Most empirical studies of money demand focus on holdings of domestic currency by domestic residents and ignore the effects of `currency substitution' holdings of domestic currency by foreign residents and vice versa although the theoretical arguments that this can be destabilizing are widely accepted and the money demand functions are frequently observed to be structurally unstable. In Miles's theory, which focuses on the opportunity cost of holding cash, and in the portfolio balance theory, a rise in the domestic rate of return reduces the demand for domestic money. In Girton and Roper's theory, which emphasizes interest-earning balances, the domestic rate of return has a positive effect on money holding. In all cases, the foreign and domestic rates of return operate in opposite directions, while the portfolio balance theory is distinguished from the other two in having negative effects on both domestic and foreign rates of return.

In Discussion Paper No. 507, Steve Thomas and Research Fellow Michael Wickens argue that currency substitution theories should be extended for the major trading currencies to include the related theory of a `vehicle currency'. This considers the use of a currency in trade and financial transactions between third- party countries, such as the pricing of oil and the denomination and servicing of much international debt in US dollars. Within Europe, the Deutschmark is increasingly fulfilling a similar role to that of the dollar world-wide, and there are also signs of the yen being used as a vehicle currency in the East.

Thomas and Wickens find that data for the dollar, Deutschmark and yen during 1973-88 provide weak evidence of currency substitution effects in conventional M1 money demand equations in all three countries and tend to favour Miles's version of the theory. The own-interest-rate terms are negative and significant, while the cross-rate effects are usually positive but not significant. Finding stronger effects of currency substitution requires greater precision about the nature of the currency holdings. For example, during 1977-88 non-resident demand for dollars and Deutschmarks reveals both own- and cross-rate effects, and there is also evidence that these demands were subject to vehicle currency influences. Thomas and Wickens conclude that currency substitution does not strongly influence conventional money demand and that the balances held by non-residents are subject to both currency substitution influences and vehicle currency effects.

Currency Substitution and Vehicle Currencies: Tests of Alternative Hypotheses for the Dollar, DM and Yen
S H Thomas and M R Wickens

Discussion Paper No. 507, January 1991 (IM)