A V Rajwade: Internal control of derivatives

WORLD MONEY/ The National Australia Bank fraud has many lessons for bank treasuries

The National Australia Bank (NAB) is the largest commercial bank in that country. The current year did not begin too happily for the bank; it had to report a loss of 180 million Australian dollars (A$) (that is, $126 million), discovered in its options portfolio.

The management expressed a "high level of confidence" that the amount represented the upper limit on the losses that were being investigated. As it happened, the investigation disclosed that the actual loss was double the initial amount "" at A$360 million.

Reports on the case by PricewaterhouseCoopers (PwC) and the Australian Prudential Regulatory Agency (APRA) came out in March 2004, and are available on NAB's website.

The reports are a rich learning source for students of derivatives and risk management, and will also be found useful by bank treasuries and management executives concerned with internal control of derivatives operations. The following discussion will hopefully vet their appetite to study the original reports.

It seems that the manipulation of books and records to hide losses and report profits started in 2000-01. At the end of that year the value of the options book was overstated by A$4 million; the amount doubled in the next year.

With subsequent profits, the cumulative over-statement had vanished by October 2002 (the bank's fiscal year is October-September) and amounted to a relatively modest A$5.5 million by the end of August 2003.

There was, however, a major loss in September and the overstatement at the end of fiscal 2002-03 was $42 million. Losses (and false entries to hide them) continued on a much bigger scale for the next few months, taking the portfolio overvaluation to A$360 million by the time the fraud was detected.

Before looking at the culture and systems that permitted losses on such a scale to be hidden, one point is worth pondering over. In none of the cases involving losses in bank treasuries that have come to light in recent years (Barings, Daiwa, Allied Irish Bank and now NAB) has any evidence come to light of the dealers profiting from the fraudulent activities.

The second interesting feature is that the fraudulent activity continued for years before it was detected. Again, in both Allied Irish Bank and NAB, the manipulation started with the objective of hiding genuine losses through reporting of false transactions.

In the case of NAB, there was also an attempt to smoothen the volatility in monthly profits by manipulating the numbers reported.

However, as losses continued, greater risks were taken in the hope of recouping them "" no doubt an illustration of the mathematical certainty that while betting over the toss of a coin, if you double the bet every time you lose, you will ultimately come out with a profit. (Incidentally, talking of betting, the practice of paying bonuses to dealers, depending upon the profit they have earned, is a one-way bet for the dealers "" if their risk-taking results into a profit, they share a part of it; if there are losses, all that the dealer may lose is his job.)

In the case of the bank, by the time the losses could no longer be hidden, it had lost hundreds of millions of dollars.

The four dealers involved in the NAB case took advantage of the weaknesses in the bank's internal control systems to enter (and reverse) false transactions to conceal losses.

This was often done by using fictitious exchange rates "" $0.0075 per Australian dollar is just one example. Such false transactions were entered just before the system calculated profits of the day's trading activity.

As soon as this procedure was completed, and before back-office verification of the deals started, the false deals were amended to reflect the correct exchange rates.

The hope was that profit in the next day's trading would more than compensate for the concealed loss, in which case, false deal entries may not need to be repeated.

Another stratagem used was the entry of one-sided internal deals (it seems that the IT system used in NAB did not provide for automatic generation of contra internal deals). This started from October 2003 when the back office stopped checking internal deals.

Even earlier, the back office considered that "they were merely responsible for ensuring that internal transactions were equal and opposite, not that they were valid."

Yet another tactic was to use incorrect market rates for revaluing the portfolio. The fraud was also facilitated by a weak risk-control culture in which multiple limit breaches were routinely permitted, transactions undertaken in products not approved for use and the value-at-risk (VaR) numbers produced by the system were known to be unreliable.

As PwC points out in its report, "If limit breaches were not being investigated properly and VaR could not be calculated reliably, then effective risk management of currency options was not possible and good business sense would require trading to be carefully managed, curtailed or ceased altogether."

This was not done until the loss reached a huge level. Another interesting observation made by PwC is that the risk-control systems and departments were "focused on process and production, with no one really attempting to understand the nature of the business being conducted."

Reliance on complex IT systems does lead to this phenomenon: the output coming from the "black box" gets taken for granted and the user of the output increasingly loses a "feel" for the numbers that come out. To my mind, the importance of a qualitative appreciation of the numbers should not be underestimated.

Email: avrco@vsnl.com