A reexamination of the uncovered interest rate parity hypothesis

Abstract

It is now widely known that the Uncovered Interest Rate Parity (UIP) hypothesis, which states that forward exchange rates are unbiased predictors of future spot exchange rates, is generally rejected by empirical evidence. We show in this paper by means of a Cox et al. (1985) model extended to a two-country economy, that a careful examination of the no-arbitrage principle leads to a different formulation for this hypothesis. Roughly speaking, this new formulation states that the expected future spot exchange rate is equal to the forward exchange rate multiplied by a term premium, whose closed-form expression can be derived. Moreover, this multiplicative term premium reduces to 1 (i.e. the standard UIP hypothesis holds) if and only if conditional variances are deterministic. In short, we show that rejection of the standard UIP hypothesis is not surprising as long as ARCH effects have been evidenced in exchange rate data. As a by-product, we provide a parametric and tractable specification for the so-called risk premium (i.e. the component of the expected exchange rate in excess of the forward exchange rate).