Forecasting Financial Variables

The recent crisis has been a timely reminder of the perils of forecasting financial markets. The problem is particularly serious when markets and economies are subject to major technological, political and institutional changes.

But anticipating major breaks in economic relations is not easy, particularly if breaks occur close to the point at which forecasts are made. One strategy of dealing with such breaks is to develop forecasting strategies that are reasonably robust to breaks. Examples of such approaches include down-weighting of observations, the use of rolling windows and pooling of forecasts obtained over different estimation windows.

At CAFE we hope to build on these developments and explore the use of panels and pooling techniques to minimize the risks associated with forecasting in presence of structural breaks.