Forecasting Russia’s GDP through the production method

Abstract

The paper analyses the variables covering real, financial and external sectors of the economy alongside various sectoral, price and survey indicators. We have obtained forecast values of gross value added by industry and an aggregated estimate of Russia’s productionbased GDP. Drawing on dynamic factor model (DFM) as the main approach, we obtained quarterly point forecasts for production-based Russian GDP and for individual sectors for 2011-21. The forecast accuracy is compared to Bayesian vector autoregression (BVAR),s imple benchmarks based on aggregated and disaggregated GDP modeling and consensus expert forecasts. The results of the study show that the dynamic factor model outperforms the benchmarks and, in some cases, also BVAR in forecast ability (measured on the basis of an out-of-sample forecast error). The superiority of the factor model can be traced back to its ability to capture sectoral information on the gross value added of individual industries. The covariance matrix analysis of sectoral forecast errors confirms that the factor model superiority is based on its ability to capture sectoral dynamics more accurately, especially during the periods of high volatility. The dynamic evaluation of point forecasts for 4 quarters ahead and comparison of modeling results with consensus forecasts of experts shows that forecasting based on the DFM model for production allows for more stable and consistent results.