Revisiting the Fisher Hypothesis for Several Selected Developing Economies: a Quantile Cointegration Approach

Abstract

This paper reinvestigates the validity of the Fisher hypothesis, for several selected developing countries. With the quantile cointegration method proposed by Xiao (2009), we find that the long-run coefficients between nominal interest rates and inflation can be affected by the shocks and, therefore, may vary over time. More specifically, in the upper quantiles there is one-to-one relationship between the two variables, supporting the Fisher effect, while in the lower quantiles, the nominal interest rate responds by a lower percentage than the change in inflation. This is known as the Fisher effect puzzle. Thus the Engle-Granger cointegration regression may suffer from model misspecification, because of the assumption of a constant cointegrating vector. A possible explanation for such an asymmetric relationship between the two variables is provided.