Global Financial Risks and Changes in Conditional Value-at-Risk

Abstract

The September 2008 collapse of Lehman Brothers was the 9/11 on Wall Street, and many articles had been written on the changes in the global risk landscape that followed. However, there is scarcity of rigorous studies using empirical data and advanced econometric methods to verify such a change and the nature of such a change. In this paper, we provide rigorous analyses of statistically significant changes in global financial risks and sharp increases in conditional Value-at-Risk after September 2008. We perform the statistical analyses using conditional distributions on the tail losses of equity portfolios constructed from the stock indexes of the major global financial markets based in U.S., U.K., Germany, France, Japan, and Hong Kong. These 6 equity portfolios represent the largest equity markets in the world and their co-movements led by the U.S. financial market capture the major market sentiments and outlook of the world. Employing the generalized marginal Pareto distribution and multivariate copula method, we provide rigorous tests of the changes in the shapes of the conditional tail losses over time, and also tests of the changes in the conditional Value-at-Risk or conditional expected losses. The Clayton copula also indicates that whenever equity portfolio losses occurred after September 2008 till June 2011, a significant increase in associations of high losses across countries occurred in conjunction with significant higher loss in each country. Thus there is now strong empirical evidence supported by rigorous methodology to assert the prevalence of heightened global financial risks and its contagion effect across the globe. An important implication arising out of these conclusions is that banks under BASEL II and BASEL III and financial institutions in the near future should not underestimate its Conditional Value-at-Risk by using the normal distribution model since under stressed situations past September 2008, the portfolio return distributions have tails that simultaneously grow longer and thinner in the direction of the loss region.