Exchange risk, exchange regimes and the portfolio balance model of the exchange-risk premium

Those portfolio-balance models of the exchange-risk premium that have been empirically tested are special cases. Two of the special-case models are the Solnik and inverse bond-demand models. There is some empirical support for the Solnik model but virtually no empirical verification of the inverse bond-demand model. In this study a general bi-lateral model of the exchange-risk premium is developed. It is shown that the inverse bond-demand model and a bi-lateral, Solnik model are special cases of the general model. The general model also includes varying inflation surprises, a feature missing from both the inverse bond-demand and Solnik models.The general model is estimated for the case of the United States and Canada. The empirical results strongly suggest that the Solnik model adequately explains the exchange-risk premium for both the adjustable-peg and floating-rate periods. None of the variables suggested by the inverse bond-demand model are significant when estimated as part of the general model.In the Solnik model, confirmed by testing of the general model, the exchange-risk premium is a function of the product of net foreign investment and the variance of the exchange rate. Net foreign investment determines the sign of the exchange-risk premium. The absolute size of the exchange-risk premium is determined by the absolute size of the product of net foreign investment and the variance of the spot exchange rate. Control of the exchange-risk premium can be exercised simply by controlling the volatility of the exchange rate, although this may produce other costs. Although net foreign investment also determines the exchange-risk premium, this is less subject to direct control.