Inflation forecasting and the distribution of price changes

Abstract

This study shows that replacing the traditional measure of asymmetry that is skewness in the inflation forecasting model with an alternative asymmetry measure that captures the joint influence of both skewness and variance on inflation significantly improves the forecast at various horizons. The empirical evidence suggests that it is more appropriate to use such measure of asymmetry in inflation forecast model as it has edge over simple measure of skewness in predicting inflation. These findings are consistent with the prediction of menu cost model that the variance of cross sectional distribution of relative price changes amplifies the impact of skewness on inflation.