William G. Ferrell President, Ferrell Capital Management Greenwich, Conn. For the one-day horizon of bankers and dealers, Value-at-risk can be applied to all instruments. For the longer horizon of fund managers, simulation and downside risk enhance VAR to more accurately measure the risk in securities and derivatives with embedded options. These models represent only the first part of a thorough risk management function. We look at risk management as a combination of science and art, and we understand that there are events that can and will happen that are not embedded in anyone's model. Portfolio managers should consult with a pool of seasoned market professionals, either internal or external, to formulate reasoned judgments about a portfolio's risk to contain risk (defense) or to relocate capital to reflect changing risk (offense). Ferrell Capital has developed powerful tools, such as Heat Maps, to measure extreme market scenarios by shocking key drivers to risk. Demonstrating the aggregate portfolio's VAR under varying market scenarios provides a road map for shifting risk exposure. Additionally, we use Alpha VAR to monitor a portfolio manager's tracking risk, the excess risk between the benchmark and actual portfolio. The intent is not so much to predict the future in terms of where markets are headed. Rather, VAR is a scientific way to address what important changes in volatility and covariance of positions and markets will have on bottom line performance.
Tanya Azarchs Director, Standard & Poor's Financial Institutions Ratings New York Right now, there isn't enough uniformity in value-at-risk systems for anyone to get numbers they are comfortable with. Maybe the numbers are not comparable between institutions. There is model risk associated with VAR, and while they appear to have performed well in predicting the level of variation in daily earnings, you still cannot be confident that they could predict what would happen if there were another indicator that shocked the marketplace. It works from historical data, and it assumes that the future is going to be a lot like the past a most recent past. Strict VAR does that, but in addition, you will need scenario testing where you can input newer stress tests. Peter Gallant Treasurer, head of market risk policy,Citicorp, New York Value-at-risk is one of the tools to monitor risks, with its main purpose in the quantification of market risk in an effective market-related manner. On its own it would not be sufficient to monitor trading risk in derivatives or any other financial products.Peter Gallant Treasurer, head of market risk policy,Citicorp, New York Value-at-risk is one of the tools to monitor risks, with its main purpose in the quantification of market risk in an effective market-related manner. On its own it would not be sufficient to monitor trading risk in derivatives or any other financial products.
Garrett Glass Chief market risk officer, First Chicago NBD Corp., Chicago Conceptually it cannot be done, and practically it cannot be done. In our bank, we have a unified model task force, (where) our experts look at our model and put into the equation the credit exposures and value-at- risk exposures and work out our capital and return-on-capital accordingly. So this is a cross-risk model. It is possible but rather daunting to do that. Just within the capital markets area, it is difficult enough to put in some of the more exotic derivatives that are not easily amenable to analytic solutions. Statisticians have simulations on top of simulations and it gets rather complex. There are some theoretical problems with bringing certain complex instruments into any single model, even within risk types and product types. Then there are the practical problems. It is still very daunting to get all this information in. Even within our capital markets, there is a part we have to capture manually complex derivatives, municipals, securities transactions because they are very specific to a market and we don't get regular feeds on the valuations.
Tanya Styblo Beder Principal,Capital Market Risk Advisors Inc., New York Value-at-risk provides a valuable to tool to measure risk on a relative basis, but must be paired with other risk management techniques, which are both quantitative and qualitative. I have researched and compared VAR results under eight common techniques and, given the same portfolio, the VAR displayed different results. What this means is comparing one VAR to another can produce dramatically different views of risk. It is a valuable technique. As a sole measure of risk, no it's too dangerous. If it is tiered with other items stress tests, checks-and- balances with appropriate guidelines, and management oversight then I believe it can be a very valuable tool. People also perceive VAR as an absolute measure of risk. But it's greatest power is that it is a relative measure of risk. Looking from period-to- period and using the same technique, you should ask how does my risk change?
Till M. Guildimann Executive vice president,Infinity Financial Technology Inc., Mountain View, Calif. I advise businesses to use multiple methodologies only one will give you the view of the problem from one area. Since risk is a multidimensional problem, it is good to look at it from several areas. Value-at-risk of the parametric type, as described in RiskMetrics, it is a simple way to look at it. It is nice to have one number. Then if the position includes options, to calculate the risk more accurately, you will need more variables. You need a distribution of potential variables like the Monte Carlo simulation. Whenever there are options, you need a simulation. The number of simulations depends on the number of variables that you use. On the complexity of your position generally 10,000 or 50,000 variables are normal it depends on the problem. There are limitations. The first is that the number may be skewed because the positions entered are incorrect, and this was what created the biggest accidents Daiwa Bank, Barings, etc. The second problem is that all these systems rely on historical data. When you have surprising, new market developments, then these systems will give you wrong answers.
