Evolution of tail risk during the global financial crisis (2008)

Abstract

The primary objective of this thesis is to understand the behaviour of extreme events taking place in the foreign exchange (FX) and stock exchange (SX) markets around the Global Financial Crisis using parametric and non-parametric methods. Specifically, it models the evolution of the extreme values and Black Swans measures of daily stock returns and volatility proxies from a set of more than thirty SX and FX markets and compares their statistical properties in the pre- and post-crisis periods.My findings suggest that irrespective of the modelling approach followed, there are substantial differences in the tail-behaviour of the two periods that cannot be accounted for by the mainstream mean-variance model although the identification of latent Non-linearities in the underlying mean and/or variance dynamics does indeed lead to a significant reduction in tail asymmetry and kurtosis. Moreover, the mainstream parametric models prove insufficient in encompassing the volatility tails. Consequently, this study demonstrates that during times of extreme financial turbulence in financial markets the existing financial risk models are inadequate to provide guidance for the probability of extreme events taking place and subsequently underlines the need for additional research on parsimonious models and for the use of encyclopaedic exercise of judgment in the real-world risk management.