ORIENTATION AND APPLICATION OF VaR IN THE STRUCTURED DERIVATIVES MARKET

by Philippe Carrel, Global Head of Alternative Investment Strategies, New York, Reuters

Has VaR become obsolete?

The fast growth of structured financial instruments and multi-asset based derivatives has raised new issues for managing the market risks in spite of cross-asset correlation effects. Pricing, trading and processing cross-asset financial instruments with complex potential leverage effects is no longer possible on an industrial scale using silo systems linked one to another with interfaces and middleware. Similarly market risk views require more than adding up each of the risk components.

In addition to this, the entire concept of risk management is changing as the most recent trends lead banks to consider the entire exposure to clients as portfolios of credit and market risks. This deeply modifies the approach to correlations and risk mitigation effects.

Structured instruments

What are those new instruments which require special risk management? It all started in the mid-90s with yield enhancement deposits created especially to keep Japanese yen holders engaged despite plummeting yields. Bundling fx options with deposits added mild spicy optionality in the most widely used short term instruments and the innovation was well received by the market. The idea was quickly replicated to gold, stock indices, and later to credit events. A new concept was born -creating cross asset layered products wrapped around well known client exposures and appetite for risk. Since then, there are new structures each month. The concept evolved from multi-currency instruments such as power reverse duals toward more complex structures involving several asset classes such as credit linked notes, interest rates swaps with several layers of barrier options attached. There seem to be no limit to the creativity of the quants. This evolution is expected to change a great deal of the usual risk management and methodologies.

The basics of all VaR methodologies -including Monte-Carlo simulations- is to estimate the future volatility of the financial instruments individually, thereby measuring a standalone risk. Then the portfolio risk is derived by adding new estimations essentially based on correlations. It is somewhat striking to notice the dramatic effect a simple change in the assumptions regarding correlations can have on VaR of the portfolio.

The sophistication so far remained in modelling the distributions within the various portfolios so that they best match the actual distributions observed on the market, hence replicating natural correlations. It worked satisfactorily most of the time however, major failures and critical deviations from the expected VaR were recorded when unexpected correlations arose from unpredictable market sentiments and extreme trading conditions.

With the new generation of structured products, these type of failure should logically occur more frequently as the variety and complexity of conditional payments is clearly a potential source of errors. By definition, cross-asset correlation surprise effects are bound to multiply since events in an asset class can create exposure in another. Moreover the assumptions are multiplied. For example, some long term quanto swap would require pricing and evaluation methodologies to be strictly aligned as well as perfect consistency in the very order the risk factors are taken in consideration. The risks are combined by the systems but some cross-gamma effects due to correlations are not entirely hedgeable. Analysing a long history of correlations, if that were possible, would provide little reliability as there would be no guarantee that history would reproduce. Most importantly, these products are by nature short lived. The exposure they intend to hedge is usually specific and in most cases contingent to the evolution of several market factors combined together.

Arguably this type of problem only impacts the very complex and tiny share of the trading book of financial institutions. One may think that overall, the models and methodologies are still good enough for the bulk of the trades. This type of reasoning has been largely invalidated through each and every crisis that hit the financial markets. Not only unexpected correlations turn hedging tools into new risk exposures, but as it all typically happens in extreme market conditions they may trigger panic moves and liquidity holes on the vanilla markets as well for hedges need to be unwounded and underlying exposures need to be covered.

Portfolio approach to client exposure

The aftermath of Enron, Worldcom and other fallen angels is that most market players have grown totally risk averse. The new regulatory framework FAS133 helped clarifying the boundaries of acceptable risk taking. As a result, banks in search of innovative high margin products resorted to sell unquestionable hedges such as credit derivatives or complex multi-asset structures.

The recent interest for credit mitigation tools is increasing that momentum with series of synthetic securitisation being carried out through credit derivatives based structures. One may think that credit derivatives are only sensitive to measurements of the creditworthiness of the issuers, but those are themselves derived from market simulations since the Merton methodology now largely prevails. The method consists of estimating a probability of default as a put option on the asset of the company which strike would be its level of recoverable liabilities. The complexity increases with the new trend to price credit default swaps (CDS) on baskets rather than on a single issue.

In other words, there is an increasingly high market risk component in the estimation of the credit risk which is again primarily linked to assumptions of correlations. Moreover, the credit and market risk averse strategies and the subsequent full hedges create concentration effects that were usually known only in portfolio management.

High risks for risk averse markets?

Are we progressively building a time bomb through intangible market factors no one will be able to control the day they become visible? It is probably the case as it has always been with new financial instruments and new market trends. Each market crash, each major financial crisis has unveiled unexpected market factors and unpredictable effects. It is usually the result of high speculation in the context of a bullish market and the result of innovations in financial engineering that were initially designed to mitigate risk. The rule of hedging is yet simple: Risk never disappears. It is merely transferred to someone else. As the financial market are increasingly growing market and credit risk averse, most of those risks are actually transformed in operational risks.