The first seeds of a new failure of risk management?

The history of financial and economic crises has been one of repetition. Booms lead to manias. A herd-like rush into assets commences and risk-aversion dissipates, toppling prudent risk management in the process. An asset bubble swells. A shock, often higher interest rates, causes the bubble to collapse. The fallout of that collapse spreads into the broader financial system and macroeconomy. The contagion can spread rapidly across international borders. Later, the lessons from the “go go days” of the mania and the extraordinary breakdown in risk management that highlighted those halcyon days where it really seemed possible for people to get rich quick with little effort are forgotten.

Already, in what could be a disturbing early indication that forgetfulness is already emerging or perhaps that not all financial firms have learned from the cascading destruction of both risk management and investment strategies built on quantitative models, Reuters reported that “quants” remain in demand. Reuters revealed:

But far from going into decline, those with financial engineering degrees are still in demand as hedge funds and banks seek ways to measure previously unforeseen risks and factor them into their models…

Some institutional investors have said quants become too enamored of their creations to notice when they turn into mathematical Frankensteins, especially in new, untested markets such as securities based on bundled mortgages…

Nassem Taleb, a former trader who wrote the best seller "Black Swan: The Impact of the Highly Improbable," is even more outspoken. "Quants and quant programs are dangerous to society," he said…

Taleb is not alone. Nobel laureate Joseph Stiglitz explained, “The important role of the financial sector in the economy—especially when financial institutions face problems—cannot be well encapsulated within the standard models…” Back in 1999, Henry Kaufman, one of the leading private-sector economists and investment strategists in the U.S., addressed the role of quantitative models with respect to the collapse of the Long-Term Capital Management (LTCM) hedge fund. He observed:

…mathematics and the computer technology that permits financial market participants to exploit its power has a dark side. Until the events of last year, a strongly held belief existed that financial risks are knowable, can be calculated with mathematical precision by massaging historical data, and can be diversified. These were always fallacies, but it took the near-collapse of LTCM, perhaps the most storied user of mathematical model-based investing, to prove the point.

Regrettably, the lessons inherent in the failure and liquidation of LTCM were forgotten, as often happens during the passage of time. Already, even as the current post-housing bubble crisis continues, it appears that some firms are ignoring the lessons that quants are no miracle workers when it comes to risk management.

In my opinion, quants can play a constructive role in dealing with specific, narrow and highly-structured problems. However, when it comes to largely ambiguous or unstructured risk where causal relationships are unclear they are literally useless. Human behavior, particularly when highly complex, sometimes opaque, instruments are involved is not servant to mathematical equations. It is largely that unstructured risk that is at the heart of the manias that fuel asset bubbles and then the contagion that results when such bubbles invariably collapse.

Precisely on account of those realities, modeled relationships typically breakdown as market volatility increases. Mathematical models cannot make risk vanish. On the contrary, they likely amplified risk by breeding a sense of complacency. The history of finance is littered with the carcasses of hedge funds and financial institutions that lived and then died from having placed their faith in quantitative models to part the proverbial sea of economic and financial risk. The present situation is par for the course.

Given the nature of the risk involved, the renewed interest in “quants” as a solution to the breakdown in risk and portfolio management that occurred during the ongoing economic and financial system crisis is little more than a repetition of a failed strategy founded on the assumption that “this time will be different.” That interest, especially if it translates into renewed reliance on quantitative models could well be planting the seeds for a future crisis.

Unfortunately, such interest represents a persistent quest in the human experience for a holy grail that assures success and vanquishes risk. In that context, the continuing naïve and exaggerated faith in the capabilities of quants and power of quantitative tools is not at all surprising. Back in 1998, The New York Times reported, “Despite such setbacks [as the failure of LTCM], some experts in the field still contend that it is only a matter of time before computers running mathematical models replace human judgment as the dominant force in stock picking.” The article might as well have added risk management.

In reality, such models offer some guidance and some insight into risk. They do not provide the whole answer. They do not provide even the largest part of the solution. Historic experience, which is both qualitative and quantitative in nature, provides a much better understanding of risk and, therefore, insights into effective risk management.

Historic experience from the Panic of 1873, Panic of 1907, Great Depression, and S&L crisis of the late 1980s and early 1990s argued that financial institutions that concentrated their loan portfolios in mortgages and mortgage-related securities were particularly vulnerable to risk. Historic experience warned that banks that had asset portfolios that were largely long-term in nature and liabilities that were mainly short-term in nature or vice versa were disproportionately exposed to risk.

Ignoring history has a high price. To date, that historic experience has only been reinforced in devastating fashion by the present economic and financial crisis. The hedge funds and financial institutions that are seeking to more effectively manage risk would do far better to focus on recruiting graduates who have a firm historic understanding of past financial crises than turning anew to quants.

The biggest thing lacking in the financial sector is someone who steps back from the numbers and models, and applies a dose of reality and long term thinking to the situation. The subprime crisis was based around a system that worked really well in the short term, and certainly looked great on revenue projections. However, nobody bothered to think about how it would play out in reality. In the real world, people who buy a really cheap rate mortgage are unlikely to be able to pay a sudden jump in rates a few years down the line. A teaser rate is a an established business practice in the industry, but only to cover the costs of a new homeowner. Expecting the income of the owner to magically increase a few years down the line was simply faulty. The people in the industry aren't stupid, but are too blinded by shorter term thinking and spreadsheets. A computer is an excellent tool for optimizing productivity in a specific area, but is completely unable to handle external variables it hasn't been programmed for. I hope that people will come the realize that computers are good for short term picks with their superior rationalizing capability, but in the long term cannot understand social trends that ultimately decide the market.

Expanding on the debate over excessive reliance on quantitative models and their limitations, the Turner Review, a British effort that examined mainly UK's regulatory response to global financial crisis has some information that addresses the broader issue of quantitative modeling. In fact, the report is quite harsh in arguing that rather than helping improve risk management, some of the models, including the almost universally-applied VAR (Value-At-Risk) model undermined effective risk management and increased firms' risk.

The Review's report, which was published earlier this month, states:

The very complexity of the mathematics used to measure and manage risk...made it increasingly difficult for top management and boards to assess and exercise judgment over the risks being taken. Mathematical sophistication ended up not containing risk, but providing false assurance that other prima facie indicators of increasing risk (e.g., rapid credit extension and balance sheet growth) could be safely ignored.

With respect to VAR, in what could represent a paradigm shift with respect to risk measurement, the report declared, "There are...fundamental questions about the validity of VAR as a measure of risk." The report then went on to mention specific defects in VAR including:

• Short-term observation periods (or an insufficient sample) along with the assumption of a normal distribution (a practice Nassim Taleb attacked in his The Black Swan) underestimated the actual probability of extreme loss events.

• VAR does not capture developments that result from market linkages.

• VAR assumes that each firm's actions are segregated to itself and do not have broader market impacts that feed back on the firms.

• Systemic risk exposure "may be highest" when risk measured by VAR "is lowest, since low measured risk encourages behavior which creates increased systemic risks."

With respect to the ongoing financial system crisis, the report observed:

From about 2003 onwards, there were significant increases in the measured on balance sheet leverage of many commercial and investment banks, driven in some cases by dramatic increases in gross assets and derivative positions... This was despite the fact that 'risk adjusted' measures of leverage (e.g., weighted risk assets divided by tier one capital, or Value at Risk (VAR) relative to equity) showed no such rise... It is clear in retrospect that the VAR measures of risk were faulty and that rquired trading book capital was inadequate.

The Turner Review is just the latest finding to weigh in on, among other matters, the dramatic breakdown in risk management that took place and the contribution overreliance on modeling made toward that outcome. In the future, a rigorous approach that combines stricter capital requirements, ties derivative positions to underlying assets (rather than naked exposure), maturity matching, accounting rules that limit the ability of financial firms to create large off-balance sheet vehicles, and a balance between qualitative and quantitative risk assessments will probably offer a vastly improved approach to risk management than the preceding regime that disintegrated as market stresses began to increase.