The carry-over effect and its value in forecasting annual growth rates

Summary

The statistical overhang (carry-over effect) is the contribution of the previous year to growth in the current year. Practitioners use the statistical overhang routinely to pin down and rationalize short-term forecasts. This article analyses the statistical overhang ‘statistically’ and quantifies its effect on forecast uncertainty. For quarterly data, knowing the statistical overhang at the end of the previous year reduces forecast uncertainty for annual growth in the current year to 68 percent. With monthly data, relative forecast uncertainty reduces to 56 percent. Against the background of the recent financial and economic crisis the analytical results are confronted with empirical forecasts. For ex post mean value forecasts the reduction of forecast uncertainty is broadly in line with the theoretical profile. With regard to the ex ante Bundesbank and Consensus forecasts, the decline of forecast uncertainty is somewhat slower than expected on the basis of information about the carry-over effect.