Uncovered interest parity tests and exchange rate expectations

Empirical studies reject uncovered interest parity. Experimental and survey data studies reject rational expectations and find evidence of adaptive, regressive, bandwagon and distributed lag expectations. In this paper we investigate how these two findings are related. We show that uncovered interest parity test coefficients can be expressed as functions of the parameters of the expectations mechanisms. Negative values for uncovered interest parity test coefficients are explained by adaptive expectations with a high speed of learning and distributed lag expectations, while positive values are caused by adaptive expectations with a low speed of learning, regressive expectations and bandwagon expectations.