Abstract
The problem of spatial inequality in the regions of Kazakhstan has been relevant for many years and in recent years this problem has worsened due to socio-economic changes in the areas, the consequences of the COVID-19 pandemic, the negative impact of inflationary processes, increased internal migration and other factors. The purpose of this study is to analyze spatial inequality between 16 regions of Kazakhstan covering the period from 2001 to 2017. The following scientific methods were used in the study: historical method, and statistical method. During the research, a new class of spatial econometric models was developed, which are modifications of the Durbin spatial model. These models are characterized by variable coefficients with spatial lags of the dependent and independent variables. The models were evaluated based on information about Kazakhstan’s regions, using the regional gross domestic product per capita as a dependent variable. The findings of the study show the advantages of the SDM model with fixed effects compared to alternative models, which is confirmed by the results of the assessment using the criteria of the Akaike Information Criterion (AIC) and Bayes (BIC). According to the SDM model, a 1% increase in gross regional product per capita in the base year leads to an increase in the growth rate of gross regional product per capita, all other things being equal. It is also worth noting that an increase in the unemployment rate by 1% contributes to an acceleration in the growth rate of the gross regional product per capita by 0.451, all other things being equal. An increase in government spending per unit in the region contributes to a decrease in the growth rate of the gross regional product per capita in the neighboring region, all other things being equal. The spatial lag coefficient indicates that changes in the indicators of the domestic regional product per capita in one region have an impact on changes in the domestic regional product per capita in the neighboring region. The results of the study indicate the need to use spatial weights when evaluating regional regression models.