover, the
ownership structure of banks in emerging markets typically adds
to their riskiness. Non-financial companies own the banks, which
in turn lend to the companies that own them. Moreover, bank
licenses are frequently issued on the basis of party politics,
nepotism, or corruption. For all of these reasons, standard
prescriptions for developed economies, such as a minimum capital-
to-assets ratio of 8 percent, are likely to be grossly inadequate
in the emerging markets context.
2
II. Responding to International Financial Crises
Crisis prevention, alas, is only part of the story. Around
60 developing countries have experienced extreme financial crises
in the past decade, in the form of debt defaults, debt
reschedulings, or acute inflationary episodes. In the past two
years, serious financial crises have hit Argentina, Latvia,
Mexico, Turkey, and Venezuela, and lesser financial crises have
hit many of the economies in transition. Dozens of developing
country governments are still grappling with heavy burdens of
foreign indebtedness, and new proposals are in the works for
easing the burdens on the world’s poorest countries of an
overhang of multilateral debts.
The International Monetary Fund was set up in 1944 in part
to serve as a kind of international lender of last resort (ILLR),
especially in order to provide temporary financing to help member
governments maintain pegged exchange rates during a period of
internal adjustment. With the collapse of the pegged exchange
rate regime in 1971, that specific function has diminished (if
not quite disappeared), though the IMF still plays the role of
central arbiter of developing country financial crises, with
regard to advice, balance of payments lending, and coordinator of
debt reduction operations.
In the past two years (Sachs, 1994, 1995) I have argued that
the IMF’s procedures lack an adequate conceptual and
institutional basis, so that there is undue and costly
inefficiency in the IMF’s handling of international financial
crises, and mainly of its role as the ILLR. These arguments have
10
recently been echoed in Eichengreen and Portes (1995), who concur
with the need for revising the IMF’s ILLR functions. It is
analytically useful at this stage to examine the IMF’s functions
with respect to the three types of financial crises described in
the previous section: fiscal crises; foreign exchange crises; and
banking crises.
1. Fiscal Crises
In the past 15 years, dozens of developing country
governments have experienced extreme financial embarrassment,
ranging from acute illiquidity, leading to a temporary default on
international obligations, to insolvency, requiring a permanent
cancellation of some portion of the government’s foreign debt
obligations. (Notice that domestic obligations are rarely
defaulted, since domestic-currency debts can be met through
credits provided by the central bank, at the cost of indirect
partial repudiation via inflation). An elaborate, if not wholly
successful, institutional pattern has emerged to handle sovereign
financial distress, including IMF loans, Paris Club relief on
bilateral credits, and London Club relief on bank credits.
An efficient system for financial workouts would offer the
following features. First, it would reduce moral hazards, by
rewarding countries for maintaining their debt servicing and
punishing countries adequately for managing their affairs in a
manner which leads to default on debt obligations. Second, it
would ensure the effective functioning of the debtor government
at all stages of the financial workout, from the onset of the
crisis to its final resolution. Third, it would create
mechanisms for coordinating the actions of creditors, to achieve
an efficient arrangement for renegotiating the financial terms on
the problem debts.
Fifteen years of experience with the international debt
crisis has demonstrated that current institutional arrangements
do not meet these needs, even though the international system has
evolved over time in a more efficient direction. As a recent IMF
report (1995, p. 11) states, “Because there exists no well-
defined and accepted legal process that is applicable in such
cases, the process of debt resolution by involuntary
restructuring is necessarily ad hoc with an uncertain outcome.”
The implications of the lack of a legal framework be appreciated
by comparing the ad hoc international arrangements with the legal
structure that governs financial workouts under the bankruptcy
laws of the advanced economies. The U.S. Bankruptcy Code,
chapter 9 (municipalities) and Chapter 11 (corporations),
provides a particularly relevant contrast to the international
arrangements, since the these chapters of the U.S. Bankruptcy
Code are explicitly designed to handle cases of financial
restructuring rather than liquidation.
11
The U.S. Bankruptcy Code recognizes that efficient workouts
require a regulatory environment at three stages of the workout.
At the outset of insolvency (or at least the outset of creditor
recognition of insolvency), the Code provides for a automatic
standstill on debt servicing. Creditors are not allowed to
pursue legal remedies to seize assets or to force the payment of
debts. The law recognizes that each creditor, acting
individually, has the incentive to enter a “grab race” for
assets, to their mutual detriment. The automatic standstill
prevents the premature liquidation or impairment of the insolvent
entity, which would lower the overall value of the entity to the
collective detriment of the creditors.
The second stage occurs between the onset of the stay and
the exit from bankruptcy through reorganization. During this
interim period, the insolvent entity (whether a municipality
under Chapter 9 or a corporation under Chapter 11) will generally
need access to new working capital. The overhang of pre-
bankruptcy debt precludes a routine return to the capital
markets, and high transactions costs among the current creditors
normally precludes new lending from the existing creditors (since
each creditor would like the others to make the emergency loans).
The Bankruptcy Code solution is “debtor-in-possession (DIP)
financing,” through the assignment of priority to the repayment
of the new, emergency loans ahead of the pre-bankruptcy claims.
The bankruptcy court must approve the DIP financing, to ensure
that the new loans actually enhance the value of the entity
(thereby increasing the value of the pre-bankruptcy debts). Note
importantly that, unlike the IMF, the bankruptcy court does not
dispose of any money itself; the court’s ability to deliver
working capital lies in its power to assign priority to new
market borrowing by the insolvent entity.
The third stage of the workout is the final balance sheet
reorganization (debt reduction, debt service reduction, new
loans, debt-to-equity swaps, etc.), usually combined with an
operational reorganization (closure of loss-making units,
divestiture, change in management, etc.). The key role of the
bankruptcy law is to provide a negotiating framework which: (1)
brings together all of the parties; (2) establishes mechanisms
for cross-the-board settlements involving all classes of
creditors; and (3) discourages free-riding or holdouts by
individual creditors, and which pushes the process towards an
expeditious resolution. The main tricks to avoid free-riding and
holdouts are: non-unanimity among creditors in the confirmation
of the reorganization plan; and the possibility of a court-led
“cramdown” of a plan that is resisted by a particular group of
creditors. A plan is confirmed if it is accepted by two-thirds or
more of each creditor class by amount of claims, and by more than
one half of each creditor class by the number of claimants.
12
By these standards, the shortcomings in the international
system are clear. There is no automatic standstill. An
insolvent government typically faces legal harassment, or at
least the threat of legal harassment, by individual creditors for
months or years after the onset of the crisis. There are
continued threats to seize airplanes, ships, and bank accounts,
and to otherwise disrupt trade credit lines. The legal basis of
these threats usually does not subside until the final
confirmation of a Paris Club and London Club deal, and by that
time, individual creditors may have done great damage to the
debtor government’s effectiveness.
There is essentially no debtor-in-possession financing,
other than the IMF lending (and perhaps other official lending)
itself. Thus, the only available funds for short-term working
capital are typically in the form of internationally mobilized
taxpayer dollars under IMF supervision. (Early in the debt
crisis, the IMF orchestrated a few “involuntary lending” programs
of the existing creditors, but these loans were very hard to
arrange, and the IMF gave up by the mid-1980s. In the case of
Mexico in 1995, for example, there was no active attempt to round
up “involuntary” or voluntary private financing). The fact that
working capital depends on taxpayer dollars has important
political economy ramifications. The IMF is naturally extremely
hesitant to lend “other people’s money” into a highly visible
crisis. Therefore, there are typically very long delays between
the onset of insolvency and the start of IMF lending. During
this period, the debtor government is starved for working
capital, and often loses much of its capacity to govern. Russia
during 1992 and 1993 is a vivid case of a government starved for
working capital, thereby undermining the economic reform program
(see Sachs, 1995, for details).
Finally, there are no effective rules to ensure an
expeditious overall settlement that overcomes free-riding and
holdouts. The Paris Club and London Club have evolved norms of
behavior to keep the major creditors fairly united. Nonetheless,
there are still sharp divisions among various creditor classes
(bilaterals versus multilaterals versus commercial bank versus
suppliers; large banks versus small banks) which incite strategic
moves among the creditors, and which in extreme cases lead to
individual holdouts that significantly delay a debt rescheduling
or debt reduction operation. The process requires a large input
of time and energy from the U.S. Treasury and other leading
finance ministries. Large, high-profile countries (e.g. Poland,
Mexico, Argentina) get more attention and more adequate remedies
than smaller countries off of the world’s political radar screen.
On the basis of earlier writings (Sachs 1994, 1995) I have
been interpreted by others as recommending an “International
Chapter 11,” i.e. an international bankruptcy court. This is too
13
literal an interpretation of my critique of current arrangements.
I am recommending that we recognize and attempt to establish the
functional equivalents of the key bankruptcy-code mechanisms:
automatic standstills, priority lending, and comprehensive
reorganization plans supported by non-unanimity rules. We do not
need a literal bankruptcy court to move in the direction of such
mechanisms. We could begin with a clear statement of IMF
operating principles regarding each stage of the workout. We
could search for ways to establish emergency priority lending
from private capital markets, under IMF supervision. The IMF and
governments could recommend model covenants for inclusion in
future sovereign lending instruments that would allow for
emergency priority lending and for efficient renegotiations of
debt claims, should the future circumstances require.
2. Foreign Exchange Crises
There is rarely a role for IMF lending to support a
particular nominal exchange rate of a member country. As the
preceding analysis has stressed, pegged exchange rates are
probably ineffective, and unsupportable, in most circumstances,
especially in light of the high capital mobility now prevailing
in the international economy. Of course, individual central
banks may well have cause to intervene from time to time in their
own currencies, but there is much less case for internationally
coordinated interventions on behalf of a particular developing
country.
There are, however, exceptions to this rule. International
lending on behalf of a nominal exchange rate target makes sense
when a government is attempting to establish confidence in a new
national currency, or to reestablish confidence in an existing
currency that has suffered a bout of extreme inflation. In both
cases, money-to-GDP ratios tend to be very low, and there is the
risk of a self-sustaining collapse of the currency, in which
market panic induces a flight from the currency which in turn
incites inflation and further currency flight. A descent into
hyperinflation is at least a theoretical possibility, which
become more likely the more tenuous is the use of the national
currency in the domestic economy (as evidenced by extremely low
ratios of money to income). Note that before the Plan Real,
Brazil exhibited a ratio of high-powered money to GDP of less
than one percent. Similarly, Kyrgyzstan’s new currency, the Som,
had an extremely tenuous role in the economy upon its
introduction, with a ratio of M2 to GDP of around four percent in
1994.
In these cases, there is a strong argument for an
internationally provided stabilization fund, to help back the new
currency, or newly stabilized currency, for the first year or
two. The $1 billion Polish Stabilization Fund in 1990 had a
14
powerful impact in raising confidence in and political support
for Poland’s ambitious reform program. Estonia, fortunately, was
able to create its own stabilization fund to support the new
Kroon in 1992, using repatriated gold reserves that had been held
by Sweden and the U.K. since World War II. Other countries,
however, have not been so fortunate as to have the requisite
reserves or to win the international support for a stabilization
fund.
In most cases, the short-term pegging of the exchange rate
should be viewed as a temporary stage until confidence in the
currency is established and partial re-monetization of the
economy proceeds. Therefore, a pegged rate might last for around
one year, to be followed by a more flexible arrangement, such as
a crawling band exchange rate system. The only realistic cases
for a long-term pegged rate are in very small open economies,
such as the Baltic States, which may viably choose a currency
board arrangement.
3. Banking Crises
The basic international arrangement regarding banking crises
is that central banks are responsible for the oversight and
prudential regulation of their nation’s banks, whether those
banks are operating domestically or as subsidiaries in foreign
countries. This division of labor may break down, however, when
domestic banks have extensive short-term liabilities denominated
in foreign currencies. In this case, the Central Bank may need
substantial foreign exchange reserves to meet a sudden withdrawal
of funds denominated in foreign exchange. Lacking adequate
foreign reserves, the domestic central bank may be unable to play
the role of lender of last resort.
There is, at this point, no international regulatory regime
governing the use of foreign-exchange denominated accounts in
national banking systems. When Argentina’s dollar-based banking
system succumbed to crisis early in 1995, Argentina required an
emergency international infusion of funds to stabilize the
banking sector, since the Argentine Central Bank lacked the
instruments and adequate reserves to play lender of last resort
on its own. This was especially true as a result of the
currency-board rules of the game which prevail in Argentina.
Under the Argentine rules of the game, prevailing since April
1991, the Central Bank is enjoined from extending domestic
credit, even to illiquid commercial banks. And even if the
Central Bank were to violate its operating principals, the
market’s confidence in Argentina’s exchange rate peg would be
rapidly undermined. In the event, the Central Bank eased various
reserve requirements, which amounted to a modest extension of
credit to the banks. Reserves fell from some $18 billion to $14
billion in a couple of months. To staunch the continuing
15
hemorrhage of deposits from the banking system, Argentina
arranged an international package of official and private support
for the banks, on the order of $10 billion.
As described earlier, the most important way to reduce the
frequency of this kind of crisis would be for governments and
central banks to restrict the use of foreign currency deposits
within a domestic banking system. At a minimum, the regulators
should require that commercial banks’ short-term foreign
liabilities (deposits and CDS) should be matched by short-term
and liquid foreign assets. Of course, in a world of derivatives,
swaps, and other off-balance-sheet means of converting the
currency denomination of assets and liabilities, regulators will
occasionally be caught short, even if strong restrictions on open
positions are in place.
In addition to regulatory changes such as these, there will
be the need, from time to time, for emergency cross-border
lending to support banking systems under siege as the result of
the abrupt withdrawal of foreign-denominated funds. As in any
lender-of-last-resort operation, the governing principle for
emergency international support should be a combination of: ex
ante prudential standards to avoid moral hazard; strong
conditionality in the event of a bailout; timely lending to avert
or to stem a panic; and closure (or merger) of insolvent, as
opposed to illiquid, financial institutions. Probably the
greatest shortcoming in proceeding with a general international
system of support is the lack of adequate prudential supervision
in most developing countries, and the lack of detailed attention
at the level of the IMF and the BIS to the state of banking
regulation in the developing countries. This suggests that it is
a high priority for the BIS and IMF together to establish a far
more rigorous system of monitoring of prudential arrangements
(capital adequacy, reporting and disclosure, portfolio
diversification, risk ratings, reserve provisioning, off-balance-
sheet monitoring and risk management, deposit insurance) of
developing countries, as a prelude to enhanced international
facilities for emergency lending operations to banking systems in
distress.
16
List of References
Calvo, G. (1995). “Varieties of Capital Market Crises,”
unpublished manuscript, University of Maryland, April.
Diamond, D. And P. Dybvig (1983). “Bank Runs, Deposit Insurance
and Liquidity,” Journal of Political Economy, June.
Eichengreen, B. and R. Portes (1995). Crisis? What Crisis?
Orderly Workouts for Sovereign Debtors,” London: Centre for
Economic Policy Research.
Hanke, S. H., and K. Schuler (1994), “Currency Boards for
Developing Countries,” Sector Study Number 9, San Francisco:
International Center for Economic Growth.
International Monetary Fund (1995). International Capital
Markets: Developments, Prospects, and Policy Issues. Washington:
International Monetary Fund, August.
Jao, Y.C. (1992). “Monetary system and banking structure,” in
H.C.Y. Ho and L.C. Chau (eds.), The Economic System of Hong Kong,
Hong Kong: Asian Research Service, 1992.
Obstfeld, M. and K. Rogoff (1995), “The Mirage of Fixed Exchange
Rates,” NBER Working Paper No. 5191, July .
Sachs, J. (1994). “The IMF and Economies in Crisis,” presented
at London School of Economics, July 1994, forthcoming in LSE
publication.
Sachs, J. (1995). “Do We Need a Lender of Last Resort,”
unpublished manuscript, Frank D. Graham Lecture, Princeton
University, April.
Sachs, J., A. Tornell, and A. Velasco (1995). “Lessons from the
Mexican Peso Crisis: What have we Learned?”, HIID Discussion
Paper, forthcoming in Economic Policy, 1996.
Wigmore, B. 1987. "Was the Banking Holiday of 1933 a run on the
dollar rather than the banks?" Journal of Economic History, 47:
739-56.
doc_195241266.pdf
Developing countries fall into international financial crises for a variety of reasons, including fiscal profligacy, exchange rate mismanagement, international financial shocks, financial liberalization, and weaknesses in the domestic banking sector.
Alternative Approaches to Financial Crises
in Emerging Markets
Jeffrey D. Sachs
Abstract
Developing countries fall into international financial
crises for a variety of reasons, including fiscal profligacy,
exchange rate mismanagement, international financial shocks,
financial liberalization, and weaknesses in the domestic banking
sector. Market expectations may play an independent role in a
financial crisis, by triggering a self-fulfilling financial
panic. International public policy should be aimed first and
foremost at avoiding financial crises, but must also be prepared
to ameliorate financial crises after they begin.
Despite ample experience with financial crises in the past
decade, there are still serious differences of opinion with
regard to best means of their avoidance, and their proper
management once they occur. These difference relate to the
appropriate roles of exchange rate policy, banking policy, fiscal
policy, and the international institutions. The purpose of this
paper is to review the lessons of the past decade, in order to
draw some policy conclusions for developing country governments
and for international institutions such as the IMF, World Bank,
and the Bank for International Settlements.
Jeffrey D. Sachs is Director of the Harvard Institute for
International Development (HIID) and Galen L. Stone Professor of
International Trade, Harvard University. The paper was prepared
at background for discussion during meetings in Basel, SZ,
December 9-10, 1995.
1
Alternative Approaches to Financial Crises
in Emerging Markets
Jeffrey D. Sachs
I. Avoiding international financial crises
At the risk of gross analytical oversimplification (and in
lieu of formal models), I will focus on three main types of
international financial crises that plague emerging market
economies:
(1) fiscal crises: the Government abruptly loses the ability to
role over foreign debts and to attract new foreign loans,
possibly forcing the government into rescheduling or default of
its obligations;
(2) exchange crises: market participants abruptly shift their
demands from domestic-currency assets to foreign-currency assets,
depleting the foreign exchange reserves of the central bank in
the context of a pegged exchange rate system;
(3) banking crises: commercial banks abruptly lose the ability to
role over market instruments (CDS) or to meet a sudden withdrawal
of funds from sight deposits, thereby throwing the banks into
illiquidity and possibly insolvency.
While these three types of crises are logically distinct, and in
some cases come in a “pure” form, they often arrive in
combination, since common underlying shocks or market
expectations are likely to operate simultaneously in the market
for government bonds, the foreign exchange market, and the market
for bank assets.
For each type of financial crisis, we can find cases of
four distinct triggering mechanisms. The first is an exogenous
shock to markets that causes market agents to reassess the
ability of the government, the central bank, or the commercial
banks, to meet various intertemporal commitments. For example,
the collapse of the term of trade of an oil-rich country may call
into question the debt-servicing capacity of the government,
thereby reducing its creditworthiness. Or a terms-of-trade
collapse might lower domestic prices, in turn undermining the
solvency of the banking system, and thereby provoking a banking
crisis. Frequently, a rise in world interest rates undermines
the ability of the government or the banks to obtain further
loans.
The second possible cause of financial crisis is an
inadvertent policy shock, in which a policy reform in one market
2
triggers an adverse market reaction in another part of the
economy. For example, money market liberalization might cause a
rise of money market interest rates which in turn leads to
disintermediation of the commercial banks, and to a banking
crisis. Alternatively, if the government was borrowing at low
domestic interest rates, internationalization of financial
markets might cause a hardening of terms for government
borrowing, leading to a fiscal crisis.
The third possible cause of crisis is the exhaustion of
borrowing limits, whether by the government, the central bank, or
the commercial banks. All market borrowers have borrowing limits
determined by their intertemporal solvency and institutional
constraints (e.g. prudential limits on borrowing that govern the
borrowers and the creditors). In some cases, these limits
produce “boom-bust” cycles, in which the borrower incurs debts
very quickly up to the borrowing limit, and then suddenly is
unable to borrow further beyond that point. This may lead to a
crisis of the borrower itself (if the arrival at the borrowing
limit is not properly anticipated), or to a crisis among other
borrowers in the economy, if the sudden cutoff of funding leads
to a sharp shift in market conditions.
Suppose, for example, that after the end of capital
controls, a country’s commercial banks borrow very heavily in
international capital markets, until their borrowing limit is
reached. Initially the borrowing will lead to a boom in domestic
spending, as the banks intermediate a large flow of foreign
loans. This initial boom will also tend to produce a real
exchange rate appreciation. Once the banks are fully borrowed up,
the inflows stop; the domestic credit markets tighten; domestic
demand declines; and their is almost surely the need for an
exchange rate depreciation, to reverse the preceding
appreciation. This kind of boom-bust cycle in bank borrowing
abroad contributed to the onset of financial crises in Argentina
and Mexico in the past year.
The fourth possible cause of crisis is a self-fulfilling
panic. In almost all financial markets, asset prices may be
subject to multiple “rational” equilibria, including highly
adverse equilibria such as government bond crises, foreign
exchange crises, or banking crises. These types of panics are
well known and have been discussed and debated for nearly two
centuries, even if there remain important differences of opinion
about the likelihood and frequency of self-fulfilling panics.
Consider three cases of panic, one for each of our paradigmatic
financial crises.
A self-fulfilling fiscal crisis emerges when the markets
“rationally” expect the default of an illiquid, but solvent,
government borrower (see Sachs, 1984, and Sachs, Tornell and
3
Velasco, 1994). Suppose that the government has a solvency limit
of $100 billion (equal to the discounted value of future taxation
that can be devoted to debt servicing), and a current debt of $50
billion in short-term liabilities. Suppose also that the maximum
debt-servicing capacity in any one year is far below $50 billion.
Individual creditors will be happy to roll over their claims on
the government, and even to increase their exposure, if they are
confident that other creditors will continue to lend. If a panic
begins, however, in which each creditor believes that the other
creditors will withdraw their short-term claims, then the
individual creditor will also call in his claims. The government
will find itself unable to borrow, and may be pushed into default
since it is unable to come up with $50 billion in short-term
revenues.
A self-fulfilling foreign exchange crisis can similarly
arise (see Obstfeld, 1995). Suppose that the government is
defending a mildly overvalued exchange rate, trading off higher
unemployment for lower inflation in the short term. An exchange
rate attack begins, in which market participants believe that the
Central Bank will choose to devalue in order to reduce short-term
unemployment. In order to defend the exchange rate, the Central
Bank would have to raise interest rates sharply, which would do
grave damage to the banking system and the real economy.
Therefore, the Central Bank decides to forego a defense of the
currency, and resets the rate at a more competitive level in line
with the sudden shift in market expectations. In this way, there
may be two rational market expectations: no devaluation (and no
panic), or panic followed by devaluation.
A self-fulfilling banking panic is the most familiar of
these cases (see Dybvig and Diamond, 1983). A bank’s business is
to engage in maturity transformation, taking short-term
liabilities and long-term assets. This maturity transformation
exposes banks to the possibility of a bank run. Individual
depositors begin a panicked withdrawal of funds on the belief
that other depositors are also withdrawing their funds. Each
depositor knows that the bank lacks the funds in the short-run to
meet all demands for withdrawals. In this way, an otherwise
solvent bank can be pushed into insolvency.
Based on theory and recent experience, can we draw certain
policy prescriptions about how to avoid financial crises, other
than the obvious ones of fiscal discipline, realistic exchange
rate policy, banking supervision, and the like? I think the
answer is yes, with the proviso that a full treatment of the
issues here requires a vastly more detailed analysis and
argumentation. There are, nonetheless, several general lessons
that can be gleaned from recent cases.
4
1. Strictly pegged exchange rates raise the risks of financial
crises, and therefore should be used only in very specific
circumstances.
Consider the various sources of financial crises. Adverse
external shocks, e.g. a fall in the terms of trade, a rise in
world interest rates, a shift in world demand against a country’s
exports, tend to require a real depreciation that is extremely
difficult to achieve through internal deflation (as opposed to
currency depreciation). Similarly, the exhaustion of external
borrowing limits (e.g. when a government has accumulated a heavy
load of foreign debt) typically requires the shift from current
account deficit to current account surplus, which in turn
requires a real exchange rate depreciation.
Markets know that pegged exchange regimes are fragile in the
face of adverse external shocks. Market participants are thus
are likely to speculate against pegged rates in the face of such
shocks. Either the defense of the pegged rate will prove very
costly in the face of such speculation (as in Argentina in 1995),
or more likely, the market pressures will lead to the abandonment
of the pegged rate (as in several ERM countries in 1992, or
Mexico in 1994). See Obstfeld and Rogoff, 1995, for similar
conclusions.
Perhaps less obviously, self-fulfilling panics are also much
easier to handle, or can be obviated entirely, by floating
exchange rate regimes compared with pegged rate regimes.
Consider the case in which the government owes debts of 50
billion in local currency (e.g. pesos), with a solvency limit of
100 billion pesos. We noted earlier that a self-fulfilling panic
can ensue, even though the government is not at its solvency
limit, if each creditor believes that other creditors will stop
lending, thereby pushing the government into default. In a
floating exchange rate regime, the Central Bank can stop the
panic by standing behind the government, promising to offer
credits to the government so that the government can avoid
default on its market debts. The market knowledge that the
Central Bank stands ready to provide credit to the government to
prevent a default is enough to rule out a default as a rational
equilibrium. Under fixed rates, by contrast, the Central Bank
would be less able or willing to act as a lender of last resort,
unless, of course, the Central Bank has 50 billion pesos worth of
foreign exchange, or an international line of credit of that
amount.
For the same reason, a bank panic is much more likely in a
fixed exchange rate regime. In the event of a banking panic, the
Central Bank can act as a lender of last resort in a floating
regime, but not under a pegged rate regime without jeopardizing
the exchange rate peg itself. Wigmore (1987) has argued that the
5
failure of the Fed to protect the U.S. banking system in the
winter of 1932-3 was the result of the Fed’s fears that lender-
of-last-resort credits to the banks would undermine the link of
the U.S. dollar to gold. In this way, the commitment to the gold
standard opened the U.S. to a self-fulfilling banking panic.
Similarly, in 1995, Argentina’s strict currency board
arrangements opened it to the threat of a banking crisis, since
the Argentine Central Bank was unable to act as a lender of last
resort.
What, then, are the appropriate circumstances for a pegged
exchange rate regime for an emerging market. The first, and very
rare, case is a true optimal currency union with one or more
countries. In this case, a common central bank can act as lender
of last resort for the whole union. Note that many cases of
“permanent” exchange rate pegs, such as the CFA Franc in West
Africa, or the dollar peg in Liberia, were not even remotely
appropriate as optimal currency areas, and therefore produced
situations of prolonged currency overvaluation.
A second exceptional case for pegging may be an extremely
open and diversified economy with an extremely flexible labor
market. Such an economy can adjust to external shocks through
internal deflation if necessary, rather than depreciation. Also,
by virtue of diversification, such an economy may be less likely
to be hit by serious external shocks than an economy highly
specialized in a few export goods, particularly natural
resources. Hong Kong and Estonia, which both use currency boards,
are possible examples. Even in these cases, however, there is
still the need to maintain a lender of last resort mechanism. It
is rarely recognized, but extremely pertinent, that Hong Kong has
relied on the Bank of China to serve as a lender of last resort
in the event of banking crises.
A third exceptional case for pegging, at least in a
temporary manner, arises in the wake of hyperinflation or in the
case of the introduction of a new national currency, as in the
successor states of the Soviet Union and Yugoslavia. In both
cases, the economy tends to be under-monetized, with extremely
low ratios of M to GDP. A temporary peg of the exchange rate
serves several purposes: it increases confidence in a fragile
currency; it stops a self-fulfilling flight from the new (or
newly stabilized) currency; and most importantly, it provides an
automatic mechanism for remonetization of the economy -- through
the balance of payments. Thus, after Estonia pegged the Kroon to
the deutschemark, Estonia ran a significant balance of payments
surplus, as Estonia households and firms repatriated foreign
exchange in order to build up money balances in the new currency.
Foreign reserves and the money supply rose by more than 10
percent of GDP in the first two years.
It is at least arguable that Argentina presents a unique case.
1
After 40 years of chronic high inflation, punctuated by episodes of
hyperinflation, the economic team in 1991 judged that only the straightjacket
of a currency board could break the historical cycle of monetary disarray and
the virtual complete absence of credibility of stabilization policies. The
currency-board arrangements have succeeded remarkably in reducing Argentina’s
inflation to low single-digit rates, indeed making Argentina one of the lowest
inflation countries in the world. Moreover, economically policy has become
directed to vitally needed “real” reforms: of labor markets, public
administration, social security, tax policy, public ownership, and other
areas.
Thus, the undoubted costs of Argentina’s monetary straightjacket,
including the banking panic of 1995 and Argentina’s high labor costs in
international terms, must be balanced against the historic break in its deeply
disfunctional monetary policies over four decades. No other country failed
for so long to find other institutional bases (e.g. central bank independence,
pegged but adjustable exchange rates, etc.) to limit inflationary pressures.
For that reason, few if any other countries should take Argentina’s gamble on
a strict currency board arrangement. The long-term consequences of
Argentina’s own bet on the currency board system are still to be seen.
6
History has shown that the stabilization of a
hyperinflation, and the introduction of a new currency, tend to
be less costly in terms of lost output, with the use of a pegged
rate rather than a floating rate (see Sachs, 1995, for further
discussion of this point). The floating rate deprives the
economy of the mechanism for automatic remonetization, and the
currency also seems to suffer from a lack of confidence relative
to a pegged rate. At the same time, however, it is very important
to switch to from the pegged rate to a more flexible exchange
rate arrangement (a float, crawling band, or some other flexible
arrangement) as soon as the remonetization is substantially
accomplished. If the pegged rate is maintained too long, then
the economy is subjected to the problems noted earlier:
inflexibility in the face of shocks (as in Mexico, 1994); and
vulnerability to panics (as in Argentina, 1995, in the aftermath
of the Mexican crisis).
1
2. The Central Bank should discourage the dollarization of
accounts in the domestic banking system. Similarly, the
government should avoid the dollarization of its own short-term
debts.
In many developing countries, particularly those recovering
from high inflation, central banks often encourage re-
intermediation of the banking system by allowing dollar-indexed
assets and liabilities. Even when these assets and liabilities
are roughly balanced in overall totals (as is often required),
there is usually maturity transformation, so that the dollar
liabilities are short-term while the assets are longer-term. The
main problem in this case is that the Central Bank is less able
to act as a lender of last resort, even under floating exchange
7
rates, since it must provide foreign exchange (dollars) in the
event of a precipitous withdrawal of dollar balances. In many
cases of dollar balances, therefore, dollar withdrawals are
frozen when a bank run ensues; eventually, the dollar-denominated
balances are then converted into domestic currency with some
degree of confiscation.
Similar logic militates against the reliance of governments
on short-term dollar denominated debts, such as the Mexican
Government’s issuance of tesobonos which played such havoc with
Mexico in late 1994. Mexican Government indebtedness was not
extraordinarily high in late 1994 (perhaps 35 percent of GDP),
but it was dollar denominated and very short term. Since dollar
debts were much higher than dollar foreign exchange reserves at
the end of 1994, a market panic could and did develop. Investors
in Mexican government debts refused to roll over the outstanding
tesobonos and Mexico was pushed to the brink of default. Because
the liabilities were dollar denominated, the Banco de Mexico was
unable to stand behind the government’s obligations via a line of
domestic credit to the government. There was, contrarily, much
less, if any, risk of default on the government’s peso-
denominated liabilities. (See Sachs, Tornell, and Velasco for
further details on the Mexican case).
3. As a corollary to (1) and (2), currency board arrangements
should be avoided, except in the case of very small, very open
economies. Even then, the Central Bank should maintain some
scope for lender-of-last resort actions, and should arrange
international lines of credit in advance in order to be able to
respond to a future banking crisis.
“Permanently” fixed exchange rates have almost always
provoked a serious crisis -- whether in West Africa, Liberia,
Panama, or more recently in Argentina (where the economy is now
suffering from 20 percent unemployment, with no respite in
sight). They are an invitation to chronic overvaluation, banking
crises, and often both together. Moreover, despite the claims
made by Hanke and Schuler, 1994, that currency board arrangements
run smoothly despite the lack of a lender-of-last resort
facility, even Hong Kong has had to rely on the Bank of China, in
conjunction with the Hong Kong and Shanghai Banking Corporation,
to act as a lender of last resort in the event of the failure of
Ka Wah Bank in 1995 (see Jao, 1992, p. 58 for details). And, of
course, in 1995, Argentina required an emergency international
loan package to help stave off an intense banking crisis (though
see the discussion in footnote 2).
8
4. The liberalization of controls on capital inflows should be
managed gradually, to avoid a boom-bust cycle.
Many financial crises in the past 10 years were directly
related to financial market liberalization, especially the
elimination of controls on international financial movements.
Crises linked to financial market liberalization include
Argentina, Mexico, Venezuela in Latin America, and Israel, Sweden
and Norway in Europe. Each case has important distinctive
features, but all exhibited a boom-bust cycle, in which a sharp
but temporary wave of capital inflow accompanied a pegged
exchange rate. When the inflow subsided, the exchange rate
needed to be devalued. The subsequent devaluation was often
delayed until after a serious macroeconomic crisis had
transpired.
The exact nature of the boom-bust cycle in capital inflows
is hard to characterize precisely (Are these cycles rational? Are
they based on myopic euphoria, which then bursts). In any case,
the basic contours of the cycle are clear. Once capital controls
are lifted, domestic banks and other financial intermediaries are
able to tap the international capital markets. A sizeable stock
of foreign debt is rapidly accumulated, until eventually the
banks and other intermediaries hit a borrowing limit, determined
by their capital base and institutional rules. When the limit is
hit, the borrowing slows dramatically.
This repeated experience probably justifies the decisions of
countries such as Chile and Korea to liberalize gradually, so
that domestic financial institutions can not suddenly take on
foreign debt. The borrowing limit is reached more gradually, so
that the boom-bust cycle is greatly damped if not eliminated.
The real exchange rate does not appreciate as much as in the case
of rapid liberalization, and the subsequent devaluation needed
after the slowdown in capital inflows is much less dramatic. A
reasonable sequencing of the liberalization of capital inflows
would probably allow immediately for free movements of foreign
direct investment (including repatriation of capital and
profits), followed in sequence by liberalization of portfolio
equity investment, long-term borrowing by non-financial
enterprises, short-term borrowing by non-financial enterprises,
and then finally, by short-term commercial bank borrowing.
5. The BIS prudential standards for commercial banks are usually
not sufficient for developing countries. More stringent capital-
asset ratios should be encouraged, as well as rigorous minimum
capital levels for banks.
Financial crises are stoked by undercapitalized banks, which
are more vulnerable to self-fulfilling panics and more prone to
excessively risky borrowing. (Raising the risk of the bank’s
While it is far beyond the scope of this paper, I should note that in
2
many developing countries a separate regulatory regime should be instituted
for micro-finance institutions, which are specialized in providing small,
uncollateralized loans to the poorest parts of the population (e.g. Grameen
Bank in Bangladesh).
9
portfolio effectively transfers wealth from the deposit insurance
fund, usually provided by the government, to the bank’s
shareholders). Most emerging markets lack effective institutions
for bank supervision, so that the book value of bank capital
often greatly overstates the amount of shareholder equity (e.g.
bad loans may not be written down expeditiously). Moreover, the
ownership structure of banks in emerging markets typically adds
to their riskiness. Non-financial companies own the banks, which
in turn lend to the companies that own them. Moreover, bank
licenses are frequently issued on the basis of party politics,
nepotism, or corruption. For all of these reasons, standard
prescriptions for developed economies, such as a minimum capital-
to-assets ratio of 8 percent, are likely to be grossly inadequate
in the emerging markets context.
2
II. Responding to International Financial Crises
Crisis prevention, alas, is only part of the story. Around
60 developing countries have experienced extreme financial crises
in the past decade, in the form of debt defaults, debt
reschedulings, or acute inflationary episodes. In the past two
years, serious financial crises have hit Argentina, Latvia,
Mexico, Turkey, and Venezuela, and lesser financial crises have
hit many of the economies in transition. Dozens of developing
country governments are still grappling with heavy burdens of
foreign indebtedness, and new proposals are in the works for
easing the burdens on the world’s poorest countries of an
overhang of multilateral debts.
The International Monetary Fund was set up in 1944 in part
to serve as a kind of international lender of last resort (ILLR),
especially in order to provide temporary financing to help member
governments maintain pegged exchange rates during a period of
internal adjustment. With the collapse of the pegged exchange
rate regime in 1971, that specific function has diminished (if
not quite disappeared), though the IMF still plays the role of
central arbiter of developing country financial crises, with
regard to advice, balance of payments lending, and coordinator of
debt reduction operations.
In the past two years (Sachs, 1994, 1995) I have argued that
the IMF’s procedures lack an adequate conceptual and
institutional basis, so that there is undue and costly
inefficiency in the IMF’s handling of international financial
crises, and mainly of its role as the ILLR. These arguments have
10
recently been echoed in Eichengreen and Portes (1995), who concur
with the need for revising the IMF’s ILLR functions. It is
analytically useful at this stage to examine the IMF’s functions
with respect to the three types of financial crises described in
the previous section: fiscal crises; foreign exchange crises; and
banking crises.
1. Fiscal Crises
In the past 15 years, dozens of developing country
governments have experienced extreme financial embarrassment,
ranging from acute illiquidity, leading to a temporary default on
international obligations, to insolvency, requiring a permanent
cancellation of some portion of the government’s foreign debt
obligations. (Notice that domestic obligations are rarely
defaulted, since domestic-currency debts can be met through
credits provided by the central bank, at the cost of indirect
partial repudiation via inflation). An elaborate, if not wholly
successful, institutional pattern has emerged to handle sovereign
financial distress, including IMF loans, Paris Club relief on
bilateral credits, and London Club relief on bank credits.
An efficient system for financial workouts would offer the
following features. First, it would reduce moral hazards, by
rewarding countries for maintaining their debt servicing and
punishing countries adequately for managing their affairs in a
manner which leads to default on debt obligations. Second, it
would ensure the effective functioning of the debtor government
at all stages of the financial workout, from the onset of the
crisis to its final resolution. Third, it would create
mechanisms for coordinating the actions of creditors, to achieve
an efficient arrangement for renegotiating the financial terms on
the problem debts.
Fifteen years of experience with the international debt
crisis has demonstrated that current institutional arrangements
do not meet these needs, even though the international system has
evolved over time in a more efficient direction. As a recent IMF
report (1995, p. 11) states, “Because there exists no well-
defined and accepted legal process that is applicable in such
cases, the process of debt resolution by involuntary
restructuring is necessarily ad hoc with an uncertain outcome.”
The implications of the lack of a legal framework be appreciated
by comparing the ad hoc international arrangements with the legal
structure that governs financial workouts under the bankruptcy
laws of the advanced economies. The U.S. Bankruptcy Code,
chapter 9 (municipalities) and Chapter 11 (corporations),
provides a particularly relevant contrast to the international
arrangements, since the these chapters of the U.S. Bankruptcy
Code are explicitly designed to handle cases of financial
restructuring rather than liquidation.
11
The U.S. Bankruptcy Code recognizes that efficient workouts
require a regulatory environment at three stages of the workout.
At the outset of insolvency (or at least the outset of creditor
recognition of insolvency), the Code provides for a automatic
standstill on debt servicing. Creditors are not allowed to
pursue legal remedies to seize assets or to force the payment of
debts. The law recognizes that each creditor, acting
individually, has the incentive to enter a “grab race” for
assets, to their mutual detriment. The automatic standstill
prevents the premature liquidation or impairment of the insolvent
entity, which would lower the overall value of the entity to the
collective detriment of the creditors.
The second stage occurs between the onset of the stay and
the exit from bankruptcy through reorganization. During this
interim period, the insolvent entity (whether a municipality
under Chapter 9 or a corporation under Chapter 11) will generally
need access to new working capital. The overhang of pre-
bankruptcy debt precludes a routine return to the capital
markets, and high transactions costs among the current creditors
normally precludes new lending from the existing creditors (since
each creditor would like the others to make the emergency loans).
The Bankruptcy Code solution is “debtor-in-possession (DIP)
financing,” through the assignment of priority to the repayment
of the new, emergency loans ahead of the pre-bankruptcy claims.
The bankruptcy court must approve the DIP financing, to ensure
that the new loans actually enhance the value of the entity
(thereby increasing the value of the pre-bankruptcy debts). Note
importantly that, unlike the IMF, the bankruptcy court does not
dispose of any money itself; the court’s ability to deliver
working capital lies in its power to assign priority to new
market borrowing by the insolvent entity.
The third stage of the workout is the final balance sheet
reorganization (debt reduction, debt service reduction, new
loans, debt-to-equity swaps, etc.), usually combined with an
operational reorganization (closure of loss-making units,
divestiture, change in management, etc.). The key role of the
bankruptcy law is to provide a negotiating framework which: (1)
brings together all of the parties; (2) establishes mechanisms
for cross-the-board settlements involving all classes of
creditors; and (3) discourages free-riding or holdouts by
individual creditors, and which pushes the process towards an
expeditious resolution. The main tricks to avoid free-riding and
holdouts are: non-unanimity among creditors in the confirmation
of the reorganization plan; and the possibility of a court-led
“cramdown” of a plan that is resisted by a particular group of
creditors. A plan is confirmed if it is accepted by two-thirds or
more of each creditor class by amount of claims, and by more than
one half of each creditor class by the number of claimants.
12
By these standards, the shortcomings in the international
system are clear. There is no automatic standstill. An
insolvent government typically faces legal harassment, or at
least the threat of legal harassment, by individual creditors for
months or years after the onset of the crisis. There are
continued threats to seize airplanes, ships, and bank accounts,
and to otherwise disrupt trade credit lines. The legal basis of
these threats usually does not subside until the final
confirmation of a Paris Club and London Club deal, and by that
time, individual creditors may have done great damage to the
debtor government’s effectiveness.
There is essentially no debtor-in-possession financing,
other than the IMF lending (and perhaps other official lending)
itself. Thus, the only available funds for short-term working
capital are typically in the form of internationally mobilized
taxpayer dollars under IMF supervision. (Early in the debt
crisis, the IMF orchestrated a few “involuntary lending” programs
of the existing creditors, but these loans were very hard to
arrange, and the IMF gave up by the mid-1980s. In the case of
Mexico in 1995, for example, there was no active attempt to round
up “involuntary” or voluntary private financing). The fact that
working capital depends on taxpayer dollars has important
political economy ramifications. The IMF is naturally extremely
hesitant to lend “other people’s money” into a highly visible
crisis. Therefore, there are typically very long delays between
the onset of insolvency and the start of IMF lending. During
this period, the debtor government is starved for working
capital, and often loses much of its capacity to govern. Russia
during 1992 and 1993 is a vivid case of a government starved for
working capital, thereby undermining the economic reform program
(see Sachs, 1995, for details).
Finally, there are no effective rules to ensure an
expeditious overall settlement that overcomes free-riding and
holdouts. The Paris Club and London Club have evolved norms of
behavior to keep the major creditors fairly united. Nonetheless,
there are still sharp divisions among various creditor classes
(bilaterals versus multilaterals versus commercial bank versus
suppliers; large banks versus small banks) which incite strategic
moves among the creditors, and which in extreme cases lead to
individual holdouts that significantly delay a debt rescheduling
or debt reduction operation. The process requires a large input
of time and energy from the U.S. Treasury and other leading
finance ministries. Large, high-profile countries (e.g. Poland,
Mexico, Argentina) get more attention and more adequate remedies
than smaller countries off of the world’s political radar screen.
On the basis of earlier writings (Sachs 1994, 1995) I have
been interpreted by others as recommending an “International
Chapter 11,” i.e. an international bankruptcy court. This is too
13
literal an interpretation of my critique of current arrangements.
I am recommending that we recognize and attempt to establish the
functional equivalents of the key bankruptcy-code mechanisms:
automatic standstills, priority lending, and comprehensive
reorganization plans supported by non-unanimity rules. We do not
need a literal bankruptcy court to move in the direction of such
mechanisms. We could begin with a clear statement of IMF
operating principles regarding each stage of the workout. We
could search for ways to establish emergency priority lending
from private capital markets, under IMF supervision. The IMF and
governments could recommend model covenants for inclusion in
future sovereign lending instruments that would allow for
emergency priority lending and for efficient renegotiations of
debt claims, should the future circumstances require.
2. Foreign Exchange Crises
There is rarely a role for IMF lending to support a
particular nominal exchange rate of a member country. As the
preceding analysis has stressed, pegged exchange rates are
probably ineffective, and unsupportable, in most circumstances,
especially in light of the high capital mobility now prevailing
in the international economy. Of course, individual central
banks may well have cause to intervene from time to time in their
own currencies, but there is much less case for internationally
coordinated interventions on behalf of a particular developing
country.
There are, however, exceptions to this rule. International
lending on behalf of a nominal exchange rate target makes sense
when a government is attempting to establish confidence in a new
national currency, or to reestablish confidence in an existing
currency that has suffered a bout of extreme inflation. In both
cases, money-to-GDP ratios tend to be very low, and there is the
risk of a self-sustaining collapse of the currency, in which
market panic induces a flight from the currency which in turn
incites inflation and further currency flight. A descent into
hyperinflation is at least a theoretical possibility, which
become more likely the more tenuous is the use of the national
currency in the domestic economy (as evidenced by extremely low
ratios of money to income). Note that before the Plan Real,
Brazil exhibited a ratio of high-powered money to GDP of less
than one percent. Similarly, Kyrgyzstan’s new currency, the Som,
had an extremely tenuous role in the economy upon its
introduction, with a ratio of M2 to GDP of around four percent in
1994.
In these cases, there is a strong argument for an
internationally provided stabilization fund, to help back the new
currency, or newly stabilized currency, for the first year or
two. The $1 billion Polish Stabilization Fund in 1990 had a
14
powerful impact in raising confidence in and political support
for Poland’s ambitious reform program. Estonia, fortunately, was
able to create its own stabilization fund to support the new
Kroon in 1992, using repatriated gold reserves that had been held
by Sweden and the U.K. since World War II. Other countries,
however, have not been so fortunate as to have the requisite
reserves or to win the international support for a stabilization
fund.
In most cases, the short-term pegging of the exchange rate
should be viewed as a temporary stage until confidence in the
currency is established and partial re-monetization of the
economy proceeds. Therefore, a pegged rate might last for around
one year, to be followed by a more flexible arrangement, such as
a crawling band exchange rate system. The only realistic cases
for a long-term pegged rate are in very small open economies,
such as the Baltic States, which may viably choose a currency
board arrangement.
3. Banking Crises
The basic international arrangement regarding banking crises
is that central banks are responsible for the oversight and
prudential regulation of their nation’s banks, whether those
banks are operating domestically or as subsidiaries in foreign
countries. This division of labor may break down, however, when
domestic banks have extensive short-term liabilities denominated
in foreign currencies. In this case, the Central Bank may need
substantial foreign exchange reserves to meet a sudden withdrawal
of funds denominated in foreign exchange. Lacking adequate
foreign reserves, the domestic central bank may be unable to play
the role of lender of last resort.
There is, at this point, no international regulatory regime
governing the use of foreign-exchange denominated accounts in
national banking systems. When Argentina’s dollar-based banking
system succumbed to crisis early in 1995, Argentina required an
emergency international infusion of funds to stabilize the
banking sector, since the Argentine Central Bank lacked the
instruments and adequate reserves to play lender of last resort
on its own. This was especially true as a result of the
currency-board rules of the game which prevail in Argentina.
Under the Argentine rules of the game, prevailing since April
1991, the Central Bank is enjoined from extending domestic
credit, even to illiquid commercial banks. And even if the
Central Bank were to violate its operating principals, the
market’s confidence in Argentina’s exchange rate peg would be
rapidly undermined. In the event, the Central Bank eased various
reserve requirements, which amounted to a modest extension of
credit to the banks. Reserves fell from some $18 billion to $14
billion in a couple of months. To staunch the continuing
15
hemorrhage of deposits from the banking system, Argentina
arranged an international package of official and private support
for the banks, on the order of $10 billion.
As described earlier, the most important way to reduce the
frequency of this kind of crisis would be for governments and
central banks to restrict the use of foreign currency deposits
within a domestic banking system. At a minimum, the regulators
should require that commercial banks’ short-term foreign
liabilities (deposits and CDS) should be matched by short-term
and liquid foreign assets. Of course, in a world of derivatives,
swaps, and other off-balance-sheet means of converting the
currency denomination of assets and liabilities, regulators will
occasionally be caught short, even if strong restrictions on open
positions are in place.
In addition to regulatory changes such as these, there will
be the need, from time to time, for emergency cross-border
lending to support banking systems under siege as the result of
the abrupt withdrawal of foreign-denominated funds. As in any
lender-of-last-resort operation, the governing principle for
emergency international support should be a combination of: ex
ante prudential standards to avoid moral hazard; strong
conditionality in the event of a bailout; timely lending to avert
or to stem a panic; and closure (or merger) of insolvent, as
opposed to illiquid, financial institutions. Probably the
greatest shortcoming in proceeding with a general international
system of support is the lack of adequate prudential supervision
in most developing countries, and the lack of detailed attention
at the level of the IMF and the BIS to the state of banking
regulation in the developing countries. This suggests that it is
a high priority for the BIS and IMF together to establish a far
more rigorous system of monitoring of prudential arrangements
(capital adequacy, reporting and disclosure, portfolio
diversification, risk ratings, reserve provisioning, off-balance-
sheet monitoring and risk management, deposit insurance) of
developing countries, as a prelude to enhanced international
facilities for emergency lending operations to banking systems in
distress.
16
List of References
Calvo, G. (1995). “Varieties of Capital Market Crises,”
unpublished manuscript, University of Maryland, April.
Diamond, D. And P. Dybvig (1983). “Bank Runs, Deposit Insurance
and Liquidity,” Journal of Political Economy, June.
Eichengreen, B. and R. Portes (1995). Crisis? What Crisis?
Orderly Workouts for Sovereign Debtors,” London: Centre for
Economic Policy Research.
Hanke, S. H., and K. Schuler (1994), “Currency Boards for
Developing Countries,” Sector Study Number 9, San Francisco:
International Center for Economic Growth.
International Monetary Fund (1995). International Capital
Markets: Developments, Prospects, and Policy Issues. Washington:
International Monetary Fund, August.
Jao, Y.C. (1992). “Monetary system and banking structure,” in
H.C.Y. Ho and L.C. Chau (eds.), The Economic System of Hong Kong,
Hong Kong: Asian Research Service, 1992.
Obstfeld, M. and K. Rogoff (1995), “The Mirage of Fixed Exchange
Rates,” NBER Working Paper No. 5191, July .
Sachs, J. (1994). “The IMF and Economies in Crisis,” presented
at London School of Economics, July 1994, forthcoming in LSE
publication.
Sachs, J. (1995). “Do We Need a Lender of Last Resort,”
unpublished manuscript, Frank D. Graham Lecture, Princeton
University, April.
Sachs, J., A. Tornell, and A. Velasco (1995). “Lessons from the
Mexican Peso Crisis: What have we Learned?”, HIID Discussion
Paper, forthcoming in Economic Policy, 1996.
Wigmore, B. 1987. "Was the Banking Holiday of 1933 a run on the
dollar rather than the banks?" Journal of Economic History, 47:
739-56.
doc_195241266.pdf