10 - Adjustment of Forecasts

Summary

In practice it often happens that forecasts from econometric models are manually adjusted. There can be good reasons for this. Foreseeable structural changes can be incorporated. Recent changes in data, in measurement or in the relevance of variables, can be addressed. A main issue with manual adjustment is that the end user of a forecast needs to know why someone modified a forecast and, next, how that forecast was changed. This should therefore be documented. We discuss an example to show that one may also need to know specific details of econometric models, here growth curves, to understand that even a seemingly harmless adjustment by a priori fixing the point of inflection leads to any result that you would like. In this chapter we discuss why people manually adjust forecasts. We discuss the optimal situation when it comes to adjustment and the experience with manual adjustment so far. A plea is made to consider model-based adjustment of model forecasts, thus allowing for a clear understanding of how and why adjustment was made.