This blog post was originally published on TabbFORUM.
Late October 2012 saw an RMA conference on risk management in Dallas, the annual FIA conference in Chicago, and a catastrophic natural disaster in and around New York. While these were separate events with very different impact levels, all three were connected by the complexity of modern risk management, what it means, and how its outputs can be understood and used.
Conversation at the Risk Management Association event in Dallas was focused largely on developing, implementing and managing a risk appetite within a financial institution. This tackled from multiple angles, including liquidity risk, market risk and credit risk. The recurring theme of the discussions was the difficulty in developing a framework that works both top-down from an enterprise-wide risk tolerance perspective, and bottom-up from an individual risk contributor’s perspective.
Risk appetite definitions ranged across benchmarked “active” risk budgeting (VaR against benchmarks), economic capital, collateral and funding costs, VaR and VaR shortfall. As has been noted before, it is interesting that in the aftermath of a financial crisis which saw VaR receive critical attention, its use is expanding into bilateral calculation of collateral and central clearing margins.
It was those same central clearing margins that took center stage at Chicago’s Futures Industry Association conference. As a business more used to exact calculations built on lot sizes and observable prices, the relative complexity, convexity and simulation-based margining of interest rate swaps are causing many to strengthen their technical architecture and approach. As the single largest OTC derivative class heads to the margin model, FCMs and their clients are becoming familiar with historic simulation, decay factors and look-back periods. At the same time, the margin posted is looking more and more like an alternative for economic capital, even using the same techniques to generate the number.
The VaR number itself, of course, is essentially predicting the minimum loss that would be expected to happen at the specified percentile. For example, a 99th percent VaR predicts the minimum loss that should be expected to happen once in every hundred days.
New York, New Jersey, and Neighboring Areas
As the conferences in Dallas and Chicago were happening, the tri-state area of New York, New Jersey and Connecticut were actually experiencing a devastating weather event.
The parallels between a VaR calculation and this “perfect storm” are as inescapable as they are shocking. Defined variously as a tail event or a 2.33 standard deviation event, the VaR cases are typically characterized by multiple loss-causing events occurring within the same simulation. This is a low percentile by definition, but has an extremely large impact. In terms of what faced the northeast U.S., a high tide and a hurricane create the worst possible outcome for the shoreline areas. Here, risk management hits closer to home.
Risk management is a control function that mitigates, to a degree, the impact of the extremes. It is also a fact that events deep in the tail are often uncontrollable. It was with shock that reports of Hurricane Sandy’s effects were viewed, and with a sense of humility that the aftermath restorations are being tackled.
Risk can often be seen as a dry academic subject, as more theory than reality, and the discussions in Dallas and Chicago could lend weight to that opinion. However, the events in the tri-state area graphically and violently demonstrated that tail events do happen, and enormous damage can be caused when extremes collide.
While you’re here…