The evolution of reserve currency diversification

BIS Economic Papers  |  No 18  | 
01 December 1986


In the late 1970s, when the US dollar was under strong pressure, reserve diversification in the form of a sharp shift on the part of central banks towards holding currencies other than the dollar in their foreign exchange reserves was believed to have been a contributing factor. Some viewed this development with concern, fearing that it might destabilise financial conditions in the international monetary system. This provoked the international debate of 1979 as to whether and how SDRs might he substituted for the "dollar overhang". When the dollar gained strength in the 1980s, however. the issue of reserve diversification lost much of its earlier interest. In addition to the recovery of the dollar on the exchange markets, other developments on the international monetary scene help to explain this loss of interest: firstly. the currency composition of aggregate foreign exchange reserves seems to have been relatively stable since 1980; secondly, there is a growing perception among researchers and practitioners alike that the impact of official reserve management on exchange rates has weakened with the rapid expansion of private portfolios and exchange transactions. The dollar began to depreciate once again in early 1985, but there has been no loss of confidence in the US currency comparable with that happened in the 1970s. Accordingly, concern about a "dollar overhang" in private or official circles has so far been less evident than was previously the case.

The aim of this paper is to shed some light on the facts of reserve diversification since the early 1970s. Two points need to be made before we examine the available information on the subject. Firstly, the extent of diversification cannot be judged by looking at changes in the currency composition of global exchange reserves, since this is affected not only by changes in the currency composition of individual countries' reserves but also by other factors. For example, changes in the distribution across countries of total exchange reserves can affect the currency composition of global exchange reserves, since countries do not all hold identical baskets of reserve currencies. Thus, if the reserves of countries which tend to hold a high proportion of dollars in their portfolios increase more rapidly than those of countries which hold a smaller proportion of dollars, the share of the dollar in global exchange reserves will increase, and vice versa. Secondly, no information is available on the currency composition of most individual countries' exchange reserves, so that the discussion of diversification is limited to changes in the currency composition of the reserves of groups of countries.

Section 2 of this paper shows that, after adjusting for factors which affect the currency composition of global exchange reserves but which are unrelated to changes in countries' currency preferences, there was no large-scale diversification of reserves out of dollars into other currencies during the period under review by groups of countries. In the 1970s the adjusted proportion of dollar holdings in their foreign exchange reserves remained virtually unchanged, whereas in the early 1980s it showed a decline in some instances. In a statistical sense, therefore, reserve diversification occurred in the latter rather than in the earlier part of the period.

The absence of large-scale reserve diversification out of the dollar is perhaps surprising both in view of the increase in international trade and other transactions denominated in secondary reserve currencies, such as the Deutsche Mark and the Japanese yen, and also in the light of the potential gains to be realised from currency diversification in terms of return/risk in the portfolio management of floating currencies. This paradox is examined in Section 3, which tentatively concludes that reserve diversification in a number of countries has been limited by the extent to which reserves are held as transactions balances. These findings are reviewed in Section 4 from the perspective of recent exchange rate developments and reserve conditions.

In this paper gold, SDRs and IMF reserve positions are not taken into account. ECUs created within the EMS, on the other hand, are treated as if they were gold and dollars, i.e. dollars swapped against ECUs are included in the dollar holdings of EMS member countries and gold swapped against ECUs is excluded from foreign exchange reserves. This treatment is consistent with the legal characteristic of the ECU whereby ownership of the reserve assets swapped against ECUs remains with the individual member countries. As far as country groupings are concerned, the countries which report reserve data to the IMF are classified into industrial countries, oil-exporting developing countries and non-oil developing countries. Although this traditional distinction has recently become tenuous, it remained meaningful during most of the period under review.