An empirical examination of the generalized Fisher effect using cross-sectional correlation robust tests for panel cointegration

Abstract

This study examines the generalized Fisher hypothesis as applied to common stocks by using the recently proposed second generation panel cointegration tests. Unlike their predecessors, these new tests assume the existence of cross-section dependence in the data. For the sample analyzed, we report that these new tests, but not their predecessors, provide strong support for the existence of cointegration between stock and goods prices. Moreover, further analysis cannot reject the hypothesis that the cointegration relation is linear. Finally, our Fisher coefficient estimates are in the range between 0.68 and 1.27 and give support to the generalized Fisher hypothesis.