Forecasting Non-Oil&Gas Revenues in the Russian Budget

Abstract

This paper examines methods for forecasting non-oil&gas revenues in Russia’s budget system in order to construct a model with the smallest forecast error in conditions of the incomplete availability of the necessary data. We use mixed data, error correction, and autoregressive distributed lag models to forecast non-oil&gas revenues at different levels of aggregation: non-oil&gas revenues minus one-time receipts; cross-border and domestic duties; and eight groups of major taxes. The quality of the models is estimated using the root mean square errors of a pseudo out-of-sample forecasts on one- and three-year horizons. The forecasts of the models constructed turn out to be more accurate than the forecasts of the Ministry of Finance of Russia for most forecast periods.