Money,inflation and untested common factors

Bairam's (1990) model of world inflation is reconsidered. This model gives the surprising result that increases in the money supply act to decrease inflation rates. I find that this result is due to the imposition of an untested (and invalid) common factor restriction. This restriction is implicit in the generalized least squares (GLS) transformation for autocorrelated regression residuals. Bairam's autocorrelated residuals appear to be caused by mis-specified dynamics. In a more correctly specified model money supply increases have the expected positive influence on inflation rates. The prevalence of the GLS transformation, and the rarity of testing for implied common factor restrictions, suggests that there may be other, equally mis-specified, models existing in the literature.