Modeling the dynamics of income distribution in Russia

In this paper, the four-parameter generalized beta distribution of the second kind (GB2) is applied to simulate the distribution of the income of Russian population based on the quarterly micro-data of household income for the period from 2003 to 2015. The distribution parameters were estimated via the maximum weighted-likelihood method, and the distributional parameter estimates were aggregated into quarterly time series. The time series have undergone the decomposition by the STL method. ARIMA and exponential smoothing models were applied to the trend component of the time series, and the distributional parameter forecasts were produced. Based on the predicted values of the distribution parameters, several inequality measures was estimated, such as at-risk-of-poverty rate, relative median poverty gap, quintile share ratio and Gini index. Thus, the robust estimates of inequality measures were obtained, the prediction accuracy of which was about 5%. An analysis of the dynamics of distributional parameters yielded an interesting conclusion that during the crisis periods the nominal level of income inequality decreases, in contrast to common apparent belief that negative macroeconomic shocks induce higher inequality.