ADVERTISEMENT

Barclays Got Amber Alert After Failing to Predict Trading Losses

Barclays Got Amber Alert After Failing to Predict Trading Losses

(Bloomberg) -- As an activist investor targets Barclays Plc for taking on too much risk, the London-based lender has disclosed how its internal models failed to predict a series of trading losses at its investment bank.

Barclays issued a so-called amber alert last year after the issues with its trading arm’s value-at-risk modeling, according to a recent filing. Tests of its models identified more unexpected losses than international rules allow for “green” status, leading to the downgrade of its risk-management systems during a six-month period in 2018 -- well before the selloff that roiled markets at the end of last year.

While Barclays’s trading results have outpaced many European rivals, the amber alert -- and others that have occurred since 2016 -- show the potential perils from Chief Executive Officer Jes Staley’s push into more lucrative but complicated trading strategies.

“If your model is throwing up lots of exceptions, but other banks’ are not, it’s your model that has the problem,” said Paul Sharma, a former executive director of the Bank of England and co-head of the financial industry advisory practice at Alvarez & Marsal. “It’s pretty rare, outside of severe market disruptions, to see a model in the ‘amber’ zone.”

Dropping from green to amber can suggest a problem with a bank’s ability to gauge the amount of risk that its traders are taking, and can allow regulators to apply a capital buffer penalty.

"We stay in regular close contact with our regulators on all aspects of market risk, including value-at-risk exceptions,” a spokesman for Barclays said. “During amber periods, additional risk-weighted assets weighting is applied automatically, and this is already reflected in our capital ratios."

Complex Models

Banks are allowed a maximum of four exceptional losses before they lose green status. As banks became more dependent on complex computer models, the three-color system to measure their accuracy was introduced about two decades ago by the Basel Committee on Banking Supervision.

Tim Throsby, a former JPMorgan Chase & Co. colleague of Staley’s, runs the investment bank, and has said the lender needed to rediscover its "commercial zeal" after losing "a suitable focus on opportunities to deploy risk." The two men’s enthusiasm for risk-taking is at odds with Edward Bramson, who has amassed a stake of more than 5 percent in Barclays and wants the investment bank pared back.

Among other major lenders, Royal Bank of Scotland Group Plc said its NatWest Markets Plc trading unit breached the green model threshold in December, when it recorded a fifth backtesting exception, largely driven by losses in its rates business. Like Barclays, RBS has had to increase the capital buffer for its trading arm.

One banking industry analyst said that falling into the amber zone is not necessarily a serious problem.

"It’s not a good sign, but I would be surprised if Barclays use value-at-risk models as the key tool for managing risk,” said Edward Firth of Keefe Bruyette & Woods Inc. “At most, it will be one of a number of tools. For many areas, it is only really used for regulatory filings."

The final quarter of 2018 was difficult for most banks as volatility spiked. Conditions, though, were mild compared to the global financial crisis in 2008, when daily trading losses were well beyond anything predicted by internal models.

In 2015, the bank reported that all major trading units were “green.” A year later, it disclosed that Barclays Capital Securities Ltd., a largely equity-focused unit, dropped to amber as its model failed to predict several losses. BCSL’s model remained in amber status in 2017, the bank said.

To contact the reporters on this story: Harry Wilson in London at hwilson57@bloomberg.net;Stefania Spezzati in London at sspezzati@bloomberg.net

To contact the editors responsible for this story: Ambereen Choudhury at achoudhury@bloomberg.net, Keith Campbell, Marion Dakers

©2019 Bloomberg L.P.