The Gateway to Algorithmic and Automated Trading

Look in VaR!

Published in Automated Trader Magazine Issue 19 Q4 2010

Sometimes, the old tools can be usefully applied in new ways. Here, Xavier Bellouard, co-founder of Quartet FS, argues for an IT-driven reappraisal of the value of VaR in forward-oriented risk modelling.

Xavier Bellouard

Xavier Bellouard

Let's just remind ourselves of where we are, shall we? The credit crisis has had huge ramifications worldwide. It's not just that the global economy simply cannot afford another failure on the scale of Lehman, but also, it's plainly clear that large banks and financial institutions cannot continue to make fat, risky bets and expect a government bailout. It is also apparent that banks and investments firms will have to answer to tighter regulatory controls and be required to report on their exposures across asset classes and geographies.

If the plain fact that there can't be a 'next time' has focused minds wonderfully on the importance of effective risk management, it is also true that the enormous cost of the banking bailouts, estimated at some £850bn1, has left the reputation of risk modelling in tatters. Financial risk management has long been a concern of regulators and financial executives, but the recent financial meltdown has not only placed even greater emphasis on its importance, but also, ultimately, left risk managers questioning the very essence of risk modelling. Banks and regulators have begun to question whether current methods can be relied upon to ensure there is not a repeat of the economic chaos witnessed in 2008 and 2009.

So how do we evolve from here?

Traditionally, investment firms have relied on fantastically complex mathematical models for measuring the associated risk in their various portfolios, primarily to reassure investors that all is well. While investors use a variety of mathematical models, the most widely used is Value at Risk (VaR), which is heralded for its ability to express market risk as a single number. Built around statistical ideas and probability theories that have been around for centuries, VaR was developed and popularised in the early 1990s by a handful of scientists and mathematicians. It describes the probability of losing more than a given amount of assets, based on a current portfolio. Alongside its ability to express risk as a single number, it is the only commonly used risk measure that can also be applied to just about any asset class.

Pre-crisis, this measurement allowed banks and investment firms to make quick and simple predictions on the potential losses of trading. However, more recently, VaR has been criticised for underestimating the tail risk. While its application has been extended in many ways to reflect liquidity risk and take into account operational risk and basic stop losses, it is still backward-looking and can fall short if there is an extreme change in price.

That said, it is important to remember that the crisis was not a result of market risk alone, and that therefore, VaR is still an extremely useful tool that can be used in conjunction with other risk management approaches, for example in evaluating worst-case scenarios. Furthermore, many argue that the recent financial crisis was ultimately a crisis of the modern metrics-based approach to risk management. It follows from this that, while undoubtedly risk modelling - as we know it - needs to evolve to cope better with market risk, the truth is that it is more banks' and traders' approach to, and use of, VaR that needs to change, rather than the actual tool itself.

Using VaR today

Historically, investment firms were using VaR primarily as a reporting tool to keep the regulators and shareholders happy. They were providing 'after the fact' analysis. Today, however, VaR must be viewed as an operational, rather than a reporting, metric. First, VaR needs to be broken down and analysed by traders. Instead of relying on a single number, traders need to look beyond the top line, delve into the complex mathematical calculations, and gain a better understanding of the type of risk they're taking and how it can best be mitigated. If they do this, VaR will become a valuable management tool, alongside other useful reference points such as the Profit and Loss sheet.

Secondly, as well as better analysis, traders need to receive VaR calculations in a timely manner. When used simply as a reporting tool, receiving VaR calculations within 24 or 48 hours is adequate. However, if traders are to have the ability to act on the information provided within the VaR calculation, they require the information much more quickly. While real-time VaR might not be necessary, getting the number within the trading day is crucial. A company's ability to understand its risk position in near enough real-time can allow it to take on more and/or greater positions and trade across complex products with more confidence.

Of course, many of these considerations are currently being mandated by the regulators and the emerging regulations will not only require a change of culture but also a review of banks' systems. As it stands, many financial institutions simply do not have the right systems in place to deliver in-depth VaR analysis in a timely fashion.

This is particularly true when you consider Marginal VaR. Marginal VaR is important as it analyses the impact - in terms of risk - of a particular asset class or country on the business. The Marginal VaR of a position with respect to a portfolio can be thought of as the amount of risk that the position is adding to the portfolio. Calculating Marginal VaR is enormously beneficial as it allows traders to understand where the largest risk is sitting within the business. However, at present few traders use it because they lack the tools required to compute Marginal VaR rapidly.

But where are we now?

The general lack of sophisticated technology within banks sits at the heart of the risk management problem. Broadly speaking, banks are no better prepared now for another financial meltdown than they were prior to the recent recession. This is not necessarily down to complacency but more due to the fact that financial institutions have not had long enough to implement new risk management methods and make the necessary changes to their IT infrastructure. Faced with pressure from the regulators, IT projects are under way, but the overall results and improvements are still a few years off.

While Wall Street has a number of plans to improve risk modelling, in the UK, the Financial Services Authority (FSA) has been the first to take action, issuing a record number of fines in 2009/10 according to the city

law firm Reynolds Porter Chamberlain. This was an increase of 21 per cent on the levels of the previous year and the report states that the regulator handed out penalties totalling £33.1 million over the course of the year, with eight fines of more than £1 million being levied.

The FSA has also strengthened its stress testing regime by requiring firms to improve their stress testing capability, enhance their capital-planning stress-testing, and introduce a reverse stress-testing requirement. The Basel committee is also modifying its Internal Model Approach to assessing regulatory capital for market risk by introducing a "Stressed VaR" charge2, and an increasing proportion of financial institutions are supplementing the use of VaR results with stress tests when reviewing their allocation of economic capital.

And where are we going?

The financial crisis has undoubtedly acted as somewhat of a wake-up call, exposing the poor quality of internal systems and revealing that it was impossible for banks and others to build a complete global picture of risk, including exposure, currency and counterparties, across their portfolios using the current systems. In 2010 the regulatory changes from the FSA and Basel Committee will require financial institutions to invest further in IT that can support the more demanding requirements. Once in place these solutions can be applied to intra-day sensitivities and P&L reporting too.

Let's end with two predictions. First, as investment firms get to grips with the new regulatory requirements, it's clear that a technology refresh will be on the cards, and as a result we will see banks and financial institutions beginning to install more risk-mitigation technology. Secondly, as we look beyond the recession and into recovery, many institutions will start trading more and more high-risk products again.

My final point is that the widespread institutional reliance on VaR is only a gamble if traders do not have the right technology solutions in place to help them analyse and break down VaR in near real-time. One can only speculate on the regulation and legislation that lies ahead but one thing that is certain is that it will be increasingly difficult to rely on homegrown systems that can't easily accommodate change, whatever form that change may take.