Poverty Study Using Nonparametric Estimation of Recurrent Survival Function

Abstract

The article analyses households’ poverty and nonpoverty duration. For this purpose survival function estimators for recurrent events were used: Wang-Chang estimator and two estimators proposed by Pena, Strawderman and Hollander (IIDPLE and FRMLE). We can conclude that survival probability for a long time out of poverty is greater than in the case of survival in poverty. Based on the graphical method we can conclude that the best estimator of survival in poverty and out of poverty is FRMLE. It means that we cannot assume that interoccurrence times within households are independent and identically distributed.