Extreme dependence and risk spillovers across north american equity markets

Abstract

This study analyzes risk spillovers across North American equity markets over 1995–2016. Downside and upside Conditional Value-at-Risk (CoVaR) are estimated after modeling the dynamic dependence structure for each equity market pair using generalized autoregressive score (GAS) copulas. US-CAN and CAN-MX dynamic correlations trend upwards over the sample period while the US-CAN correlation fluctuates around a higher long-run average. Conditional tail dependence is symmetric and significantly higher following the Global Financial Crisis (GFC) in all cases, implying greater co-movement under extreme economic conditions. Downside and upside risk spillovers are significant and asymmetric along two dimensions for each equity market pair, where downside risk spillovers are more severe and the degree of asymmetry by conditioning direction is rank ordered by relative equity market sizes. Aside from the US-CAN pair, downside and upside risk spillovers are significantly larger following the GFC as compared to the pre-crisis period. Observed asymmetric and time-varying behavior is consistent across high/low risk and high/low risk spillover sub-periods.