Recent evidence suggests that many economic time series are subject to structural breaks. In the presence of breaks, including historical data prior to the most recent break to estimate a forecasting model will lead to prediction errors that are biased but also may have a smaller variance. This...