Causality testing and sensitivity to detrending: The money-income relationship revisited

Abstract

The money-income causal relationship is examined using the Hsiao (1979) causal testing method for alternative definitions of money and different detrending transformations. The results for Barbadian data suggest that each detrending transformation will induce different lag specifications in the causal regressions and that the implied causal directions are sensitive to both the definition of money and the detrending technique used. An unambiguous result, based on an examination of the components of money, is that quasi-money Granger-causes income. A recommendation of the paper is that tests for unit roots be used prior to performing the causality tests.