Nowcasting Japan's GDP
62 Pages Posted:
Date Written: May 15, 2020
This paper backtests a nowcast of Japan's GDP growth. It follows the mainstream nowcasting literature in its use of a state-space model and a standard set of indicator variables. Its distinguishing features are use of genuine real-time data for Japan, a comparison with a market consensus forecast at 13 monthly horizons, and a new methodology for the revision analysis to account for the nowcast's response to new information. We find the nowcast to be competitive with the consensus at relatively short horizons and at several longer horizons. Our revision analysis indicates that the several-month-ahead forecast by the nowcast during disasters can be erratic due to parameter instability.
Keywords: nowcasting, real-time data, dynamic factor models, state-space models
JEL Classification: E17, E37, C53
Suggested Citation: Suggested Citation