A Dynamic Micro Panel Data Model on Individual Expenditure Habit: Hierarchical Bayesian Method

This research uses a hierarchical Bayesian method on dynamic panel data model to examine how unobserved individual expenditure habit affects parameters of inference. It is noticed that not accounting for the heterogeneity (individual differences) produces inconsistent estimates of the mean autoregressive coefficient, even for a panel with large N and T. Therefore, a great deal of interest was placed on hierarchical Bayesian estimation of unobserved individual heterogeneity of dynamic panel models, in order to improve on a static panel model. The method allows for unit-specific coefficients to be different across observations and imposing a stability condition for individual autoregressive coefficient drawn from a beta distribution (0, 1). The theoretical findings are accompanied by the use of primary data via a comprehensive questionnaire approach and extensive Markov Chain Monte Carlo (MCMC) experiments. The examination of all the figures and tables indicate that the Hierarchical Bayesian method effectively handled the complicated pattern exhibited by the individual habit especially at MCMC experiment as the dimension of N is large and T is small.