Lead PI: Prof. Gregory Connor

This WP concerns the use of semi-parametric estimation methods in security market risk modelling. Nonlinearities in factor risk exposures, together with multivariate risk dynamics discussed in work package 10, may help to explain the sudden value collapse of a large number of quantitative equity hedge funds in the three-week period beginning on August 9th 2007. The sudden price decline for hedge funds with leveraged long-short exposures to value and momentum factors could be ascribed to nonlinearities in the behaviour of these factors. Deepening our understanding of portfolio risk is of critical importance to asset managers generally, and hence this work package is of direct industry relevance.

Copula modelling of primary asset returns has particular relevance for European capital markets. Longin and Solnik (2001) and Poon, Rockinger and Tawn (2004) both find that copula-based tail dependencies between equity markets are strongest between the two large continental European markets (Germany-France). This high tail dependence can be ascribed to the greater interdependency of European markets compared to other national equity markets. This tail dependency is a very important concern in modelling the catastrophe risk of European equity market investment. As European capital markets continue to integrate and extend their high level of integration eastward; measuring the commonalities in extreme movements (particularly extreme losses) is an important subject of research. Copula models, and related models of tail dependency, are the natural approach.

In current business applications, copula models are particularly important for portfolio-based derivatives such as collateralised mortgage obligations (CMOs) and credit derivatives written against CMOs. These models have come under particular scrutiny following the collapse of the CMO market in 2007 and 2008 and deepening our understanding of these models is of relevance to the asset management industry.