Extreme conditional value-at-risk : a scenario for risk management

M.Comm. (Financial Economics)Systemically important international institutions that were too “big to fail” such as American International Group, Citi Group, Merrily Lynch, UBS and MF Global to name a few, were bailed out from their financial problems by their respective government. Besides the immediate substantial financial costs that were incurred globally, banking sector problems associated with the US mortgage-backed securities spread to other countries and had a significant negative impact on their real economies – many countries went into recession, unemployment increased and production levels declined. It is now 2012, three years after the crisis and global economic activity is yet to return to pre-crisis levels. Given the substantial losses that were incurred globally and claims that the financial crisis was caused by the failure of risk management, an investigation of the inadequacies of risk management as a discipline that developed to safeguard the world from such financial havoc is not only necessary but undoubtedly required. Therefore, this mini-dissertation investigate the inadequacy of risk management, specifically, the inadequacy of current models that are used to estimate market risk. Traditional Value-at-Risk (VaR) models and two extreme value theory (EVT) distribution models: the generalised extreme value distribution (GEV) and the generalised Pareto distribution (GPD) are used to estimate potential losses in order to evaluate the appropriate model for estimating losses in extreme market volatility. The main findings are that EVT-based models accurately estimates both downside and upside losses during extreme market volatility, and therefore must be used as internal bank models for market risk