A new model of personal income distribution : specification and estimation

Abstract

The research deduces a new model of income distribution by size from a set of elementary assumptions. Its main properties are analyzed and five methods of parameter estimation are proposed. The mathematical form of the Lorenz curve and the Gini concentration ratio associated with the specified model are also deduced. The model is fitted to the observed income distributions of four very dissimilar countries : Argentina, Canada, Sri Lanka and the USA. The fits obtained for the USA in 1960 and in 1969 are compared with those obtained using the lognormal, the gamma and the Singh-Maddala models, working with the sum of squares of deviations and the bounds for the Gini concentration ratio proposed by Gastwirth. In conclusion, the specified model fared better than the others for almost all of the fourteen properties introduced in this paper.