Journal Issue

The Financial Costs of Holding Reserves

International Monetary Fund. External Relations Dept.
Published Date:
January 1989
  • ShareShare
Show Summary Details

International borrowing rates affect the costs of foreign exchange reserves

Foreign exchange reserves are central not only to international economic activity but also to domestic economic policymaking. A country with a healthy reserve position is not likely to default on its bank or trade credit and threaten financial stability in other countries, while a country with low reserves has fewer funds to draw on in a crisis and is more likely to have payments problems more frequently than countries with a better cushion. From the country’s point of view, reserves provide a buffer against the possibility of future crises; they help to maintain confidence in a country’s financial management and facilitate access to trade and other credit; they enable a country to protect its current consumption from transitory or seasonal declines in export earnings, or sudden increases in the prices of imports; and they help support exchange rate policy, whether it be by smoothing exchange rate fluctuations or maintaining a currency’s value against a peg or less formal guideline.

But while it might seem that the higher reserve holdings are, the more stable the situation will be for all concerned, reserves are not without cost to the country concerned. They are financial assets—hard currencies that are often scarce and could be used for other purposes. How should one measure these costs? Do central banks in fact take them into account as they manage reserves on a day-to-day basis? And if these costs affect reserve demand, what implications do they have for international monetary stability? These questions are not new, but they have been lent a sense of urgency by the recent changes in global financial markets. Swings in access to international credit, along with the volatility of international interest rates, have put the balance of payments and reserve management of many countries under intense pressure.

Since countries do not publicize how they manage their reserves, one has to deduce the principles they follow. Traditional analysis recognized that the cost of reserves as financial assets was important in theory, yet few studies were able to find any empirical significance. But the issue turns on the definition of the appropriate alternative use of hard currencies if they are not held as reserves—that is, the proper selection of the “opportunity cost” of reserve holding. A 1985 study by Sebastian Edwards (No. 1532 in the National Bureau of Economic Research Working Paper series) did turn up such an empirical significance by testing the hypothesis that the correct opportunity cost was a country’s “net” borrowing rate (i.e., the rate at which it could borrow on international markets less the rate it earned on the assets in which it invested its reserves).

The study on which this article is based updates and extends Edwards’ analysis (see box). The results show that the rate at which individual countries borrow on international financial markets has a clear effect on reserve holdings. It also shows that the reserve holdings of countries that have debt-servicing difficulties are particularly sensitive to the cost of holding these assets. This means that economies with perhaps the greatest need for reserves economize the most when international borrowing conditions become difficult.

This article is based on a more detailed study by the author, “The Demand for International Reserves and their Opportunity Cost,” published in the September 1989 issue of IMF Staff Papers, available from IMF Publication Services, Washington, DC 20431 USA.

The period covered, 1978–86, was one that embraced dramatic changes in the international financial markets. In the mid-1970s and early 1980s, most developing countries were able to borrow unprecedented amounts from the international banking system at costs that were relatively low, though variable. After 1982, the pendulum swung in the other direction, as most of these nations found it difficult to raise loans—for some, credit ceased completely. The sample of 25 countries excluded the large “money center” nations, because their borrowing costs tend to be close to the rate they can earn on the short-term assets in which reserves are held. It also excluded the group of debt-problem countries after 1982, because these countries then faced credit rationing, meaning that their borrowing rates no longer determined either the amount of credit they received or reflected prevailing credit conditions.

Framework of study

The concept of the study is that the level of reserve holdings should be conditioned by their costs as financial assets, as well as the need to cover prospective payments deficits. There seems to be no generally applicable rule of thumb that defines reserve adequacy; reserves that are adequate for one country might well be inadequate for another, or even for the same country at a different time. In general, countries that are more open to international trade tend to hold higher reserves than countries that are less so. Beyond this, countries with limited capacities for earning or borrowing foreign exchange, or with highly concentrated or variable sources of earning, should need higher reserves than those with the same foreign exchange needs but sounder earning capacities and better credit standings.

A common way to judge reserve adequacy is to link reserve levels to a country’s foreign exchange commitments reflected in its imports. But reserve-to-import ratios can only be a useful guide where other factors, particularly revenue potential, stay constant, and this is a risky assumption—in fact, countries with similar import levels generally do hold different reserve levels. Most of the more detailed empirical work on the demand for reserves has, therefore, associated reserve holdings with a measure of the country’s openness (to capture external revenues and commitments), and with variations in past external payments deficits (to capture the future need for reserves). The results show that both factors are important in determining reserve holdings.

But this approach ignores that reserves are also financial assets and entail a cost. Foreign exchange used to increase reserves could have been put to some other use. Given that reserves tend to be held in short-term secure assets, which pay lower interest rates than longer-term or less secure assets, this cost can be calculated as the difference between these lower earnings and returns on the highest-yielding alternative.

The crucial issue thus becomes how to identify that alternative. In principle, countries will borrow abroad until returns on domestic investment drop to match the cost of borrowing, implying that all rates eventually equalize. But recent analysis suggests that because of the various risks associated with lending, particularly to developing countries, domestic investors will cease borrowing well before rates meet. As a result, measured domestic returns may remain permanently above international borrowing rates, the difference being explained by existing risks. Even so, researchers using domestic interest rates or some proxy for returns on domestic assets generally have found no impact on reserves; governments apparently do not view domestic investment as the best alternative to holding reserves.

The argument here is that governments look on repaying external debt as the appropriate alternative to—that is, the correct “opportunity cost” of—holding foreign exchange in reserves. This is so because countries place a high priority on making timely debt payments; by maintaining creditworthiness, they have better prospects of borrowing what they need on favorable terms. To test the idea, this study analyzed the impact on the sample countries of this opportunity cost—defined as the unit cost of a country’s foreign borrowing, less the return it earned on investing the reserve in a short-term liquid asset (the three-month US Treasury bill in this case).

Costs can be high

Just a cursory review of the relevant data shows that when the net cost of borrowing is used to calculate the costs of reserve holdings, these costs can be relatively high, in terms of an important source of foreign exchange earning, such as exports (see chart). Calculating the total cost of reserves as the net cost (as defined in the note to the chart) times the level of reserves, Colombia’s reserves cost the country the equivalent of 4 percent of export earnings in both 1981 and 1982 (compared with often considerably less than 1 percent in Australia), at a time when both reserves and exports were rising at about the same rate. India’s reserves, in the late 1970s, absorbed from 2–3 percent of exports. Net reserve costs were also high in the Philippines in 1982. Although the data show that these costs absorbed only half the share of exports that they did in 1981, total reserve holdings in 1982 were almost 250 percent less than they were at the end of 1981.

The concept underlying the empirical test of reserve demand is the same as that for money demand, according to which people hold cash as a precaution against income shortfalls, people tend to demand more cash as their wealth increases, and people hold less cash when returns on other assets—such as savings rates—rise. In formal terms, the study assumed that countries minimize the total costs of reserves held as a precaution against deficits by minimizing opportunity costs arising out of two situations: reserves held as a precaution against deflation of the domestic economy, and reserves held as an asset. For each country in the sample, it was predicted that optimum reserve holdings would rise with a measure of “scale” that reflected its economic size (imports in this case), with the likelihood that deficits would occur, and with the expected cost of deflation (captured by past deficits). To capture the appropriate opportunity cost of reserves as an asset, it was assumed that reserves would fall as the rate increased on its external debt less the return on a secure liquid asset.

… and are taken into account

When this test was run on the whole group of countries for 1978–82, as well as for all countries excluding the problem debtors from 1978–86, all the variables were shown to be important determinants of reserves and to have the expected effect. Higher potential deficits caused the authorities to increase reserves; greater openness led them to reduce reserves (a high ratio of imports to GDP was associated with low reserves); and higher costs of borrowing relative to investments in short-term assets also caused the authorities to reduce reserves. Holdings also rose with economic size, as reflected in higher imports.

In an effort to determine whether these results also held true for different groups of countries when tested separately, the test was rerun for 1978-82 on the industrial countries and two groups of developing countries—those that were and those that were not to develop problems in managing their debt after 1982. The one striking difference with the test for the whole sample in the 1978–82 period was that the net cost variable proved important only for the countries that were to have debt problems.

If, as this study suggests, the debt- problem countries gave more weight to borrowing costs than others in managing their reserves, there may be cause for concern. Monetary authorities manage their reserves both to provide a buffer against future crises and to maintain confidence in the country’s financial management. Even as early as 1978, observers were aware of the size of debt being accumulated by some countries. The first multilateral reschedulings were arranged in 1975; there were two in 1976, three in 1977, four in 1979 and 1980, and many more thereafter. Before 1982, risk considerations were captured by interest rates, which were also generally higher for the large debtor countries (although there was quite a degree of variation). At the same time, their current account deficits were high, owned reserves were likely to be low, and one might reasonably suppose that their reserve holdings were partly borrowed. Faced with higher borrowing costs, these countries seem to have been concerned to continue scheduled repayments and to maintain their reputation in international financial markets for as long as they could, even if they had to reduce reserves to do so. There may also have been solvency problems; the authorities in the debt countries simply could not afford to hold expensive reserves.

Cost of reserve holdings for selected countries

(Millions of US dollars, percent)

Sources: Bank of England (or syndicated loan costs; Data Resources Inc. Pot the Treasury bill rate; and IMF, International Financial Statistics for reserve and export data.

Notes: Net cost data are calculated as the spread over 6-month LIBOR, plus the short-term LIBOR rate, less the short-term US Treasury bill rate on an annual basis. Total costs are calculated as reserve holdings times net cost, in millions of dollars. The percent of exports figures are calculated as the total cost of reserves as a percent of exports of goods, in dollars.

The apparent lack of importance of interest rates in the non-debt-problem countries suggests that higher borrowing costs were not associated with lower reserves. In these countries, current foreign exchange earnings might have been sufficient both to meet higher interest payments and to increase reserve holdings.


There are, of course, many problems with drawing either firm or general conclusions from a narrow empirical test such as this. First, its scope is partial. Only reserve demand has been estimated, but both reserves and the factors determining reserves are affected by other economic forces, both policy determined and exogenous, that could at best bias and at worst completely distort the results. Second, the determining factors themselves beg a number of questions. Some, such as the use of imports as a scale variable rather than GDP, could be answered by a different selection of variables; some could be answered by a different methodology (this would cover whether banks set interest rates by looking at countries’ reserves rather than vice versa); while other questions—for example, whether past variability in deficits is always a good proxy for future deficits—are simply unanswerable.

Nevertheless, given the little we know about what determines actual reserve levels, the finding that holdings are sensitive to borrowing costs increases our understanding of reserve management. Given the importance of reserve holdings for international financial stability, the results also underscore the risks inherent in a system that causes countries with debt problems to economize on reserves when borrowing costs rise. This goes back to the earlier notion that if countries reduce reserves below levels considered “adequate”—however adequacy is defined—financial stability may be at risk.

Other Resources Citing This Publication