RDP 9406: Reserve Bank Operations in the Foreign Exchange Market: Effectiveness and Profitability 2. Previous Profitability Studies

2.1 The Connection Between Profits and Stabilising Intervention

In 1953 Friedman noted that “speculation can be destabilising in general only if speculators on the average sell when the currency is low in price and buy when it is high”.^[3] Conversely, profitable speculation would be stabilising because, to make profits, speculators would need to buy low and sell high, thereby reducing variability in the exchange rate. As a central bank intervening in the market acts (in some respects) much like an ordinary speculator, Friedman suggested that its success or failure in stabilising the rate be judged on the basis of an ordinary speculator – “there should be a simple criterion of success – whether the agency makes or loses money”.^[4]

This simple criterion elicited many articles examining the link between profitability and the effectiveness of foreign exchange intervention. The research focussed on:

1. the extent to which central bank foreign exchange operations had been profitable; and
2. the possibility that profitable intervention may not be stabilising or that stabilising intervention may not be profitable.

2.2 The Profitability of Intervention

There have been several empirical studies on the profitability of intervention for major countries. One of the first was that published by Taylor (1982), which examined nine industrial countries early in the floating period, from the early 1970s to the end of 1979.^[5] According to his estimates, central banks lost more than $US11 billion over the whole period. Losses varied substantially for individual countries and, more importantly, as the period over which the calculation was carried out was altered.

Subsequently, several authors challenged Taylor's results, reworking his calculations using several refinements. By lengthening the sample period and taking account of the interest differential between investing in foreign currencies and the local currency, Argy (1982), Jacobson (1983), and the Bank of England (1983) found that these large losses were in fact profits. For the US, for instance, Jacobson estimated that losses totalled around $US500 million for the 1973–79 period, but over the entire 1973–1981 period, net profits amounted to almost $US300 million. Moreover, including a measure of net interest earnings increased profits by up to $US470 million over the longer period.

Many of these studies suffered from the fact that they were based on data which only approximated the exchange rate at which transactions occurred. For example, they used published data which were often on an end-of-month basis. Given significant intra-month movements in exchange rates, this is not a good basis for reliable studies, as the rates at which a central bank deals in the market can be very different from the published series.

A later paper by Leahy (1989) on the profitability of US intervention sought to reduce some of the aggregation problems by using daily rather than monthly observations. Over the entire study period, from 1973 to early 1988, profits amounted to almost $US5.5 billion when interest income was included and $US3.8 billion when excluded. Again, however, Leahy noted the sensitivity of his calculations to using shorter periods, pointing to sub-periods when the authorities incurred large losses on intervention operations.

Murray, Zelmer and Williamson (1990) from the Bank of Canada applied Leahy's basic analysis to the Canadian experience. Unlike many earlier studies, however, they used actual data on the amount and the rate at which daily interventions were undertaken. Over the full sample, from mid 1975 to mid 1988, profits from intervention amounted to about $C1.6 billion ($US1.2 billion), with most of that coming from net interest earnings. Like other studies, the calculation of profits varied among sub-periods. Although profits were relatively large over the period as a whole, substantial trading losses were realised during some sub-periods.

2.3 The Relationship Between Profitability and Effectiveness

In making his claim that stabilising intervention would be profitable, Friedman might have had in mind a situation such as that depicted in Figure 1. The exchange rate is expressed here as the price of the foreign currency – i.e. as the amount of domestic currency required to buy one unit of foreign currency. This is the way most countries quote their exchange rate. A low reading indicates that foreign currency is cheap (the value of the domestic currency is high) and a high reading indicates that foreign currency is dear (the value of domestic currency is low). For those readers who are used to working mostly with Australian dollar exchange rates, this definition of the exchange rate may, unfortunately, be confusing as the Australian dollar is quoted in foreign exchange markets in reciprocal form (e.g. US70 cents per Australian dollar). The more widely accepted definition is used in this paper to be consistent with international usage and to simplify the understanding of the algebra used later in the study.

In terms of Figure 1, a central bank which bought foreign currency at point A and sold at point B would make a profit, since it bought when the price was low and sold when the price was high. It is also likely that by buying foreign currency at point A, the central bank would have tempered the domestic currency's appreciation (so that the exchange rate followed the path traced by the solid line rather than the dotted line) and, by selling it at point B would have reduced the local currency's depreciation.

In this example, there is a very clear relationship between profitable and stabilising intervention. However, many researchers did not see this as proving that there was a link between profitable and stabilising intervention. They went on to examine cases where there was no relationship between the two.

Some argued that a central bank's operations might be profitable, but have little impact on the exchange rate and therefore not be stabilising. Referring again to Figure 1, this would be the case if the central bank bought at A and sold at B, but the exchange rate followed the dotted line. This is theoretically possible, but unlikely. While a central bank might be able to conduct its operations in such a low key way that it would have no impact on the exchange rate, it is hard to think of reasons why it would do this on a continuing basis, unless its operations were always aimed at something other than stabilising the exchange rate.

More recent researchers^[6] have pointed to the possibility that profitable speculation could lead to more, rather than less, exchange rate variability. In these models, a group of speculators is aware that some market participants tend to buy as the price rises and sell as it falls. Initially, these “knowledgeable” speculators push the price away from equilibrium, knowing that the price will move even further away as the latter group enters the market. At this point, they turn their positions and the price moves back towards its equilibrium. While such a model may reflect the behaviour of some private speculators, who are trying to induce instability into the rate, it makes little sense for central banks which are trying to stabilise the rate. Therefore, while the situations outlined by these researchers are theoretical possibilities for private speculators, for all practical purposes they can be put aside.

Others^[7] argued that a central bank's operations may be stabilising but not be profitable. Again, it is not hard to find illustrations of this. They fall into two broad categories. One involves measuring profitability over very short periods. For example, if a central bank had bought foreign currency but the price of foreign currency fell further soon after, then over that time span, those operations would not have been profitable. Central banks, however, would not normally judge the success of their operations over a short time frame. They would be looking to see whether their activities had a net stabilising effect over a period as long as an economic cycle; for example, did it contribute to reducing the degree of overshooting at the trough and at the peak. The effectiveness of intervention, and its profitability, therefore need to be measured over a time frame consistent with the central bank's objectives.

Second, central banks are not profit maximisers; they do not wait until the exchange rate reaches the bottom before buying (even if they knew where the bottom was). A central bank that was aiming to stabilise its exchange rate would start to buy foreign currency a long way before the exchange rate reached its probable low point. Unless the domestic currency immediately stops appreciating when the central bank starts buying foreign exchange, something which is unlikely, the central bank's operations are likely to show a loss for some time. In terms of Figure 1, a central bank might start to buy foreign currency at point C. By the time the foreign currency had depreciated against the domestic currency at point A, those operations would be showing a loss. It is not until the price of foreign exchange started to move back up that the operations would show a profit. In short, if a central bank has a longer term objective in mind, the success of those operations can only be assessed over the span of a complete cycle. At various points along the cycle, operations are likely to show a loss, even though they may have been stabilising.

Another situation in which stabilising intervention may not be profitable is if the exchange rate is on a consistent upward or downward trend. Drawing on Mayer and Taguchi, this is illustrated in Figure 2.^[8] The exchange rate fluctuates around a downward trend. A central bank which bought foreign currency at point A and sold it at point B may be stabilising the exchange rate (in the sense of reducing the variation around the trend) but those operations would not be profitable as the rate at which the currency was bought would be higher than the rate at which it was sold.

Various researchers, however, have pointed out that in the circumstances outlined above, if profitability calculations take into account not only the trading profits on foreign exchange operations but also the differences in net interest earnings as a result of the switch from foreign to domestic assets (or vice versa), then the relationship between stabilising foreign exchange operations and profitable operations could be restored.

Taking account of net interest earnings will in most cases add to the profitability of central bank intervention as long as there is some tendency towards uncovered interest parity, i.e. as long as interest rates on the depreciating currency tend to be higher – on average – than on the appreciating currency. For example, if the central bank whose currency is depreciating sells foreign currency to buy its own currency, it would increase interest earnings on its assets as it moved from foreign to domestic assets. Similarly, for a country whose currency is appreciating, central bank purchases of foreign currency would result in trading losses in the short term but higher interest earnings. Instances where intervention has a negative effect on net interest earnings are confined mainly to periods when the domestic currency is appreciating due to tight monetary policy and the central bank is intervening to slow the appreciation.

In summary, we would conclude from the results of the available research that, while there can be instances where stabilising intervention is not profitable and profitable intervention is not stabilising, it is likely that they would be exceptions rather than the rule. Such cases tend to occur when measurement is over a period that is relatively short or when the exchange rate has a strong trend over the long term. In the latter case, allowing for net interest earnings can help. In general, however, particularly when measured over reasonably long periods (covering at least one complete economic cycle) evidence that foreign exchange operations are profitable would suggest that those operations are also stabilising. On this analysis, the profit criterion as set out by Friedman remains a reasonable test of the effectiveness of central bank foreign exchange operations in stabilising the exchange rate.

Footnotes

Friedman (1953), p.175. [3]

ibid, p.188. [4]

The United States, Japan, Germany, United Kingdom, France, Italy, Canada, Switzerland and Spain were included in his study. [5]

See Shleifer and Summers (1990). [6]

See Bank of England (1983). [7]

See Mayer and Taguchi (1983), Diagram 3 on page 15. [8]