A V Rajwade: Proprietary trading risks

WORLD MONEY

Till the recent crisis, most focussed on only the rewards.

In an article of the same title (Business Standard, June 8, 2007), I had analysed the trading profits of some of the major commercial and investment banks, in relation to the average value at risk they had disclosed in their 2006 accounts. While I had to make some robust assumptions in calculating the expenses, I had concluded that:

The net trading income after expenses amounted to several times the capital charge for market risk (in other words, a return on equity of a few hundred percent!); and

Given the level of profits, was the risk was being underestimated.

Recent events in the credit derivatives market and the substantial losses many major banks have reported clearly evidenced the underestimation of the risk. As a report in Bloomberg analysed, there is also little correlation between the disclosed value at risk and the losses sustained. For example, Merrill Lynch, which reported the second lowest VaR amongst the five major US investment banks (Goldman, Morgan Stanley, Merrill Lynch, Lehman Brothers, Bear Stearns) reported a $ 10 billion loss in Q4 2007, the largest in its century-old history "" and way beyond the reported VaR.

Subsequent reports about the huge loss sustained by Societe Generale (SocGen) further evidence the weaknesses of the traditional VaR methodology in measuring risks. Nassim Taleb (The Black Swan) and Benoit B Mandelbrot (The (Mis) Behaviour of Markets: A Fractal View of Risk, Ruin And Reward) have long argued that the normal distribution methodology used by most banks and blessed by regulators, to measure market risk, underestimates the frequency and scale of extreme events, that the outcome distributions have far fatter tails.

As is well-known by now, in a move that surprised the market, SocGen reported a trading loss of almost Euro 5 billion, in its equity derivatives portfolio (besides, a loss of Euro 2 billion in its portfolio of credit derivatives). The surprise was all the more because SocGen has pioneered some of the most complex instruments in international financial markets, and boasted of a very sophisticated risk management system.

The entire loss has been blamed on the activities of a single trader who was supposed to be doing arbitrage trading (a la Nick Leeson of Barings): He would go long in a portfolio of (underpriced) financial instruments, at the same time selling another portfolio of (overpriced) instruments with similar characteristics, hoping to make some profit on the price differences. To be sure, this was not arbitrage in the strictest sense, since obviously, a basis risk was being taken. In fact, it seems that, since 2005, the trader had been taking naked, directional bets (once again exactly what Leeson did), and reporting fictitious trades on the opposite side so as to keep his total portfolio within the risk limits. (Last year, even Eurex had questioned the size of some of his trades.) He would reverse the fictitious trades before confirmations went out, and enter in fresh bogus deals in replacement, using his colleagues' passwords, and the knowledge he had gained of the systems when he was working in the bank's mid-office. When the fraud was discovered on the weekend of January 19/20, the loss was of the order of Euro 1.5 billion on a portfolio of Euro 50 billion. The top management, audit committee and board, which met on Sunday, decided to unwind the trades before disclosing the loss. Given that the equity market was already jittery and falling, the unwinding, spread over three days, increased the loss to Euro 5 billion, as the liquidation of the bank's positions further weakened the equity prices. It was only after the liquidation was complete that the bank disclosed the loss. The only sure thing is that it has reported the largest-ever trading loss sustained in financial markets. Incidentally, in which business can one relatively junior employee lose so much money for his employer?

For students of risk management, there are several questions and issues that are worth thinking about:

How many times are unauthorised, over-the-limit trades done, but not reported as they get unwound at a profit or without undue losses?

How was the cash used in the trader's "arbitrage" activity monitored?

Should gross value of trades be monitored?

But this apart, if SocGen management, with all its sophisticated risk management systems, could not detect what the trader was doing for three years, is it any surprise that Indian managements, facing huge MTM losses on complex derivatives, also remained unaware of what was happening for a long time?

avrajwade@gmail.com