In recent months, there has been an interesting twist in the story of VaR, the central measure of market risk, capital requirements, and one of the targets of criticism in the aftermath of the credit crisis. Not much comment has been made of it, but VaR has been adopted as a key plank in the central clearing of derivatives, precisely because of its ability to produce a meaningful single number. This is ironic given that it was this single number, and its attendant lack of insight into the deeper tailed risks, that led to much debate about VaR’s suitability as the key metric, and the role it played in risk management leading to the crisis. And coincidentally, it seems the story of VaR is paralleled well by the journey of young Anakin Skywalker in George Lucas’s prescient epic Star Wars.
To fully appreciate the turnaround, it’s worth looking briefly at the origin of Value-at-Risk.
Obviously complex metrics have their roots in many areas, and tend to be cross-fertilized evolutions rather than single-point revolutions, and VaR is no different. What can be said though is that the emergence of complex derivatives and the growth of those instruments in the late 80s and early 90s did pose a problem for institutions looking to capitalize adequately for adverse market movements. This is not unlike the issue faced by the Jedi Council as they searched for the “one to bring harmony to the Force” at a time of growing tension within the Galactic Republic.
The financial industry’s risk problem was twofold. The increasing complexity of the derivative books created risks that were hard to quantify under traditional small movement measures. Simultaneously, the siloed nature of financial firms led to “natural hedges” (or risk offsets) between trading positions being lost in the risk aggregation. The result was an interesting problem where the risks were either being (vastly) understated or (vastly) overstated.
The answer, VaR, appeared a perfect solution to this twin-horned dilemma. By determining the sensitivities of all of the traded instruments at a firm, to a range of the most commonly observed drivers (interest rates, volatility, FX rates), the problem could be reduced to then using those sensitivities in a range or distribution of scenarios, itself determined by a correlation process across the observable market data, and finding the worst results at required percentiles (i.e. the 99th worst result). This was uncannily similar to the discovery of the gifted pod racer, Anakin, whose natural powers would be molded by his master-to-be, Obi-Wan Kenobi so he could become the future Jedi hero.
Determining this “parametric VaR” involved many approximations and smoothing, but was a huge step forward in the search for a number that best expressed the risk position of an organization.
Improvements in the underlying mathematics continued, as second order movement in the underlying risk drivers were added, but it was the exponential increases in computing power that paved the way for improvements.
The drawbacks in the parametric approach became obvious, and are well documented, so the next step was to move to full revaluation of the underlying trades under the distribution of scenarios. Two main avenues were explored and adopted: Historic VaR and Monte Carlo VaR. Historic VaR uses actual histories to directly create the scenario distribution, while Monte Carlo uses statistical analysis and correlation data to generate a range of possible future scenarios. Portfolios are then valued under each of the scenarios and the worst cases at specific percentiles are again found and reported.
It is at this high point, VaR’s very own “Battle of Coruscant,” that things start to go awry.
The aim of VaR was always that the “minimum loss that should occur, at the frequency of the percentile selected.” What this means is that the VaR number, when working, should never be a maximum loss number, but was always the minimum loss that the portfolio would incur at a frequency determined by the selected percentile (once every 100 days for 99th percentile). What was never intended was that this should be seen as an indicator of the losses that could occur with less frequency, but be far more devastating.
The very neatness, scalability and simplicity of VaR – its ability to capture so much complex information into a number – led regulators to pick it up and embed it as the risk measure, rather than a key risk measure. Once it was linked so completely with the capitalization demanded by the regulators in most jurisdictions, with the only choices left being which of the three VaR methods to use, its centrality in any subsequent crisis was all but assured. It was truly on the path to the “dark side.”
Once the crisis hit, and pre-crisis risk management was scrutinized with the crystal clear lens of hindsight, VaR was found to be imperfect. Many pointed out that it did not deal well with extreme events, deep tail risks, emergent risks and specifically regime change risks. This was not news to those who had conceived and/or used VaR for a long time, echoing master Yoda’s initial and ongoing misgivings about Obi-Wan’s gifted protégé. It was simply restating certain limitations that had always been caveats of the metric. The interesting thing is that even given the storm around the calculation, no suitable replacement could be found, a situation that remains to date.
Of course, the recent crisis was rooted in credit and liquidity. One of the driving changes across the world has been to take individual institutions out of the systemic default loop via the mechanism of central clearing. This most importantly covers swaps and credit default swaps.
Central clearing relies on daily margin calls by the central clearing agencies, so it’s natural that some kind of measure would be required to cope with complexity while delivering a single number that could be used for this margin. After analysis and experimentation, the most appropriate number was found to be…Historic VaR.
Value-at-Risk, the same measure implicated as a contributor to the credit crisis, was called forth to be the central tenet of the solution to that crisis. This is possibly the greatest twist of fate since Lord Vader rose from his knees to cast the emperor into the deep chasm running through the second Death Star.
Different clearing houses use different driving parameters, but the method remains the same. To experienced risk folks, none of the above is new, and is in fact a ridiculously abbreviated history of such an important metric, but it does show the unintentional irony of industry genuflections and trends.
May the Force be with you…
While you’re here…