Estimating the Income Distribution in Latin America Using Limited Information

Abstract

Latin America is one of the most unequal regions in the world. In fact, this situation has been prolonged from the seventies to the nineties. The lack of convergence over time is becoming absolutely alarming. This trend notwitstanding, welfare has increased in that region and poverty has been reduced, thus indicating that inequality remains constant despite economic growth. Hence, an accurate estimation of income distribution is crucial to implement policy initiatives and to assess results in a region where high inequality is almost a tradition. In this paper, we estimated the income distribution in Latin America using limited information combining a measure of inequality (i.e. the Gini index) which is taken from the UNU-WIDER database and an indicator of average income (per capita GDP) provided by the World Bank database. The proposed methodology to estimate the income distribution of Latin America from country level data is comprised of two steps. In a first stage we estimate individual country distributions using two versions of the Singh Maddala [Singh, S.K., Maddala, G.S., (1976). A function for the size distribution of incomes. Econometrica] and the Dagum [Dagum, C. (1977). A new model a personal income distribution: Specification and estimation. Economie Appliquee] distributions. These versions are called the Lame type I and Lame type II distributions, which present several theoretical advantages [Sarabia, Jorda and Trueba, (2013). The Lame class of Lorenz curves. Communications in Statistics, Theory and Method, forthcoming]. Thereafter, the regional distribution is developed from a finite mixture of these families using population shares. Finally, several features of the Latin Int Adv Econ Res (2014) 20:231–232 DOI 10.1007/s11294-013-9445-9