Macroeconomic forecasting with matched principal components

摘要: This article proposes an improved method for the construction of principal components in macroeconomic forecasting. The underlying idea is to maximize amount variance original predictor variables that retained by order reduce involved estimating forecast model. achieved matching data window used constructing with estimation window. Extensive Monte Carlo simulations, using dynamic factor models, clarify relationship between reduction and various design parameters, such as observation length, number predictors, length horizon. also empirical application eight key US time series over horizons, where are constructed from a large set predictors. results show proposed modification leads, on average, more accurate forecasts than previously component regression methods.