Global panel data modelling of industrial output

Abstract

This study estimates panel data models of industrial output by means of fixed effects, time fixed effects, random effects and pooled panel data models. The models are specified for the main data set which includes all available countries and subsets of countries according to income levels i.e. low income countries, lower middle income countries, upper middle income countries and high income countries. Hausman tests reveal that fixed effects model is appropriate for all countries model, low income countries and upper middle income countries. Meanwhile, fixed time effects model is appropriate for high income countries and Chow stability test reveals that pooled panel data model is appropriate for lower middle income countries. Diagnostic analyses of estimated models indicate that all models suffer from the problems of cross sectional dependence, non-constant variances and serially correlated errors. As such, this study applies robust standard error estimators and derives its final conclusions based on the most reliable results. Based on robust fixed effects model, household consumption, government consumption and money supply are statistically significant regressors for the global economy’s industrial output. Meanwhile, government consumption, money supply, interest rate and trade openness are statistically significant regressors for industrial output in low income countries. Based on robust pooled panel data model, all regressors are statistically significant for lower middle income countries. Robust fixed effects model of industrial output for upper middle income countries reveal that household consumption, government consumption, money supply and inflation are statistically significant. While the regressors that are statistically significant for industrial output in high income countries based on robust time fixed effects model are household consumption, government consumption, factor years of 2009 and 2010.