A system of fixed exchange rates, or a system of exchange rate bands like the ERM, is usually maintained through the central bank’s foreign exchange market intervention using its foreign exchange reserves. Such regimes are subject to periodic speculative attacks, however: speculators take a position against a currency if they believe that the exchange rate is likely to devalue from its current parity soon enough to compensate them for the costs of speculation; high capital mobility has lowered these costs by greatly expanding the speculators’ range of alternative means of taking positions in the foreign exchange market.1
Speculators’ belief that a devaluation is likely may be self-fulfilling, since the attack motivated by this belief may exhaust the central bank’s foreign exchange reserves and leave no alternative but to devalue the currency. Market fundamentals, however, will partly determine an attack’s chance of success, and therefore its timing, size, and dynamics. This annex reviews the consensus in the economic literature on how a currency crisis is generated and the dynamics of a speculative attack on an exchange rate peg in an era of high capital mobility.2 The organizing principle used to develop these ideas centers on the pursuit of profits by market participants and their competition against each other to secure such profits. This section contains what may be characterized as the current macro-economic view of the dynamics of speculative attack—that is, without reference to the specifics of the banking system that are the subject of Annex IV. It addresses the timing and magnitude of the attack.
Dynamics of Speculative Attacks
When governments explicitly set a level or a range for the price of their currency relative to another single currency or weighted basket of currencies, they make a conditional commitment that the exchange rate will remain within certain specified bounds as long as the costs of defending it also remain within certain bounds. The precise limit on the costs—the use of intervention resources and economic and political costs of interest rate increases—of maintaining the exchange rate is rarely made explicit and will be uncertain in the minds of currency speculators. If the governments on both sides of the exchange rate policy put unlimited resources behind the exchange rate, no attack on a fixed exchange rate system could be successful. It is usually clear to speculators, however, that controlling the exchange rate is only one of the government’s many monetary commitments and is unlikely to have the highest priority. When the exchange rate policy becomes inconsistent with other policies, the exchange rate target may be abandoned or modified. At such a moment, the remaining limited resources available to the government for defending the exchange rate are in play as investors restructure the currency denomination of their portfolios in anticipation of the imminent policy change.
Speculators profit by buying foreign reserves from a central bank at the fixed exchange rate and selling them after a successful attack at a devalued exchange rate. Until recently, attacks on exchange rate regimes were treated as aberrant speculative disturbances of otherwise well-functioning markets. It is now recognized, however, that attacks can be the market’s response to policy goals that are perceived to be inconsistent with a given parity.
An exchange rate is pegged or controlled by the coordination of macroeconomic factors and policies and the use of accumulated or borrowed international reserves. These international reserve stocks bear the brunt of day-to-day intervention to control the exchange rate. Monetary and fiscal policy, in conjunction with countries’ growth and trade, are the “fundamentals” that set the longer-term tendencies for the exchange rate. Nevertheless, an attack on an exchange rate is first and foremost an attack on the government’s accumulated international reserve stock and its access to international reserve credit.
Speculators who attack and force the collapse of a fixed rate regime will sell the foreign exchange they purchased at the post-collapse market price. In the literature on attacks on exchange rate policies, the post-attack price is known as the shadow exchange rate. That is, it is the exchange rate that would prevail in the market if a successful attack occurred today. A successful attack results in a shift away from domestic currency toward foreign currency, thereby absorbing the government’s international reserves.
While the parity is maintained, the shadow exchange rate remains in the background as an unobservable variable—it becomes observable only the moment after the attack. A speculator will attack whenever the shadow exchange rate indicates that the currency being attacked will be depreciated by the attack. During the period when the exchange rate peg is viable, the shadow exchange rate indicates a stronger domestic currency than does the visible pegged rate. Policies that weaken the shadow value of the currency weaken the viability of the exchange rate peg and bring an attack forward in time.
These principles can be illustrated by exploring three examples. The first two are well known in the academic literature, while the third is an extension of the literature that may be particularly relevant for the exchange rate turmoil experienced during the fall of 1992 in various countries in the EMS. The first example examines how competition among speculators can determine the timing of an attack so that speculators do not profit at the government’s expense. This example also distinguishes between an attack based on market fundamentals and one based on speculative excesses. The main point of the first example is to lay out the attack as the market’s predictable response to inconsistent government policies. In the second and more realistic example, some speculators profit and some lose in the period surrounding the attack. In the third example, speculators, on average, profit at the government’s expense.
All the examples center on a hypothetical country, country H, that has fixed the price of its currency in terms of the currency of a stable foreign partner, country F. The “exchange rate” means the H-currency price of the F currency. Fixing the exchange rate links country H’s goods prices and interest rates to those prevailing in country F. Suppose that country H finances its domestic fiscal deficit at least in part by printing money. If that money financing takes place at a faster rate than the growth of domestic money demand, then country H will lose international reserves. In all the examples, there is a policy conflict: long-term money financing of the deficit in H is in conflict with long-term maintenance of a fixed exchange rate coupled with long-term price stability in F. If F were to give up its long-term price stability and inflate at the same rate as H directly or indirectly—by allowing H to increase the F money supply using F-currency reserves and loans to support the H currency—there would be no long-term policy conflict. If country F is firm in its long-term commitment to price stability, however, then F will sterilize the increase in its money supply resulting from H’s deficits. Eventually country H will run out of international reserves, or will run out of access to reserve borrowing, and will be beset by a currency crisis.
Example 1
Speculators are assumed to know exactly what the government will do. Initially, suppose the government will create domestic credit and thereby cause reserves to run out smoothly. If reserves run out with no discontinuity, no jumps in the money supply will occur. With money financing of a structural deficit and price stability abroad, country H’s currency can be expected to depreciate when the peg is removed. The depreciation can be expected to come in two parts. First, the H-currency value of the F currency will jump to equate the real supply of domestic currency assets to the decreased demand for those assets. Second, the H currency will continue to depreciate owing to excessive monetary expansion.
Now consider a single profit-maximizing speculator who can foresee the collapse of the fixed exchange rate regime. The speculator observes that reserves are running down and knows that once they run out, the currency will begin to depreciate; H-currency interest rates will accordingly jump upward, immediately reducing the demand for domestic assets and resulting in the jump depreciation of the H currency. The speculator realizes that if he could increase his holding of foreign exchange just before the collapse, he would have a capital gain, in units of domestic currency, equal to his holding of foreign exchange multiplied by the size of the depreciation of the domestic currency. The size of the depreciation at the moment of the attack is equal to the difference between the shadow exchange rate and the controlled rate. The speculator’s strategy is thus to borrow domestic currency and buy foreign exchange from the domestic monetary authority while the monetary authority still has some foreign exchange.3
Evidently, there is an inconsistency in this initial scenario. It is not possible for the monetary authority to run out of reserves smoothly, because a single speculator has an incentive to purchase all the government’s international reserves before it otherwise runs out of reserves smoothly in a final, very profitable, speculative attack. The assumption that reserves run out smoothly is therefore incorrect. The scenario allows unusual profits to exist in a risk-free environment, so competition for those profits would be fierce. If many speculators are competing for profits at the government’s expense, they would try to “leap-frog” each other in time: anticipating that another speculator might buy before him, each speculator will attempt to buy out the government’s international reserves earlier and earlier until, at some point, the exchange rate depreciation at the time of the attack would be negligible and thus so would be the profits from attacking the currency.
Recall that with expansionary monetary policy, the shadow exchange rate of the H currency was depreciating relative to the F currency. Speculators’ pursuit of profit plus the depreciating shadow value of the currency lead to the conclusion that the attack is pushed back to the moment when the controlled exchange rate is equal to the shadow exchange rate. This is how the timing of the speculative attack is determined in the literature—it happens just at the time that potential profits are driven to zero by competition among speculators for those profits. The attack still converts a large stock of domestic-denominated assets in the hands of the public into foreign-denominated assets in the hands of the public, but this is exactly what the public demands, given the now higher rate of depreciation of the attacked currency.
This attack takes place in a hypothetical environment of certainty. It is not the result of arbitrary speculative behavior but is the market’s response to incompatible government policies. The attack does not occur at an arbitrary time or at the instant the market realizes the long-term incompatibility of the government’s policies. It takes place, predictably, when the shadow exchange rate has depreciated to the level of the controlled rate. The problem that brought on the attack was that the long-term financing of the fiscal deficit by the domestic monetary authority was incompatible with the long-term maintenance of a fixed exchange rate with finite access to international reserves. Of course, if country H had access to an indefinitely large stock of international reserves, the fixed rate could be maintained.
The policy incompatibility in this example presented the private sector with an opportunity—either the policy of deficit finance of fiscal objectives or the policy of fixed rates would collapse eventually. In this case, the fiscal objectives are given priority over maintaining the fixed exchange rate. As speculators have no effective way of forcing a balanced fiscal budget, they attack the vulnerable leg of the policy stance—the fixed exchange rate. Also, if speculators wait until the government has run out of reserves smoothly, the domestic currency will suddenly depreciate when reserves are exhausted. Speculators try to seize the opportunity to purchase the government’s foreign exchange before the government otherwise runs out and to extract for themselves a risk-free capital gain. Competition among speculators hastens the attack on the government’s reserves, increases its magnitude, and decreases the size of the currency depreciation at the moment of the attack. Competition in the private sector minimizes the extent of the currency depreciation at the moment of the attack and maximizes the volume of reserves stripped from H’s central bank.
If the speculators attacked government reserves before the shadow exchange rate equaled the fixed rate, they would have lost if the exchange rate pegging authority pulled out of the market because the H currency would appreciate. If the central bank did not pull out of the market and stood willing to buy foreign exchange at the previous controlled rate, the attack would be stillborn; the speculators would sell the central bank all the foreign exchange they had previously purchased. In this example, speculators would never attack a fixed exchange rate system in which the shadow exchange rate is more appreciated than the controlled rate.4
The first example differs from those that follow because when government actions contain certainty, speculative competition drives speculative profits to zero. When uncertainty is introduced, realized profits need not be zero.
Example 2
More realistically, suppose that speculators cannot be sure of short-run government actions. They can see, however, that in the longer term the policy of fixing the exchange rate conflicts with money financing of the deficit. The speculators’ actions depend on beliefs about how long the government can maintain the incompatible policies. They will borrow domestic money to buy foreign exchange in speculation against the exchange authority. Such borrowing will drive up the domestic interest rate until it is equal to the foreign interest rate plus the risk-adjusted expected rate of change of the exchange rate.
High short-term interest rates on domestic currency act as a brake on speculative activity. For example, if it is widely thought that there is a 50 percent chance of a 2 percent depreciation over a given weekend, short-term interest rates on domestic currency might be driven to several hundred percent on an annualized basis. A speculator who bets on a devaluation over a given weekend might have to pay more than 1 percent of the value of a loan in domestic currency for the use of domestic money over the weekend. No speculator can afford to be wrong on too many such weekends.
If a crisis occurs, speculators who have recently taken short positions in domestic currency when the exchange rate peg is abandoned will profit. Those who have either dropped out of the speculation or who have been paying high interest rates on domestic currency for some time will lose. On average, competition will, once again, remove expected risk-free profits. In this case the profits are competed away in two dimensions. First, as in the certainty case, competition hastens the final attack minimizing the final currency depreciation. As before, the actual final attack takes place as soon as the shadow exchange rate equals or exceeds the fixed parity. Second, the prospect of the successful final attack drives up short-term interest rates on domestic currency, making speculation against the currency expensive.
This interest expense limits the extent of and the return to risk-averse speculation and, as in the hypothetical example, can produce some spectacular annualized short-term interest rates. As an example, during the week before the March 1983 devaluation of the French franc against the deutsche mark, short-term (three-day) Euro-franc interest rates reached over 500 percent a year. As is discussed in Annex VI, during the September 1992 crisis in the EMS, the Sveriges Riksbank raised its marginal rate to over 500 percent a year. These gigantic annualized rates are of the appropriate order of magnitude in a setting where a lender of domestic currency might be expected to take a capital loss of 2-4 percent over a weekend when the currency is devalued. Indeed, annualizing 2-4 percent over a two-day weekend results in an annualized interest rate of 250-500 percent a year.
Example 3
The third example involves an even more powerful inconsistency in government policy. Suppose that country H’s government fixes the exchange rate, finances its deficit by printing money, and pegs the short-term interest rate on domestic currency approximately equal to the foreign interest rate. Designed to fund domestic financial institutions, this is a policy combination that speculators can exploit.5 The high short-term H-currency interest rates that acted as a brake on speculators and a drain on their profits in the previous example have now been reduced. Speculators can borrow domestic money and buy foreign exchange at a small interest rate spread. These three inconsistent policies can coexist only if the domestic monetary authority has access to an unlimited foreign exchange credit line: the three policies are then not inconsistent and are invulnerable to attack. If, however, speculators suspect that access to credit is actually limited (as it must be if country F is to preserve its long-term price stability target), they will attack the weak link in the inconsistent policy chain.
This example differs from the previous one in that interest rate competition among speculators will not reduce the aggregate speculative profit. When the H government pegs the short-term H-currency interest rate, it does not allow that rate to rise to reflect lenders’ fears of being paid back in depreciated currency. Speculators profit from the interest rate peg because their speculative position is to short the weak currency and hold the strong currency. With the H-currency interest rate unable to respond to speculative forces, the speculators are able to fund their H-currency borrowing at low interest rates with the government effectively the only new short-term lender. In this situation we can expect speculators to take massive positions against a government policy constellation that makes such positions almost risk free and costless.
Multiple Equilibria
In the examples examined above, a speculative attack may be an inevitable outcome as with a country that attempts to fix its exchange rate while operating an incessant policy of rapid domestic credit creation. An alternative is also considered in the literature: a speculative attack may be only one of two potential equilibrium outcomes. For example, a country may run a tight monetary policy that is consistent with a fixed exchange rate. However, a speculative attack that breaks the peg may trigger an easy monetary policy under a less disciplined new regime. Speculators may never attack, and the fixed exchange rate can continue indefinitely. As another outcome, however, they can attack the limited reserves devoted to defending the fixed rate and trigger the shift to the easier monetary policy that validates the attack through the now devalued and depreciating exchange rate. In this case, a central bank whose fundamentals are “right” may be attacked anyway—speculators can guess that the fundamentals will not be right for the old parity after the attack, because the attack itself, together with the response of the central bank, changes the fundamentals.6
Lessons from the Literature
In the examples presented above, speculators understand that the authorities will not curb the fiscal deficit to make it consistent with the existing exchange rate. The speculators therefore turn their attention to the exchange rate policy and the possibility of profiting from a speculative attack. International reserves, whether borrowed or previously accumulated, act as a buffer that can, at least temporarily, finance the multiple targets of the excessive fiscal deficit and a controlled exchange rate. When the international reserves, or access to reserve borrowings, evaporate, the fixed rate target is abandoned.
The three examples emphasize three distinct points. First, a speculative attack need not be the result of speculative excess; it can be the market’s response to the perception of inconsistent government policies. In particular, if a fixed or controlled exchange rate is perceived to be inconsistent with other aspects of monetary and fiscal policy, the exchange rate policy can be attacked. Second, if uncontrolled during a currency crisis, short-term interest rates on domestic currency will rise to high levels. These high short-term rates act as a brake on speculative activity and may temporarily extend the life of a currency peg. Third, if, owing to the domestic financial structure or other domestic constraints such as an output recession, the attacked country’s central bank cannot allow short-term interest rates to rise to levels that would discourage speculation, the speculators’ leverage can be multiplied many fold. The speculators can borrow domestic money short term at pegged interest rates and use it to speculate against the exchange rate peg. In essence, the speculators have set one arm of government policy—the interest rate peg—at war with an other arm of government policy—the exchange rate peg.
Even if the government abandoned the interest rate peg, only the short-term aspects of the policy inconsistency would be eliminated. In the longer term, the fiscal imbalance in the country being attacked is in conflict with the price stability policy in the partner country. Unless these policies are harmonized, exchange rate stability cannot be maintained.
The literature also raises the disturbing possibility that any fixed exchange rate policy in which central banks’ access to foreign reserves is limited is vulnerable to attack if that attack would alter other policies. If it is believed that the attack will influence other policies that are fundamental to the economy, an otherwise viable fixed exchange rate can be attacked successfully.