Paribas, the French banking giant, has begun a pilot of a system to measure "operational risk," considered the most difficult of all risks to quantify.
Operational risk includes incidents of fraud, reputation damage, technology failure, and natural disaster.
Paribas is testing OpWatch from Toronto-based Algorithmics Inc. The system is meant to detect operational weaknesses in the bank's daily flow of cash bonds. The banking company, formerly known as Banque Paribas, has deployed it in the fixed-income and equity operations of its London office.
OpWatch works to identify characteristics of past trades associated with losses. Using those characteristics, it calculates the probability of losses on those trades throughout their life cyle. The aim is to provide early warnings on similar trades as they enter Paribas' straight-through- processing systems.
Other banks have deployed operational risk systems that use external data to benchmark trades. Paribas officials said they thought that approach would be costly for data maintenance.
"We want a heads-up warning" of operational loss events, said Timothy J. Kent-Phillips, head of global operational risk control, fixed income, and equity business at Paribas.
Such risks are hard to measure, but the cost of failing to spot them can be great. For example, rogue trading brought down Baring Brothers & Co., London's oldest merchant bank, and led U.S. regulators to bar Japan's Daiwa Bank from operating in this country.
Regulatory changes are among the forces that have led banks to focus on operational risk. The Basel Committee of the Bank for International Settlements has said it plans to examine operational-risk-associated capital. With approved operational risk methodologies, banks might lower capital reserves and thus increase shareholder value.
For Paribas, operational failures are unacceptable, Mr. Kent-Phillips said. He noted that clients screening preferred broker-dealer candidates are increasingly likely to consider operational performance measures such as accurate and timely trade confirmations and settlements.
Paribas hopes eventually to allocate capital against operational risk in response to evolving regulations.
Other goals include integrating the operational risk system into new client services and using it as a "core" tool to develop a real-time picture of the bank's entire trading operation.
Paribas chose Algorithmics' system because of its ability to address different types of operational losses, such as process-related failures or rare events, Mr. Kent-Phillips said. OpWatch also supports different user groups.
"We wanted a tool to sit on our desktops that we would use day to day- not something that produces reports that get filed on shelves," Mr. Kent- Phillips said.
Competing systems include Orca from Operational Risk Inc., a new Bankers Trust spinoff; and RiskOps from Greenwich, Conn.-based NetRisk, also founded by a Bankers Trust alumnus, Eugene B. Shanks.
Both of those systems use insurance-based actuarial techniques that take advantage of publicly available loss data. But OpWatch, which Paribas is piloting, is based on causal models used in computer sciences and statistics, said Jack L. King, director of the vendor's operational risk division.