Broad uptake of FpML among banks, the near completion of the protocol’s product coverage and messaging framework and its role in current and future trade confirmation and matching services set the stage for using FpML for operational risk mitigation (see related story, page..)

The results of any FpML counterparty portfolio project are unlikely to be aired until the banks involved are comfortable with the results. In time, however, the initiative may yield a new market discipline aimed at quarterly or periodic portfolio checking between counterparties, said Brian Lynn, chief technology officer at Gem Soup and a member of the International Swaps and Derivatives Association FpML Standards Committee.

Without FpML, determining the comparability of counterparty trades ahead of trade effective dates would be an onerous task, as trades often sit on books for years, errors accumulate and internal modifications that do not require counterparty confirmation occur, such as those arising from trading desk and systems reorganisations.

For banks, the initiative is an exercise that is best undertaken before Basle II’s 2006 effective date. Ultimately, identifying problems quickly and reducing instances of mismatches can shrink bank capital requirements under Basle II, Lynn noted. The conundrum of Basle II and operational risk is threefold: the proposed capital accord is about best practices; operational risk management is a backward looking discipline; and best practices for operational risk are a work in progress.

Financial institution stock analysts have been slow to take bank operational risk profiles on board, but that will soon change and rating agency interest is acting as a catalyst, officials said. Moody’s Investors Service, for example, earlier this month released operational risk assessments on six European banks, including Deutsche Bank, Barclays and HSBC. Twenty-five more assessments are in the pipeline and an analysis of approaches to operational risk by major European banks is due in October.

Banks will use the assessments for benchmarking. “They will read theirs and other banks’, and that will assist with best practices,” said Brendon Young, operational risk specialist at Moody’s. The rating agency’s global scope should also help regulators cope with home/host issues, as well as facilitate transparency under Basle II’s third pillar.

The prospect of an operational risk risk-transfer market, akin to that in credit, has been contemplated for some time now (see IFR 1428). Some holes, primarily in the area of data, must be addressed first, however (see story on page ...).

Probabilities of operational risk failures (PoFs) are only as useful as the quality of data feeds. A predictive tool that aims to help banks transform reactive operational risk management processes to proactive ones – like any model – can only serve as a guide. Dynamic Anomaly Pattern Response (DAPR) tools and techniques, some officials say, are producing interesting results.

“What hasn’t been clear is the relationship between [operational loss data] and key risk indicators. So, we wanted to see why that is. Maybe, like GDP, it has too many drivers or perhaps there is a time lag,” said Jonathan Howitt, operational risk director at DrKW. Using a year’s worth of real equity trading data from five locations and a streamlined set of key risk indicators, DrKW recently tested the effectiveness of Z/Yen’s prediction tool, PropheZy.

“The outcome, in aggregate, was an accurate forecast of loss severity,” Howitt said, adding that a 12% differential in severity versus predicted loss severity emerged. Twice as many errors were predicted than recorded by the bank, however. “We haven’t quite cracked why that is yet,” Howitt said.

DAPR tools can be used as a prediction engine that lets the back office drive the front office by providing an end-to-end view of a unit’s profitability.

“Technology is five years ahead of people’s mindsets,” said Ben Parker, equity operations productivity and quality programme manager at JP Morgan. For the past three years, JP Morgan has been using the theoretical methodology Six Sigma to identify 70% of the variables that have an impact on trade failure, Parker said, noting that multilinear logistic regression testing was the method for doing this. The bank has not built a prediction engine that gives a probability output and compares it to the actual.

As ‘exception fails’ consume a lot of capacity, the ability to predict which trades have a greater PoF due to factors as seemingly innocuous as the day or time of the week is key. Likewise, the ability to set an impact of the severity of the event occurring and to assign a confidence level to a PoF is essential.

“You work using the data that’s available, so it is always historical and reactive,” Parker said. When information becomes more predictive, there is nothing precluding it from influencing business strategy, he added, noting that outliers will always exist.

Both JP Morgan and DrKW have been testing Six Sigma and DAPR with equity data, but the DAPR and tools like it are more applicable and more useful for derivatives and other complex products.