966
Serial number: 08/2015
Author(s): M.Mehdiyev, V.Ahmadov, S.Huseynov, F.Mammadov
Language: Azerbaijani
Date: 2015
Abstract: In this paper, we study main problems and practical issues of modeling and forecasting of macroeconomic variables in the national economy. For that, we employ astructural VAR models and estimate interdependencies among different economic variables. Initial data analysis of macroeconomic variables incorporated into the model show that there exist certain anomalies and high volatility in their dynamics. Only that fact alone can be considered among important factors which strongly contribute to weak interdependencies among variables. It is not surprising that researchers using different macroeconomic variables on the national economy cannot find statistically significant relations among them. Here, we use different statistical methods to reduce volatilities and anomalies present in their dynamics. We estimate VAR models with different number of variables and employ two forecasting methodologies (Waggoner and Zha (1999), Banbura, Giannone and Lenza (2014)) to construct our conditional and unconditional forecasts on several variables. In the end, we demonstrate practical problems of modeling and forecasting of macroeconomic variables and provide some policy recommendations to improve quality of statistical data on the national economy.
Key words: Forecasting, time series methods, Bayes methods
JEL classification: F47, C11
