Recent work by Clements and Hendry have shown why forecasting systems that are in terms of differences, dVARs, can be more accurate than economet- ric models that include levels variables, ECMs. For example, dVAR forecasts are insulated from parameter non-constancies in the long run mean of the...

After a forecast failure, a respecification is usually necessary to account for the data ex post, in which case there is a gain in knowledge as a result of the forecast failure. Using Norwegian consumption as an example, we show that the financial deregulation in the mid 1980s led to forecast...