Smooth Forecasting of Evolutionary Panels

Final Report Summary - SFEP (Smooth Forecasting of Evolutionary Panels)

The contributions of this project consist in modeling, estimating and forecasting multivariate time series that are characterized by two features: non-stationarityand dimension-reduction. These two ingredients are typically observed, for example, in economic data: the variability of the series evolves smoothly over time and the series have a common behavior.
The history of economic forecasting suggests that there are some regularities informative about future events, but also major irregularities as well (see Clements and Hendry 1999). To deal with these irregularities we need to face the problem of non-stationarity that characterizes the data.
In addition to non-stationarity, these time series can be highly correlated, which motivates the use of dimension-reduction techniques. Linear factor models have attracted considerable interest over recent years especially in the econometrics literature. One of the characteristics of the traditional factor model is that the process is stationary in the time dimension (see Forni et al. 2000, Bai 2003). This appears restrictive, given the fact that over long time periods it is unlikely that the variability remains constant over time.

Final Report Summary - SFEP (Smooth Forecasting of Evolutionary Panels)

Verwandte Dokumente

Diese Seite teilen

Herunterladen