SocGen - its gross how things fall through the net

Ever told a little lie, but then found yourself in a spiral of more deceit as you seek to preserve or cover-up the initial one?

I find it hard to tell at this point whether the only person that was locked into such a spiral was SocGen "rogue" trader, Kerviel. This is a strong assertion, but each time a new pronoucement is made to add to the story, it becomes more incredulous. Yesterday, for instance, it was announced that:

Though he was only supposed to buy futures – bets on the direction of European markets – if they were covered by a hedge – a similar position limiting any loss – he used other people’s access codes and “falsified documents” to create fake hedges, leaving the bank exposed to the full downside.

SocGen said he had evaded detection for almost a year by only choosing “very specific operations with no cash movements or margin call and which did not require immediate confirmation” and by constantly switching between different types of instrument.

By January 18, when he was finally caught, he had positions worth €30bn on the Euro Stoxx, an index of Europe’s biggest companies, €18bn on Germany’s Dax and €2bn on the UK’s FTSE.

So the Bank is disclosing that Kerviel's job was to seek to make profits by carving out off-setting positions (hedge position) that limited the possible profit or loss that would arise from market movements. Such positions would be created from two assets that were assumed to be close, if not perfect, substitutes for each other. For example, one could buy futures to create a long exposure profiting from the market rising and buy put options that would allowing you to sell an equivalent exposure with a strike price below the current market price, termed "out of the money" options.

Such a structure might look like this
In risk terms, the Bank's maximum price exposure would be deemed to be the loss of 10 and it is this on which Kerviel might have been monitored. However, if the option contracts were fictitious then the Bank would be fully exposed to the market movements. The use of purchased options would be consistent with the assertion that no margin calls were involved, but it would normally involve payment of a premium and exchanging confirmations of the trade with a counterparty to independently verify the existence of the trade.

So, another instrument possibility might be a performance or total returns swap. These trades are stuck directly between two counterparties rather than on an exchange. They agree to pay each other the returns on nominated assets. For example a simple example is where one counterparty agrees to pay the interest on £5bn in exchange for receiving the "returns" on the market from the other. The respective payments on such swaps might only occur quarterly, with no initial outlay. More importantly, the confirmation of such trades can be flaky since they follow no standard process. This type of trade would certainly fit with the comments in the press reports.

Again, from a risk perspective, provided the notional value of the swap offset the futures positions, they might have considered Kerviel to have a "flat" position. That is not to say that other risk questions would have arisen such as the exposure to the counterparty and their credit worthiness.

Importantly, such deals are precisely the areas that SocGen excelled in and traders would have been encouraged to engineer both to satisfy client demand and create "risk free" trades that generate profit.

However, this doesn't overcome the challenge for Kerviel that every trade should be confirmed independently of the trader with the counterparty and controls should be in place to chase up outstanding confirmations. The reason for such controls is precisely to address the risk of incorrect or fraudulent transactions being booked. The absence of such confirmations will normally prompt investigations and reversal of uncollaborated transactions.

However, one important element often missing from risk environment is properly dealing with backdated or cancelled trades. Inside many firms, risk reports are produced daily reflecting what happened yesterday or positions at right now. As such, one possible vulnerability is that a trader cancels a historic trade booked 30 days ago and replaces it with another that is 10 days old. The risk team wouldn't care about this because their systems would be unlikely to flag up that this created a historic position, since the current position is fine thanks to the newer trade. However, a middle office function would be required to investigate a 30 day old trade that hadn't been confirmed would be investigated. But cancelling such a trade would almost certainly halt the investigation and the 10 day old trade wouldn't yet warrant attention until later. In such a scenario, everything hinges upon the controls over cancelling and rebooking trades. Unfortunately, human failings means booking errors can easily and regularly arise, leading to complacency, with the result that controls over rectification of such"errors" are usually the laxest.

Obviously, more pieces of the jigsaw will emerge over coming days and weeks that will improve our understanding of the complete picture, but as the last few posts may have illustrated, there are always weak spots in any impregnable fortress.

Labels: , ,

posted by John Wilson @ 9:12 AM Permanent Link newsvine reddit



0 Comments:

Post a Comment

<< Home