Model uncertainty

The author makes the valuable contribution of making us aware of black swans that could be lurking in the financial market.

NASSIM NICHOLAS TALEBS book (The Black Swan: The Impact of the Highly Improbable), analysing financial systems, has become a bestseller. In the process, it has also become a must-read for financial analysts of Wall Street. Taleb, who hails from a highly respectable Lebanese Christian family and whose grandparents and great-great grandparents were high government officials in various regimes, saw his entire family fortune vanish practically overnig ht in the Lebanese civil war of 1975. One suspects that memories of this trauma have influenced his views.

What is a black swan? If you had lived all your life in the northern hemisphere, all you would have seen were white swans. You would conclude that all swans are white. This is exactly what the Europeans of the 18th century concluded before they went to Australia where they found only black swans. Thus, the metaphor of the black swan is used to denote anything that appears impossible on the basis of a limited number of observations.

In modern statistical parlance, it would be called a model uncertainty. More precisely, it is what happens when we use a model to predict some future events but the underlying model is wrong and we do not even know that the model is wrong. To paraphrase Mark Twain: Its what you dont know you dont know that gets you into trouble.

In January 1995, the Barings Bank lost 827 million ($1.4 billion), twice the banks available trading capital. It went bankrupt in February that year. The architect of that failure was a rogue trader named Nick Leeson. Since he operated as a trader in the front office and was the person who processed the trade in the back office, he was able to hide his activities from his superiors. At first, he took several uncovered positions and lost a few million pounds. Then he took bigger bets to recover the losses.

However, the Kobe earthquake of January 1995 produced a figurative earthquake in the Japanese futures market that amplified the loss to the billion-dollar range. This was an unexpected event in the market. Nobody expected Kobe to be so badly damaged, given that all possible measures had been taken to ensure that the city would be able to withstand such an earthquake. And nobody expected that an earthquake in Kobe could ruin a British bank that had been in continuous operation since 1762. This was a black swan.

Most large banks take measures to stop trading that allows both the front and the back office to be controlled by the same person. This is supposed to be a standard risk management practice. Societe Generale, Frances second largest bank, took such practices seriously. In fact, the magazine Risk declared early this year that among the large banks in the world, Societe Generale had the best risk management process in place. Yet, this did not stop Jerome Kerviel, a trader at Societe Generale, from taking an unhedged unauthorised bet in the European futures markets. Once again, it turns out that he had control of both the back office and the front office. This time, the undoing came about when the Federal Reserve of the United States unexpectedly dropped its official interest rate in late January. In the end, Kerviel caused a 4.9-billion ($7.2-billion) trading loss to Societe Generale, which could spell its end.

These are examples of operational risk gone spectacularly wrong. They are black swans. Where are these black swans in our lives? According to Taleb, they are everywhere. In particular, there are black swans in financial markets. The standard statistical models (the bell-shaped curve of the normal, or Gaussian, distribution) used in standard financial models underestimate the chances of high-risk events.

Events that actually occur once every decade are often modelled as if they occur once in 100 years. This creates the false impression that we can ignore events that should not be ignored. Black swans are real. They are even more real in developing countries such as India where the financial market is inherently much more volatile than its counterparts in the developed world.

Indians have two bad habits. First, we consider Western-trained Indians to be superior to home-grown ones. Western-trained financial experts acquire a Western mindset. They tend to ignore problems that typically arise in markets that are thin most of the companies listed in Indian stock exchanges hardly ever have any trade. Thus, their training does not prepare them to spot black swans. They are like the 18th century English who had never seen a black swan.

Second, we in India believe that the regulations of developed markets are worth imitating. As a consequence, we tend to adopt the regulatory environment of the Western markets, lock, stock and barrel. If we listen to the policymakers in the government, in the Reserve Bank of India (RBI) or in the Ministry of Finance, their long-term goal appears clear. They want to mimic the regulation adopted by the OECD (Organisation for Economic Co-operation and Development) countries. Banks are considered to be progressive if they follow so-called Basel II compliant regulations.

Is the empirical evidence from developed markets relevant to India? Events such as the Asian meltdown of 1997 are far more frequent in India than they are in the developed world. If we use American or British data to assess the operational risk in India, we would be ignoring risks without knowing that we have ignored them. In short, we would miss the black swans.

What is the way forward? The whole idea of Basel II is to make realistic assessments of risk exposures backed by data and to use proper statistical techniques to build appropriate models. To this end, the RBI could constitute a high-powered committee that could lay down the norms for the assessments of advanced statistical model-building techniques. Regulatory norms for risk management are in a nascent stage in India at present. The RBI could insist that only evidence from India would be acceptable. It could issue a clear policy statement on this matter now so that banks would encourage studies in the Indian markets. Then the black swans of the future will not be missed.

There are already worrying signs that we might be heading down the wrong path. Large credit rating agencies in India have tied up with international ones from the developed world. Many models used by these international agencies are proprietary and, therefore, black boxes by definition we do not know what goes on in them.

While Taleb makes the valuable contribution of making us aware of black swans, he rants against economists, finance professionals and academic statisticians. He argues that they have entrenched the practice of ignoring black swans. By this accusation, he ignores historical fact. In fact, many of the criticisms he makes first came from economists. For instance, when Fischer Black, Myron Scholes and Robert C. Merton laid out their model for finding closed-form solutions for option pricing problems, they assumed the normal distribution but warned that the solutions were very special cases and, thus, were not to be taken as gospel. It is hard to fault them if the industry simply ignored their comments. Even Gauss, several centuries ago, knew that there was life beyond the Gaussian distribution.

On page 261, Taleb asserts that since 1963, nothing has happened in economics and social science statistics except for some cosmetic fiddling. This is egoistic nonsense. In fact, some of the studies in economics that he cites were written after 1963. Taleb also makes many flippant comments that put a jarring note on his otherwise splendid exposition. For instance, on page 196, he notes with glee: A single butterfly flapping its wings in New Delhi may be the certain cause of a hurricane in North Carolina. Such assertions are intriguing, but there is not a shred of evidence to prove them. Unless, of course, we could interpret an earthquake in Kobe causing a bank failure in England this way.

Tapen Sinha is the ING Chair Professor of Risk Management at Instituto Tecnologico Autonomo de Mexico with a concurrent appointment as Professor of Risk Management at the University of Nottingham, U.K., and a consultant to Cranes Software International Limited, Bangalore.