Nowcasting Japanese GDP using targeted predictors

Since the GDP is unobservable over monthly periods and its sources of variation are interrelated, estimating the GDP in real time is a challenge in macroeconomics. In this paper, we estimate the real-time Japanese GDP using a dynamic factor model to address the above issues. We extract the dynamic factors of an economy following Giannone et al.(2008). In addition to the baseline model, we introduce a factor-augmented diffusion index equation targeted to the GDP for better forecasting accuracy. The variable selection scheme for this model extension follows Bai and Ng (2008). The model performed better empirically than the others in terms of forecasting accuracy.