Warrington College of Business Administration, University of Florida
By now there are several indicators suggesting that the global economy has moved away from the major collapse feared in late 2008. However, high uncertainty remains palpable among investors. This is partly due to the fact that we still do not fully grasp the detailed mechanism behind the financial crisis we just experienced. Challenges abound both in terms of mathematical modelling improvements and also in defining the right economic/financial concepts to assess the damage done and prevent a repeat of past disasters. As this journal has in recent issues, the papers contained herein all deal with issues that directly pertain to the present financial environment.
Following the recent financial crisis, many question whether it is prudent for the financial industry to continue to rely so heavily on simple, standard quantitative risk models and risk measures. Value-at-risk (VaR) and Gaussian models for asset returns have been criticized even in the popular press. Two obstacles impede the adoption of more advanced models and measures. One is the difficulty of statistically estimating the parameters for more complicated models well enough to produce a reliable estimate of a risk measure, such as expected shortfall (ES), that is more sensitive to the tail behavior of the loss distribution than VaR is. The second obstacle is that using advanced models to measure the risk of a portfolio that contains derivative securities may require a computationally expensive nested simulation. The article by Liu and Staum addresses this issue by developing a computationally efficient nested simulation procedure for ES.
Recent experience has also now taught many that unpredictable, correlated large jumps and volatility need to be accounted for when modeling asset prices. In their paper, Xu et al present a technique for dynamic asset allocation in such a context, where the optimization criterion is the trade-off between return and risk measured as conditional VaR. Their approach addresses a major difficulty in the practical implementation of stochastic programming for this risk measure and should prove very useful given the increased interest expressed by practitioners. In a numerical study they show that large jumps play a critical role, as they induce a lesser allocation to risky assets in comparison to standard models.
The recent crisis has also highlighted the need for better criteria for risk management. In particular, how should diverse capital at risk be aggregated for a single organization? This specific issue has even been brought to the fore during the bailout of some financial institutions and the unwinding of those left to fail. The standard aggregation of credit, market and operational capital does not account for the interaction among them as can be clearly illustrated in structured products. In their paper, Brockmann and Kalkbrener present a multi-period extension of a multi-factor approach that enables the incorporation of different liquidity horizons and accounts for common dependence between market and credit risk attributed to specific factors.
Concentration risk in credit portfolios is not only an important topic for individual banks, but also for the whole banking system; this was impressively demonstrated during the financial crisis. Thus, it is essential to measure concentration risk for banks’ internally, as well as for regulatory purposes. In their paper, Gürtler et al propose a methodology to adjust multi-factor models in such a way as to to measure concentration risk and to deliver results that are consistent with Basel II. Within a simulation study, the authors analyze the impact of sector concentrations on several portfolios and compare the performance of different multi-factor approaches. In this context, the authors also test whether the common criticism of VaR, typically illustrated by contrived portfolio examples, is also relevant in more realistic settings. The empirical results support the view that it is straightforward to apply VaR instead of the ES in order to measure sector concentration risk in credit portfolios.