Estimating Multidimensional Poverty Index And Its Growth Trajectories Using SEM
Author: Umair Khan


In this study, we estimate poverty in the form of a composite index by aggregates their respective indicators into a scalar score. In this study the basic idea is to capture the broader dimensions of poverty such as education, sanitation, housing, health, and other aspects as well. The analysis is based on Pakistan Social and Living Standard Measurement (PSLM) Survey 2014-15 and 2019-20 collected by the Pakistan Bureau of Statistics (PBS) is used to measure poverty at the national, regional and provincial level and providing empirical evidence on its growth trajectories as will to examine the poverty status of a particular region or province over time. In this study SEM approach is used for aggregating indicators into a composite MPI. Multiple-group comparisons in structural equation modelling were used for analysis of differences in the measurement model across provincial-wise as well as for urban and rural regions. The results of the study revealed substantial variations between urban and rural respondents in the conceptualisation of poverty. The results indicate that each sub-population consider respective items of population differently. Its means that the concept and meaning of poverty in different regions and provinces is different. The results of the study showed that the poverty level in rural areas is higher than urban areas. By comparing the poverty level in provincial wise Panjab has lowest level of poverty, and Baluchistan has worst situation as compare to the rest of provinces. For the observation of change over time in MPI we use latent growth model. The results revealed that there is reduction of poverty observed over time, but the reduction level vary for the different regions and provinces.

Meta Data

Keywords : Confirmatory Factor Analysis, Latent Growth Curve Model, Multidimensional Poverty Index, PSLM, Structural Equation Modelling
Supervisor: Ahsan ul Haq

Related Thesis​