The Accuracy of OECD Forecasts for Japan

In: Pacific economic review, Band 2, Heft 1, S. 25-44

Abstract

A variety of accuracy measures, error diagnostics and rationality tests are applied to the OECD's macroeconomic forecasts for Japan of aggregate demand and output, inflation and the balance of payments. It is found that the OECD forecasts are superior to naive no‐change predictions and forecasts generated by simple autoregressive time‐series models. Most forecasting error is nonsystematic. As predictors of direction the OECD's six‐month ahead forecasts should be considered valuable; this cannot be said for forecasts which look ahead a year and 18 months. Many forecasts fail bias, efficiency and consistency tests so that the rational expectations hypothesis is not generally supported.