FRBSF Economic Letter
1996-06 | February 9, 1996
Many countries in the world fix their exchange rate to that of another currency. In Africa, a number of countries have a currency in common the CFA Franc which they peg to the French franc; in Europe, the European Monetary System (EMS) links its members’currencies to the German Deutschemark; and in Asia and Latin America, several countries peg their currencies unilaterally to the U.S. dollar.
Countries do not often change the level at which the exchange rate is pegged, and when they do, it is typically as a result of a dramatic speculative attack. The CFA franc was devalued in 1993; repeated speculative attacks on the EMS in 1992 and 1993 forced two currencies to drop out of the system and others to devalue; fixed exchange rates all over Latin America were threatened in 1995 after the Mexican peso was devalued and then floated in December 1994.
When a currency is devalued, anyone who holds it loses money. Rational investors can seek to avoid these losses by selling domestic currency for foreign exchange in the expectation of a future devaluation. Thus an outflow of foreign exchange reserves is observed before most “typical devaluations.” To deter these reserve losses, governments frequently raise domestic interest rates, which increases the cost of selling domestic currency. But this is usually only a short-term solution. The only long-term solution to an unsustainable exchange rate peg is to fix the underlying source of the devaluation.
In this Weekly Letter, I attempt to develop a stylized picture of a devaluation of a pegged but adjustable exchange rate. I ask: Is there a typical devaluation? If so, what does it look like? Are there standard “early warning signs” that signal a coming devaluation? Is there a typical “state of the economy” that coincides with a devaluation? For instance, are devaluations caused by loose monetary or fiscal policy? And how do countries and their policy authorities typically respond to devaluations?
If devaluations are preceded by a standard set of economic conditions, then these “early warning signs” of pending danger would be of intrinsic interest to both policy authorities and private investors. They certainly would have been useful before the 1994 Mexican devaluation and the 1992 attacks on the EMS!
I address these questions using a quarterly data set covering the period from the beginning of 1959 through 1993 for 20 industrial countries: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, UK, and US.
I examine nine variables change in foreign exchange reserves, current account balance, budget deficit/surplus, money growth, interest rates, stock market index, inflation, unemployment and real output growth. The data are mostly taken from the International Monetary Fund’s International Financial Statistics. I used the IMF’s annual report on Exchange Arrangements and Exchange Restrictions to compile a list of officially declared devaluations.
The variables are usually transformed into differential percentage changes by taking differences between domestic and German annualized first-differences of natural logarithms, multiplied by 100. Current accounts and government budgets are measured as percentages of domestic output; interest and unemployment rates are in percentages.
Figure 1 contains nine “small multiple” panels. Each small graph illustrates the behavior of a single variable for a four-year window around the time of devaluations, comparing its behavior with that of a control group of observations in which no devaluation or speculative attack occurred; that is, they are “event studies.” The top left panel, for example, focuses on how foreign exchange reserves behaved around the time of devaluation; specifically, it shows how this variable deviated from its “typical” value (that is, when no speculative attacks occurred) beginning eight quarters before devaluation, continuing through the actual event (marked with a vertical line), and ending eight quarters after devaluation.
The control group excludes observations when devaluations, flotations, and changes in the widths of exchange rate bands occurred. It also excludes periods when speculative attacks were warded off by central banks, either through interest rate increases or through declines in international reserves. I also exclude observations within a two-sided one-quarter window of each event and crisis to avoid double-counting. Although there are 81 (potentially non-independent) devaluations in the data set, the number of observations that underlies an observation in any individual panel may be lower because of missing data. In this sense, the samples are not directly comparable across panels.
Average values are highlighted in the panels; a band of plus and minus two standard deviations is also provided to illustrate the extent of variation.
The patterns in Figure 1 (this file requires Adobe Acrobat) make sense. Reading across the first row, one sees a steady loss of foreign exchange reserves for several quarters prior to devaluations. These are associated with persistent weakness in the current account, which is about 2 percentage points of GDP in greater deficit at devaluations than in periods of tranquillity. (Positive values for the current account and budget balances indicate surpluses.) Following devaluation, these patterns are reversed. Reserve losses slow and end after two post-devaluation quarters. The current account turns around visibly, though more sluggishly.
The next two panels provide evidence on the policy variables that may be associated with these trends. The government budget deficit is also significantly higher before and at the time of devaluations, compared to periods of tranquillity. There is some evidence that devaluing countries run larger deficits (as always, relative to Germany) than do countries in the control group, although the two-standard-deviation bands suggest that this differential is barely significant statistically. The money supply grows faster prior to devaluations than in tranquil periods; those growth rate differentials decline (at least temporarily) after devaluation.
Short-term interest rates are higher than in the control group over the eight quarters leading up to devaluation, as if a positive probability were attached to the change in the exchange rate. As the event gets closer and probabilities of devaluation are refined, the interest rate rises significantly in anticipation of the devaluation. Stock prices are significantly lower in the period leading up to devaluation, presumably reflecting these higher interest rates. Interest rates do not decline substantially after the actual devaluation. This suggests that devaluation has credibility costs and that markets expect further subsequent attacks, requiring the monetary authorities to maintain high interest rates after the devaluation. Devaluation is, however, good for expectations of profitability: stock prices rise in the wake of the event.
The rate of CPI inflation bears the expected relationship to money growth: it is faster, by 2 or 3 percentage points per quarter, in countries about to devalue than in the control group. Unemployment is also higher than average in the years surrounding devaluations. Only output growth does not differ significantly around devaluations compared to periods of tranquillity.
Overall, these patterns are consistent with standard economic models of the causes and effects of devaluation. They suggest that countries generally devalue in response to external imbalances (falling reserves, current account imbalances, poor competitiveness, and so forth), rather than internal problems (though unemployment is high). These external imbalances are associated with unusually expansionary monetary and fiscal policies. Governments appear to react well to devaluations, tightening monetary and fiscal policies in order to lock in competitiveness gains.
There is a “typical” devaluation; all devaluations are alike in the sense that they have a number of similar characteristics, namely, losing international reserves, experiencing a current account deficit, and suffering from depressed stock prices, high interest rates, inflation, and unemployment. Devaluations are preceded by expansionary monetary and fiscal policies, which economists also think of as causes of devaluations. Just as importantly, devaluations appear to work in the sense that they correct external imbalances while apparently encouraging governments to correct the underlying policy imbalances that appear to have caused them in the first place.
Of course, not all governments react to the circumstances that precede devaluations by actually devaluing. They have other alternatives. They can change some monetary or fiscal policy instrument (other than the level of the pegged exchange rate); for instance, they can change the width of the exchange rate band, they can impose capital controls, and they even can abandon the exchange rate peg altogether. Comparing all these alternatives simultaneously is a subject worthy of much study, and one I pursue (with Eichengreen and Wyplosz) elsewhere.
Andrew K. Rose
UC Berkeley and FRBSF Visiting Scholar
Eichengreen, B., A. K. Rose, and C. Wyplosz. 1995. “Exchange Market Mayhem: The Antecedents and Aftermath of Speculative Attacks.” Economic Policy, pp. 251-312.
Opinions expressed in FRBSF Economic Letter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System. This publication is edited by Sam Zuckerman and Anita Todd. Permission to reprint must be obtained in writing.
Please send editorial comments and requests for reprint permission to
Attn: Research publications, MS 1140
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120