Hitotsubashi Institute for Advanced Study


The Efficiency of the Government’s Revenue Projections


This paper evaluates the efficiency of the Japanese fiscal authority’s revenue projections from 1960 to 2020 using real-time data. Revenue projections are not efficient, primarily due to the conditioning projections of output growth. By adjusting the forecasts based on the results of real-time forecast evaluations, this paper finds that the out-of-sample accuracy of the one-year-ahead projections could be significantly improved by a magnitude of up to 10 percent in root mean squared errors. The analysis of the disaggregated series suggests that corporate tax projections are the least efficient. The fiscal authority’s loss function is estimated to be asymmetric, making the underprediction of revenues more common.

Report No.: HIAS-E-122
Author(s): Natsuki Arai(a), Nobuo Iizuka (b), Yohei Yamamoto (c)
Affiliation: (a) National Chengchi University
(b) Kanagawa University
(c) Hitotsubashi University
Issued Date: September 2022
Keywords: Revenue Projections, Japan, Forecast Evaluation, Real-Time Data, Out-of-Sample Forecast Accuracy
JEL: C53, E62, H68.