Feeding The Risk Management Quants Garbage

We have heard the old computer adage "garbage in - garbage out" to highlight how even a sophisticated computer program will produce nonsensical output if provided nonsensical input. The world of risk management is no different. The quants on Wall Street that are hard at work developing the next best trading and risk management systems are not perfect, but their job and measured performance becomes even more difficult when they are given bad information (see NY Times blog article).

On the surface, the goal of the risk management quants seem simply - tell me how much of the portfolio is at risk, and then tell me how much I need to sell, or how much capital I need to set aside so that I can sleep at night. In a sense, prepare me for the 100 year flood. Yet the 100 year floods seem to be occurring more often. Why is this? One possible reason is the "garbage in - garbage out" phenomenon, the problem of which is exacerbated as the markets continue to become more complex. Recent case in point, Lehman Brothers. As talk continued about a potential failure with Lehman, it became almost impossible to tell what their exposure was. Who are the counterparties? What are the default rates? What are the recovery rates? And most frightening of all, was does this new product even do? If companies cannot even understand the products they are selling, how can one expect to develop an adequate risk management system to help protect against the 100 year flood when it is not clear that water damage is even the problem, or that the strength of the levees is even important?

There is no doubt that some systems on Wall Street were provided optimistic data and assumptions, or had smoothed-out historical data in order to reduce the number of times the warning bells sounded, ultimately keeping companies from scaling back positions or redeploying capital to less profitable areas. But I suspect that there was an equal number of firms that diligently tried to provide the best information possible, but were simply in the dark. Why is this the case? There are no doubt a number of reasons, many of which are financial, but the separation between those that develop such risk management systems from those that develop products that need to be managed is not helping the situation. The information gap most likely goes both ways as the financial engineers are unaware of the workings of the risk management systems, while the risk managers are blind to the real exposures of the complex structured products that are being purchased and sold.

As more retail investors enter the markets, institutional trading continues to rise, and the securitization and engineering of products increases, both volatility and the values at risk will continue to affect markets. As new products are offered to the markets, it is essential that those who develop such products are on the same team as those who manage the risk. Risk management truly needs to be an enterprise-wide proposition, with incentives in place to reward those who properly managing risk, just as they are in place for those who engineer and sell the latest structured product. Performance on Wall Street is measured in money. If risk managers start to become rewarded in a similar manner, or if the risk management of new products begins to influence how new products are rewarded, we may then begin to find that the garbage provided to risk managers will begin to smell a little better.