Detecting nonlinear dependencies in eurozone peripheral equity markets: A multistep filtering approach

Abstract

This paper assesses the presence of linear and nonlinear dynamic causal relations among the stock markets of the Eurozone peripheral countries, namely Portugal, Ireland, Italy, Greece and Spain (PIIGS countries). In addition to the pairwise analysis, this paper tests causality in a multivariate setting to take into account the effects of all variables by using a three-step multivariate filtering procedure. The paper also investigates the existence of nonlinear causal linkages of VAR/VECM filtered residuals to verify whether the observed causalities are strictly nonlinear and controls for conditional heteroskedasticity using a multivariate Dynamic Conditional Correlation (DCC) GARCH model to test the hypothesis of nonlinear noncausality. Most of the nonlinear causal linkages were purged after multivariate GARCH filtering, a fact which indicates that volatility effects primarily induce nonlinear causality. Yet in some cases that nonlinear links pertained, possibly peripheral stock returns may exhibit statistically significant higher-order moments. Our results may be useful in explaining a significant part of the (non)predictability of peripheral stock markets but more importantly in quantifying the process of financial markets and banking integration in the Eurozone.