Bayesian and frequentist modelling of West African economic growth : a dynamic panel approach

The empirical outcomes of previous studies examining the relationship between economic growth and socio-economic indicators have been inconclusive and conflicting. To further probe into the study area, the current research employed a dynamic panel model estimated via three robust dynamic panel data estimators of the generalized method of moment (GMM), frequentist instrumental variable (IV) and the Bayesian IV on real and simulated data. Various model performance criteria such as Wald statistics, leave-out-one crossvalidation and the Pareto k checks were used for validity verification. The results of the robust diagnostics checks and a model strength metric showed that the family of IV models outperformed the GMM. Thus, the estimation provided by the Bayesian IV is upheld and recommended in modelling dynamic panel data as it provides robust estimates of the parameters of interest.