Features of forecasting macroeconomic indicators based on the use of the mean-adjusted BVAR model
Abstract. The article investigates specific features of forecasting macroeconomic indicators using the mean-adjusted BVAR model. The BVAR model is widely used for analyzing economic time series, but its predictive ability can be improved by including an adjustment for the average value. The authors analyze the effectiveness of forecasting based on the mean-adjusted BVAR model using the example of various macroeconomic indicators. The study showed that the mean-adjusted BVAR model is more effective than other models for forecasting inflation, industrial production index and money supply. It copes particularly well with long-term forecasts and surpasses the traditional BVAR model due to the updated specification. The scientific novelty of the study lies in the systematic selection of the optimal hyperparameter for the a priori distribution of Minnesota and the comparison of the predictive power of mean-adjusted BVAR with competing models based on Russian data. The results of the work will help to improve the quality of economic forecasts and improve the efficiency of decision-making in an unstable economic environment.
Keywords: mean-adjusted BVAR, macroeconomic indicators, data, modeling, forecast, Minnesota prior distribution.
Highlights:
- macroeconomic data were collected and processed, and brought into a single standard format. After that, the sample was divided into educational, test, and training data sets;
- the model was estimated on the training sample and a forecast was built, the hyperparameter of the prior distribution was optimized, the mean square forecast errors were calculated for each model, and the ratio of the mean square forecast errors was determined;
- a comparative analysis of the forecast accuracy of the various models under study was conducted.
Irina A. Eremina, Vladimir V. Vallask - Peter the Great St. Petersburg Polytechnic University, St. Petersburg, Russia