Parameter Estimation of Singh Maddala Distribution by Moments

Abstract

Singh Maddala Distribution is a flexible distribution and mostly used for modeling the income, wage, expenditure and for wealth distribution of a country. To model the density of the income distribution, we have derived L-moments and TL-moments of the SMD in closed form. These moments are used to estimate the scale parameter which is related to the inequality of the income distribution. This study shows that L-moments and TL-moments are better alternatives against conventional moments and estimators of these moments equally applicable for both small and large sample size. Through Monte Carlo simulation study, we find TL-moments estimators provide least bias and smaller Root Mean Square Error of statistic, coefficient of variation, skewness and kurtosis. We also find that TL-moments are superior to L-moments if we are allowed to discord extreme value, otherwise L-moments provide the best fit.