Skipping SafeguardsBy Sean Gallagher | Posted 2002-03-12 Email Print
Allfirst Financial ran its currency trading desk without much of the software that was standard at its corporate parent, Allied Irish Banks. Was that a $690 million mistake?
In theory, the segregation could have been achieved in software AIB already had. In 1998, AIB began to install a software suite from Misys, based in the United Kingdom, at offices in Dublin, London and New York. In the back office, Misys' Opics program centralized many functions, including the tracking of foreign exchange transactions. At the trading desk, AIB installed Tropics, a software package for capturing currency trades electronically.
These two products can be directly integrated, providing what in the trading world is called "straight-through processing"computerized handling that allows trades to be fulfilled on the same day they are authorized.
"It's a fairly durable system," says Debbie Williams, co-founder and research director at Boston-based Meridien Research. "[Opics] is a pretty typical trading system, and very secure if you're using their front end."
According to an article in International Banking Systems in November 2000, AIB was one of the first banks to roll out Opics. And the bank planned to deploy Opics at its subsidiary, Allfirst. But apparently, it didn't install the front-office Tropics software there. Many processes, such as matching trades, continued to be handled manually.
Misys declined to discuss the case. "We don't have anything to add to this story," says Jeff Clark, Misys' U.S. marketing manager, from his office in White Plains, N.Y. Sean Fitzgerald, managing director of foreign exchange activities at Allfirst, refused to comment on the bank's trading systems. "I'm not at liberty to discuss that," he says.
Nonetheless, such software, if in place, could have prevented transactions that didn't have bona fide sellers and buyers. Without it, trades could be faked at the trading desk and not caught on the back end. "If trade information was hidden, or if bogus trades were entered, it would have most likely happened at the front-office end," says Harvey.
Good controls should prevent excessive trading. One function of front-office software is to generate alerts when a trading credit limit is exceeded. Warnings can be sent automatically, by e-mail or page, to the appropriate manager.
The back-office software also generates formal reports on such exceptions. So, even if managers at Allfirst weren't immediately aware of trades by Rusnak that went over his $2.5 million trading limit, they should have become aware of them when reviewing daily and weekly trading reports. Apparently, experts say, they either never noticed the patterns or simply ignored them because they trusted Rusnak's judgment.
In addition, Allfirst should have had in place a system that ensures all trades are between valid parties. But another element used by AIB's larger currency trading operations that was not used at Allfirst was an electronic trade confirmation management system.
When a trade is made, the two parties involved are supposed to exchange confirmations, to legally acknowledge the transaction. For a trade to be valid, there must be matching confirmations. Parent AIB uses the Crossmar Matching Service, an automated matching service that can confirm trades within two minutes.
Allfirst didn't use Crossmar. According to industry experts familiar with Allfirst, it handled most confirmations by fax. And in manual confirmations, it's not unusual for confirmations to go unmatched for some time.
"Bogus trades would have lacked confirmations from counter-parties," says Capco's Harvey. Without active matching of confirmations, Rusnak would have been able to enter nonexistent trades for currency options that would have countered the effect of his bad trades, essentially reducing the apparent risk to zero from the point of view of a risk manager.
The types of trades Rusnak was engaged in required even more oversight than straight currency trades. AIB confirmed that some of the trades were "forwards," or option trades based on the idea that currency rates will swing into one's favor at a future date, when the trade will be completed.