| Keywords [eng] |
pajamų nelygybė, paneliniai modeliai, daliniai kintamųjų atrankos metodai, LASSO regresija, klasterinė analizė / income inequality, panel models, partial variable selection methods, LASSO regression, cluster analysis |
| Abstract [eng] |
Econometric analysis of the factors determining income inequality in OECD countries This paper examines the relationship between income inequality and socioeconomic indicators in OECD countries in the period of 2007 - 2016. The theoretical part of the thesis reviews the scientific literature on income inequality, which allowed to classify the factors determining income inequality into five categories, that might be described as: economic growth or economic development, demographic, political, cultural - environmental and macroeconomic factors. In the practical part of this work, firstly, pooled and one-way fixed effects panel models are constructed for all countries. The residual errors of the models has not met the assumptions of the regression analysis of the panel data, and the structure of the models, where the slope coefficients are the same for all countries, does not help to identify significant factors affecting income inequality across different countries. Therefore, in order to achieve the aim of this work, a combination of cluster analysis, panel modelling and variable selection methods have been proposed as an alternative of the estimation of the different slope coefficients for each individual. After clustering that allowed countries to be grouped according to their geographical location and the social models operating in them, income inequality were modelled separately in each cluster. The K-means method used in clustering allowed to solve the problems of residuals heteroscedasticity and autocorrelation. Although double and partial selection methods has reduced standard errors of parameter estimates not in all cases, however these techniques helped to identify a statistically significant linear relationship between GDP growth and income inequality in North European and Anglo-Saxon countries and confirmed that higher tax revenue is associated with lower income inequality in four of the six clusters. |
