Control lessons from the Societe Generale fraud: what patterns emerge from rogue trading incidents of the recent past?
The recent trading losses of roughly $7.2 billion at Societe Generale, attributed to the rogue actions of junior trader Jerome Kerviel, set a new record for losses in rogue trading incidents. The most prominent of these in the recent past have been Kidder Peabody (1994), Barings (1995), Allied Irish Bank (2002) and National Australia Bank (2003). As each new incident occurs, we in the risk management/ financial controls community ask ourselves: What new lessons can we draw from this control failure? In the case of the Societe Generale incident, I will argue that, from one point of view, the answer is: Not much--Kerviel's methods for eluding scrutiny of his positions were very close to those used in previous incidents. But, from another viewpoint, we can learn quite a bit, since clear patterns are emerging when we look across episodes.
Before proceeding to an analysis of this and similar events, let me make clear that I am treating as noncomparable some other incidents of very large trading losses, such as the recent subprime-mortgage-related losses of Bear Stearns, Merrill Lynch, Citigroup and UBS and the 1998 collapse of LTCM. What the incidents I am reviewing have in common is that they involve deliberate falsification of trading records to hide the size of real positions from supervision internal to the firm. I am excluding cases in which size of positions were known within the firm, though they may have been unclear to parties outside the firm such as lenders, investors and regulators, and the true riskiness of the positions may have been misevaluated both within and outside the firm. These are very different situations requiring different remedies.
Going back 30 years or so, the oral tradition within control functions was to worry about position falsification through the technique known as "tickets in the drawer," a trader simply not entering some of his trades onto the firm's books and records. (Since all of the frauds discussed here have, to the shame of my gender, been perpetrated by males, I will use masculine pronouns.) This approach seems to be lacking in any of these recent incidents, presumably because the possibility of the fraud being exposed through an inquiry from a counterparty to an unrecorded trade is too great. What seems to have replaced it is the entry of fictitious trades designed either to offset the risk position of actual trades, making the net risk look small, or to create bogus profit and loss (P&L) to disguise actual earnings. Since fictitious trades lack a real counterparty, they cannot be exposed through action of a counterparty but can only be uncovered by internal controls. The creator of fraud is in an ongoing battle with control personnel--the control personnel have the advantage that uncovering only one clear-cut case of falsification is enough to uncover the fraud; the advantage of the creator of the fraud is the multiplicity of methods he can employ to discourage this discovery.
I divide this analysis into three parts: (1) the details of Kerviel's...