Method for Correcting Panel Data Heterogeneity in the Models of Complex Economic Systems

The article is devoted to the construction of models for panel data that take into account the influence of qualitative features on the endogenous variable. Dummy variables are an econometric tool that formalizes the influence of qualitative features. The need to include dummy variables in econometric models was recently dictated by structural changes in the economies of a number of countries caused by unprecedented Western sanctions and changes associated with the pandemic. The aim of this study is to develop and empirically test a model of Russia’s trade turnover with BRICS countries using panel data techniques that account for the impact of qualitative (time-invariant) factors on the endogenous variable, employing Fixed effects vector decomposition (FEVD) method. FEVD approach provides a more flexible model specification by combining the advantages of fixed and random effects models without relying on the strict assumption of zero correlation between individual effects and regressors, which is typical of random effects models. This enhances the capabilities of panel data analysis in econometrics and enables more accurate modeling of the influence of qualitative factors on the endogenous variable. The objectives of the study include: building models for panel data based on BRICS data, conducting their specification tests, implementing FEVD method algorithm in the R software environment, and algebraic and empirical verification of the properties of the method parameter estimates. The result of the work is adaptation of FEVD method to the specifics of BRICS economies in the context of modern economic challenges. Balanced panel data for five BRICS countries (Brazil, Russia, India, China, South Africa) for the period 2000–2020 were used as an empirical base. Particular attention is paid to the analysis of the impact of macroeconomic indicators (GDP, dollar exchange rate, oil price, pandemic shock, etc.) on Russia’s trade turnover with BRICS countries. The FEVD method made it possible to increase the accuracy of the estimation results in comparison with the traditional fixed effects model. The study contributes to the empirical base for estimating fixed effects models using FEVD method.