Income Inequality Analysis in the Period of Economic Transformation in Poland
Domanski, Czeslaw, Jedrzejczak, Alina, International Advances in Economic Research
This paper presents some statistical methods of income inequality analysis based on the theoretical income distribution models that are well-fitted to the empirical models. As theoretical curves, the lognormal, gamma, Burr Type XII, and the Dagum models were used. They were applied to the earnings distributions in Poland in the period of economic transformation from a centrally-planned to a market economy. On the basis of the Dagum model, showing the perfect consistency with the considered earnings distributions, the maximun likelihood estimators of inequality parameters and economic distance ratios between men and women were calculated. (JEL C10)
The period 1991-99 was characterized by a series of fundamental changes in the Polish economy. The changes influenced, among other things, income and earnings distributions by size. It was connected, on one hand, with greater possibilities of economic activity for different social groups, and on the other hand, with a growing polarization of personal incomes. According to the latest OECD report concerning inequality and poverty, a growing income dispersion is one of the main determinants of economic growth. Simultaneously, it is well known that a high level of income inequality may have a destructive influence on economic growth and development. Therefore, it seems worthwhile presenting income and earnings distributions and their inequality in the period of transition from a centrally-planned to a market economy.
This study presents some inequality measures useful in income distribution analysis. Besides the well-known Gini ratio, the Bonferroni index was also applied, which is advisable when the major source of income inequality is the presence of economic units with a much lower income than other members of the population. The Zenga curve, which measures the so-called point concentration, was used to test changes in the inequality level. Moreover, the Atkinson measure connected with the concept of social welfare was taken into consideration.
To assess the relative deprivation of one income group with respect to another, inequality measures between income distributions, called economic distance ratios, were used. These inequality measures were estimated by means of the maximum likelihood method on the basis of theoretical distribution parameters. As theoretical distributions, the Dagum, gamma, Burr Type XII, and the lognormal curves were considered. They were compared with the empirical distributions from the point of view of their consistency.
Earnings Distributions by Means of Theoretical Models
In many situations, it seems reasonable to use theoretical distributions, which show high consistency with empirical distributions. First, such an approach enables for the flattening of irregularities in empirical data coming from the method of gathering information. Second, the use of theoretical distributions simplifies and accelerates the analysis because all distribution characteristics can be expressed by the same parameters. Moreover, it is easier to estimate inequality measures on the basis of a random sample, knowing the mathematical form of density or cumulative distribution function.
A variety of probability functions has been suggested as suitable in describing the distributions of income by size. Among them, the Pareto, lognormal, and the gamma models were the most often used. The Pareto curve is considered to be an ideal theoretical distribution for high income groups. The lognormal and gamma distributions have the advantage that they describe the entire range of positive income. The above mentioned models depend only on two parameters that simplify the calculations, but decrease the opportunity of perfectly fitting the theoretical curve to the empirical points.
From among several three- or four-parameter curves used in the analysis of income distributions by size, only a few can be regarded as good income distribution models. …