Checking the reliability of the model of dependence of Russia's GDP on exports by testing random balances

The article discusses the construction of a paired linear regression model of the dependence of the used GDP on Russia’s exports and the assessment of its parameters. The parameters of paired linear regression are calculated for a set of 24 values of Russian exports and used GDP. The homo-, heteroskedasticity of residues in paired regression is checked. The presence of heteroscedasticity can lead to bias estimates of regression coefficients. The unbiased estimates mainly depend on compliance with the assumption of the independence of the residuals and the values of the factors. Heteroskedasticity will affect the decrease in the effectiveness of parameter estimates. The following tests are consistently considered: Goldfeld-Quandt, Spearman, Glazer, Park, White rank correlation, Broich-Pagan criterion. In the Goldfeld-Quandt test, the resulting sorted array is divided into two equal parts. Spearman’s rank correlation test is based on calculating the rank correlation coefficient between the values of the regression residuals and the values of the regressor factor. The value of the Student’s statistical criterion is determined, its value is used to judge the presence or absence of heteroscedasticity of residues. The Glazer test is based on the most general ideas about the dependence of the standard error of the random component on the values of the explanatory variable. Among the analyzed models, the model with the value for which the parameter is most significant is selected. It is proposed to use an additional functional dependence of the prologarithmic type to conduct a study on heteroskedasticity. The statistical significance of the regression parameter according to the Student’s t-criterion is checked. The Broich-Pagan test is used if there are grounds to believe that the variance of errors may depend on some set of observed variables. This information can serve as an important management tool for various economic agents of the market.