Essays in global comovements

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
University of Alabama Libraries

This dissertation looks into comovements in the global macroeconomic aggregates across countries by applying Bayesian econometric models. The first essay provides new results on the significance and relative importance of global and regional comovements in sovereign credit risk. I employ a dynamic factor model with time-varying stochastic volatility to examine the time-varying effects of global comovements. I find that the effects of the global comovements on individual countries are smaller than commonly perceived, especially after the end of the Eurozone debt crisis. Moreover, contrary to previous findings, I find that the net effects of the global and regional factors are greater than those of global macroeconomic variables. The second essay provides evidence of significant international co-movements of public debt in the form of the common global and regional factors. International events such as the global financial crisis and Eurozone sovereign debt crisis suggest the existence of global and regional factors that can generate synchronizations of public debt across countries. In contrast with previous studies that are focused mostly on domestic economic fundamentals in explaining public debt, I find distinct global factors in the public debt-to-GDP ratio, from both principal components analysis and the Bayesian dynamic factor model. I show that the global factor accounts for a significant fraction of the variation of public debt often more substantial than those explained by domestic variables in many countries. In the third and final essay, I develop a new panel cointegration model based on the well-known autoregressive distributed lag (ADL) models. I adopt the generalized method of moments procedure of Arellano and Bond, noticing that the ADL cointegration model is a special case of dynamic panel data models. The suggested procedure can overcome the difficulties found in the studies based on the panel vector auto-regressive models, which can be biased in the presence of cointegration. I apply the suggested procedure to the issue of defense-growth relationship, while capturing the inter-temporal dynamic relation between military spending and economic growth. The results reveal that there are multiple cointegrating vectors in the relationship of military spending and growth.

Electronic Thesis or Dissertation