MISSPECIFICATION BIAS IN MODELS OF THE EFFECT OF INFLATION UNCERTAINTY

In this paper we show that the appropriate measure of inflation uncertainty is relative measure. Specifically, is the standard deviation of inflation relative to the gross expected rate of inflation. Empirical studies of the effects of inflation uncertainty have misspecified their models by not using the relative measure. The bias introduced is equivalent to omitting a relevant variable. We derive the asymptotic bias introduced by this error, and estimate its magnitude for the United States and eight high-inflation countries. Estimates of the effect of inflation uncertainty from high-inflation countries, and cross country studies that include them are likely to suffer from a considerable negative bias. We conclude that future studies should employ this relative measure.