Investigating poverty and labour force participation among older population in Egypt: a multilevel simultaneous equations modeling approach

摘要

This study has investigated the relationship between elderly poverty and their labour force participation based on the Egyptian Household Observatory Survey - Round 7 dataset (IDSC, 2010). I have considered two methodological problems that plague many existent researches, endogeneity and hierarchical structure of the data. A measure of poverty in old age that captured five broad dimensions of poverty has been constructed using factor analysis technique. To investigate the main determinants of poverty and labour force participation, I have developed four single-level models and four multilevel-models for poverty and labour force participation. I found that poverty is endogenous to labour force participation. Consequently, a simultaneous equations model that correct for this endogeneity is considered. In addition, I found significant differences among governorates regarding elderly poverty and their labour force participation. To overcome the problem of the dependences among observation within each governorate and to provide more accurate results for regression parameters and their standard errors, I have developed a multilevel linear model for poverty and a multilevel logistic model for labour force participation. To consider both problems simultaneously, I have proposed a more developed model; a multilevel simultaneous equations model. I have also performed a simulation study to formally asses to what extent the endogeneity problem in the hierarchical data structure cannot be ignored. The results showed that the biasness and the accuracy of the parameters associated with the endogenous variable differs according to the strength of the endogeneity and based on the level at which the endogeneity occurs.