Unpredictability and the Foundations of Economic Forecasting

We revisit the concept of unpredictability to explore its implications for forecasting strategies in a non-stationary world subject to structural breaks, where model and mechanism differ. Six aspects of the role of unpredictability are distinguished, compounding the four additional mistakes most likely in estimated forecasting models. Structural breaks, rather than limited information, are the key problem, exacerbated by conflicting requirements on `forecast-error corrections'. We consider model transformations and corrections to reduce forecast-error biases, as usual at some cost in increased forecast-error variances. The analysis is illustrated by an empirical application to M1 in the UK