The increasing sophistication of foreign exchange markets offers investors a wide range of liquid assets. In normal times, this enables participants to manage their risks and lower their costs of transacting, and is therefore conducive to a more efficient allocation of funds in the world economy. But these same features have also expanded the range of opportunities available to speculators who believe that a currency’s sudden depreciation is impending, and have increased the private sector’s ability to marshall large resources on very short notice. Of course, during a foreign exchange crisis, when most market participants become convinced—rightly or wrongly—that a large change in the exchange rate could well be in the offing, the line between speculation and hedging becomes rather blurred.
This annex discusses the mechanics of speculative attacks, including the alternatives open to speculators and hedgers in their position-taking activities against a currency that is expected to depreciate, as well as the costs involved in such activities. It shows how these transactions feed their way through the banking system, and how they are financed.1
Bank Covering Operations for Forward Contract Positions
Speculators typically attack a currency through short sales, generally by entering forward contracts to sell the currency. Forward sales may also be associated with hedging programs implemented by fund managers, nonfinancial corporations, and market makers. These hedging programs either entail a 100 percent hedged position or the use of OTC or synthetic currency puts with a lower hedge ratio. These operations require the sale of the weak currency forward to a bank—either once and for all in a 100 percent hedge, or continuously to rebalance positions as interest differentials and volatility shift.
The international banking system handles a forward sale of a currency in the same way, regardless of whether the customer is hedging a security denominated in a weak currency or speculating against that currency in a short sale. Speculative short sales involve entering relatively long-dated (at least one month) forward contracts with banks or security houses, for example, by selling sterling for deutsche mark.
As standard practice to balance the long sterling position that this transaction initiates, the counterparty bank will immediately sell sterling spot for deutsche mark for two-day settlement. Although its currency position is then balanced, the bank still has a maturity mismatch in both deutsche mark and sterling: it will borrow sterling overnight to cover settlement of the spot sale, but it will receive sterling 30 days hence. It faces the opposite maturity mismatch with its deutsche mark position: it buys spot deutsche mark but will need it only 30 days hence so it will lend overnight deutsche mark and effectively has borrowed 30-day deutsche mark. To close this maturity mismatch, a bank typically will transact a currency swap, which entails a delivery of deutsche mark for sterling spot and a delivery of sterling for deutsche mark forward.2 These are customary wholesale operations executed by banks writing forward contracts to speculating customers—individuals or institutions. Standard banking practice thus implies a cautious response to the currency market: the bank takes on neither currency nor interest rate risk, only counterparty risk.3
Figure 1 presents a concrete example of such a forward transaction. In this example, both the forward and spot exchange rates between the deutsche mark and sterling are DM 3 per £1. In the first step, a customer sells £100 forward for DM 300 to a bank. This is an off-balance-sheet item for the bank, but it has payments implications like any on-balance-sheet transaction. The payment and receipt implications for the bank are displayed in the first panel. The bank will receive £100 and pay DM 300 in one month. These are the same movements of funds that the bank would face if it held a sterling treasury bill and shorted a deutsche mark treasury bill. To eliminate the currency mismatch, the bank immediately sells £100 for deutsche mark spot exchange, the payment implications of which are combined with those of the forward contract in the second panel. The currency positions are now balanced, but a maturity mismatch in each currency remains—long-maturity sterling is funded with rollover sterling and rollover deutsche mark are funded with long-maturity deutsche mark. Finally, to eliminate the maturity mismatch, the bank immediately enters into a one-month currency swap, exchanging DM 300 for £100 spot and £100 for DM 300, 30 days forward. The complete payments implications for the bank are displayed in the third panel: the bank has eliminated market exchange and interest rate risk through these transactions. But it has increased its counterparty risk—both with the customer in the forward contract and with the counterparty bank in the swap and spot sales.
These transactions demonstrate that, first, a sterling-deutsche mark forward contract is equivalent to a currency swap and spot exchange transaction. Second, on its origination, a forward sale of sterling by the customer immediately generates a spot sale of sterling by the bank. Third, since by entering into the forward contract the bank is exposing itself to counterparty risk, it is effectively extending credit. It limits this credit risk by placing credit line limits on its counterparties. These credit limits restrict the off-balance-sheet credit expansion that would be associated with a large volume of forward transactions, and thereby limit the volume of forward contracts that can be issued.
Finally, the customer in the forward contract may be a central bank, which can intervene in the foreign exchange market by buying its currency forward rather than spot. If the central bank’s forward purchase matches a forward sale of some other customer in the banking system, all the swap and spot transactions of the banking system will cancel; specifically, spot exchange sales will be matched with purchases at the parity exchange rate. Thus, the central bank’s intervention will absorb the spot sales of its currency without the central bank’s having to intervene directly in the spot market. Nevertheless, the forward contract must be financed by a bank, which takes counterparty risk with the central bank. Thus, a central bank that enters into a forward contract is drawing down credit from the banking system. Domestic banks may not directly limit the central bank’s credit, as it is sovereign, but domestic banks will in turn be limited in their access to foreign exchange by the prudential management of foreign banks. An equivalent on-balance-sheet impact would occur if the central bank borrowed 30-day deutsche mark from the bank, sold it for spot sterling, and placed the proceeds in a 30-day sterling time deposit with the bank.
The transactions that the bank undertakes to balance a forward sale of sterling ultimately require a counterparty from outside the commercial banking system willing to buy sterling forward. In the ERM crisis, few private parties were willing to take this position. To fuel a speculative attack, the banking system must provide credit in the attacked currency. This is evident in the first panel of Figure 1, where the bank’s sterling receipts from the forward contract constitute a one-month sterling loan to the short seller. In the ERM crisis, the lira, sterling, or franc credit provided by the banking system ultimately was a pass-through of credit from the central banks. The central banks were the counterparties in both legs of the position-balancing transactions of the banking system. For example, by entering into a one-month sterling currency swap, a bank effectively borrowed one-month sterling. When this sterling was sold on the spot market, it was bought by the central banks in their efforts to support the currencies under attack. The lenders of, say, one-month sterling through the swap acquired the sterling that they needed to deliver spot by discounting paper through the discount houses at the Bank of England. This in turn sterilized the effect of the central bank’s foreign exchange intervention. In sum, the central banks financed the attacks by providing funds at a ceiling interest rate, as the demand for credit in the currencies under attack increased.
Alternative Position-Taking Strategies and Their Costs
As an alternative to using a forward contract, a speculator can use on-balance-sheet methods to short a currency. For example, in an attack on the Swedish krona, a speculator may borrow from a bank, sell the kronor spot for deutsche mark, and hold deutsche mark bank deposits. If on-balance-sheet funds are available to the speculator at market interest rates, the speculator will profit if the krona depreciates by more than the interest rate differential between the krona and the deutsche mark; that is, if the actual depreciation of the exchange rate exceeds the discount in the forward market for kronor.
Thus far, the effect of bid-ask spread movements as cost factors in calculating the profitability of speculation has been ignored. During the ERM crisis, bid-ask spread movements in key money and exchange markets were large, reflecting illiquidities in the banking systems. As off-balance-sheet operations became illiquid and often unavailable, speculators had to switch to the on-balance-sheet operation of borrowing and selling spot. In normal times, it is cheaper to use off-balance-sheet methods because they use less of the banking system’s expensive reserves. In a crisis, however, a flight to demand deposits usually occurs (despite their low yields), as market participants seek to gain flexibility, and as market makers in securities find that they must fund themselves through the banking system. The cumulating demands for on-balance-sheet bank products increase bid-ask spreads in interbank markets, and these increased spreads are then reflected in higher bid-ask spreads throughout the financial markets.
Speculators who buy and sell securities, or borrow and lend bank funds, to take positions in foreign exchange must consider the bid-ask spreads in these markets as part of the cost of speculation, particularly during a currency crisis when these spreads are substantial. Whether a speculator can lower the costs associated with this higher spread through an appropriate choice of instruments is considered below.
Speculation Using Bank Loans
A speculator against, say, the Swedish krona can take a short position by borrowing kronor from a bank and purchasing a bank deposit denominated in deutsche mark. For example, on September 17, 1992 during the ERM crisis, a closing-one-month interbank-ask rate quoted by banks selling Swedish kronor (lending funds) in the retail market was 70 percent per annum, while the price at which they would buy (bid for) funds was 50 percent.4 The one-month interbank rate (ask) quoted for banks selling deutsche mark in the retail market was 9.25 percent, and the price at which they would buy funds was 9.15 percent. As large banks acting as market makers in Swedish kronor and deutsche mark deal among themselves at much narrower spreads than the bid-ask spreads posted by these banks to speculators, the spread in the wholesale (or interdealer) market is much lower than the spread in the retail market.
A speculator in the retail market that borrowed Swedish kronor and acquired a deutsche mark interbank deposit paid 70 percent on the Swedish krona loan and received 9.15 percent on the deutsche mark deposit, therefore facing an interest rate spread of 60.85 percent. Before the speculator could make a profit, the krona would have had to depreciate by 60.85 percent on an annualized basis. Over a one-month period, the speculator’s break-even rate of depreciation would have been 5.07 percent.5
Selling Assets
Holders of weak-currency-denominated securities will also be among currency sellers trying to liquidate their positions. For example, a U.S. pension fund that holds krona-denominated securities may wish to close this position by selling the securities and purchasing deutsche-mark-denominated assets. In this context, interest rates of securities denominated in kronor will usually be greater than those of securities denominated in deutsche mark owing to market expectations of depreciation of the krona. Although the pension fund will sacrifice some nominal yield by switching to deutsche-mark-denominated securities, it expects to make up more than this difference from the appreciation of the deutsche mark relative to the krona.
An additional cost that the pension fund must absorb is the bid-ask spread in transacting securities denominated in kronor.6 The spread must be subtracted from its speculative gains, as in the bank loan above; thus the higher the spread, the more costly is a speculative bet against a currency.
If a dealer expects that he can easily find a buyer for the securities sold to him by the pension fund, the bid-ask spread that the pension fund must pay will be relatively small. If the dealer cannot find a buyer readily, he must finance the securities as part of his inventory. If he must finance inventory for a long period with a bank loan, the bid-ask spread will be very high. When the dealer uses a bank loan, potential investors are signaling that they do not want to hold the security sold to the dealer by the pension fund; instead they usually want to hold a demand deposit, which is a considerably more liquid instrument. In turn, the bank must structure its balance sheet to transform the loan it makes to the dealer into a liquid deposit. This is expensive, and the dealer must pay for this expense in his borrowing costs. These are the costs ultimately passed on to the pension fund.
If the dealer must fund his portfolio for one month at the ask price of 70 percent, he must earn an annualized return of 70 percent on capital gains and interest on these securities when he finally sells them if he is to break even. The pension fund will be forced to accept a discount of its securities sufficient for the dealer to earn his required return.7
Finally, it is sometimes claimed that the ERM crisis was different from earlier speculative attacks because a large proportion of sellers were liquidating long positions in securities denominated in the currency under attack. According to this argument, the seller of an asset denominated in the currency under attack is not deterred by a central bank defensive squeeze aimed at raising the short-term interest rate against short-term borrowers who are making speculative short-sales. Nevertheless, it is the funding requirements of a buyer of the dumped securities that are squeezed by the central bank action. The buyer will pass through any anticipated costs of a liquidity squeeze to the seller by reducing the bid price for the security. If the market for the securities is highly liquid, the dealer will have to fund only briefly with a bank loan, and the costs of the squeeze for the seller of securities will be low compared with a short seller who is funded for a long period by rollover credit. If the market for the securities is illiquid, the seller will pay up front the same cost of the squeeze as would the simple short seller of currency, and will not be less deterred.