The Assessment of Poverty and Inequality through Parametric Estimation of Lorenz Curves
The Assessment of Poverty and Inequality through Parametric Estimation of Lorenz Curves
by Camelia Minoiu (Economics) and Sanjay Reddy
The estimation of poverty and inequality often requires the use of grouped data as complete household surveys are neither always available to researchers nor easy to analyze. This study assesses the performance of two functional forms for the Lorenz curve proposed by Kakwani (1980) and Villasenor and Arnold (1989). The methods are implemented using the computational tool POVCAL, developed and distributed by the World Bank. To identify biases associated with this method of estimating the two Lorenz curve functional forms, the authors analyze unit data from several household surveys and a wide range of theoretical distributions. They find that poverty and inequality is better estimated when the data is generated from unimodal distributions than when it is drawn from multimodal distributions. For unimodal distributions, the biases in the estimation of poverty measures are rarely larger than one percentage point. Inequality (measured by the Gini coefficient) is well estimated in most cases considered. Neither of the two Lorenz curve estimation methods provides consistently superior performance, and performance does not always improve with the number of data points analyzed.
