JPMorgan hasn't detailed how or why the New York-based lender altered the VaR formula. The changes — and the timing of the firm's disclosures about them — are the focus of an inquiry by the U.S. Securities and Exchange Commission, Chairman Mary Schapiro told a congressional panel on May 22. Shares of the bank, the biggest in the U.S., have tumbled 25 percent since April 5, when Bloomberg News first reported on Iksil's trades.

Chief Executive Officer Jamie Dimon, 56, has since suspended the bank's $15 billion share buyback program and replaced executives who oversaw the errant trades. Dimon has said the losses could grow and that it might take the rest of the year to liquidate trades at the unit, which is charged with managing the bank's idle cash to earn a profit while minimizing the company's risk.

Iksil, who joined JPMorgan in 2005 according to U.K. regulatory records, was given more leeway than many traders because he produced outsized gains during previous years — including more than $100 million in 2011, said a person close to the bank.

Whale Watching

His bosses may not have understood the complexity of his trades, said the person, who asked for anonymity because the information hasn't been released publicly. Executives and risk managers in the chief investment office were aware of Iksil's positions because they met every Thursday morning to discuss the unit's trades, the person said.

Iksil was assigned to devise hedges and make trades to counter the risk that a faltering economy might lead to a surge in losses on corporate loans or bonds. By 2010, the VaR on his trading book was about half of that for JPMorgan's entire chief investment office, which at the time also oversaw more than $300 billion of securities, according to a person with direct knowledge of the CIO's operations.

VaR represents the maximum JPMorgan traders would expect to lose on 95 out of 100 trading days, according to quarterly filings with regulators. It is calculated daily, and the average for a quarter is reported in regulatory filings.

While there's no estimate of what the losses might be on the worst days, a string of daily losses exceeding the VaR can be a warning that the formulas are flawed or that markets have turned unusually volatile. Dimon had encouraged the once-conservative CIO operation, run by Ina Drew, to boost profit by buying higher-yielding assets such as structured credit, equities and derivatives, Bloomberg News reported on April 13.

Banks and their traders have multiple computer models to estimate potential swings in profits and losses. Value-at-risk is among the most crucial because it's reported to investors in filings that are reviewed by the SEC. Any changes to the model's characteristics are supposed to be disclosed, Schapiro said.

The bank produces more than half a dozen VaR barometers for different parts of the firm, and the chief investment office gets one of its own. Toward the beginning of this year, the bank changed the mathematical formulas used to calculate VaR for that unit, Dimon said on May 10, without elaborating on the reasons.

Quarterly Rise

The new formula showed average VaR for the chief investment office stood at $67 million, according to a regulatory filing on April 13, the day JPMorgan reported first-quarter results. When JPMorgan reverted to the old model, it showed the average VaR was $129 million, and that the figure ballooned to $186 million at the end of the period, a May filing showed.

“We implemented a new VaR model, which we now deemed inadequate,” Dimon said. “We went back to the old one, which had been used for the prior several years, which we deemed to be more adequate.”

Iksil alone may have amassed a $100 billion position this year in contracts on one credit-derivative index, counterparts at hedge funds and rival banks said in April. The holdings amounted to tens of billions of dollars under the firm's own math, a person familiar with JPMorgan's view has said.

The VaR changes may have allowed or encouraged Iksil or other traders in the chief investment office to take bigger positions, said David Hendler, an analyst at CreditSights Inc.

“It's possible that when the new model said, 'Hey, we can put on more risk,' they did,” he said. “And then when they went back to the old model, they saw, 'Oh my God, our risk is much higher than we thought.'”

The bank is conducting its own probe into the CIO's losses and plans to report on the findings, said Kristin Lemkau, a bank spokeswoman. Dimon is scheduled to testify June 13 before the Senate Banking Committee, and will be asked to appear June 19 at the House Financial Services Committee, a person familiar with the plans has said.

Dimon didn't mention Iksil by name on May 10, when the CEO first disclosed the losses and the decision to revert to the old version of VaR. “There are constant changes and updates to models, always trying to get them better than they were before,” Dimon said.

In practice, such updates typically occur no more than once a year, said a former JPMorgan risk manager who asked not to be identified to avoid alienating the bank. Lesley Daniels Webster, a former head of market and fiduciary risk management at JPMorgan, said that new models are usually tested “in parallel” with old models for about three months to make sure they're working properly before being used.

Signing Off

“There's a formal approval process for the adoption of a new model,” said Daniels Webster, who retired in 2005 and runs her own risk-management firm, Daniels Webster Capital Advisors, based in Naples, Florida. “Somebody has to sign off.”

Allen, the other former JPMorgan risk manager, said that a model change big enough to reduce the VaR by half probably would need approval from the chief risk officer, especially if a trading book is unusually large.

Dimon said in April 2009 that he doesn't pay much attention to VaR and has criticized the gauge when analysts questioned him in past years about its levels. VaR is “a very imperfect number” that “bounces around all the time,” he said on a Jan. 18, 2006, conference call.

U.S. banks were warned last year by the Office of the Comptroller of the Currency to closely scrutinize the possibility that computer models used to calculate VaR might not be properly designed or calibrated.

“The use of models invariably presents model risk, which is the potential for adverse consequences from decisions based on incorrect or misused model outputs and reports,” according to the April 4, 2011, document from the OCC, which supervises JPMorgan's primary banking subsidiary. “Model risk can lead to financial loss, poor business and strategic decision-making or damage to a bank's reputation.”

JPMorgan's team that handled such matters was on the verge of a shakeup that involved at least six management changes in late 2011 or early this year within the chief risk office, chief investment office and treasury. The personnel changes may have contributed to lapses in risk management, said the person close to the company.

Barry Zubrow, 59, a Goldman Sachs Group Inc. veteran who was hired by JPMorgan in November 2007, served as chief risk officer through Jan. 12, when Dimon shifted him to oversee regulatory, public relations and lobbying strategy. His replacement was John Hogan, 45, who oversaw risks within JPMorgan's investment bank during the U.S. subprime mortgage crisis in 2008 and 2009.

People in Motion

Hogan, after being named chief risk officer in mid-January, didn't announce his new management team — including his own successor — until Feb. 13, so he was doing both the new job and the old job for about a month, according to the person close to the bank. A chief risk officer signed off on the VaR change, which took effect in January, the person said. It couldn't be determined whether Zubrow or Hogan signed off, and neither responded to phone calls seeking comment.

The new team included Irvin Goldman, 51, who had been overseeing strategy at the chief investment office, as the unit's chief risk officer. He replaced Peter Weiland, who remained with the bank as head of market risk for the investment office, reporting to Goldman.

Goldman, Zubrow's brother-in-law, had been fired in 2007 by Cantor Fitzgerald LP for money-losing bets that led to a regulatory sanction of the firm, Bloomberg reported on May 20. Regulators didn't accuse Goldman of wrongdoing.

Evan Kalimtgis, 42, who co-headed risk management for the $355.6 billion book of securities in the investment office, quit in March after learning that Goldman would become his new boss, people with knowledge of the move said. Kalimtgis, who until mid-2011 was overseeing market risks in the London CIO office where Iksil worked, said he couldn't comment. Keith Stephan succeeded Kalimtgis as head of market risk in the London office, two people with knowledge of the matter said.

Andrew Abrahams, who had been head of quantitative research and model oversight at JPMorgan, reporting to the chief risk officer, retired in May, according to his profile on the website LinkedIn. He's now an instructor at Stanford University near Palo Alto, California, where in January he taught a seminar on “model risks, safeguards and new directions,” according to the school's website.

Leaving the Bank

“My departure from JPMorgan was planned from the middle of last fall, and had nothing to do with the current news story,” Abrahams said in an interview. He said that while his official retirement date was in May, he wasn't physically present after the end of 2011. Abrahams was succeeded by C.S. “Venkat” Venkatakrishnan, according to a Feb. 13 memo from Hogan.

Joseph Bonocore, 44, a former finance chief in the CIO who was promoted in November 2010 to become the bank's treasurer, quit in October to join Citigroup Inc. The job stood open for five months until prime-brokerage and futures chief Sandie O'Connor, 45, was promoted to the post in March. The treasurer shares responsibility with the chief investment officer for managing “capital, liquidity and structural risks of the firm,” according to JPMorgan's annual report.

Drew, 55, who as chief investment officer oversaw Iksil's trades, resigned on May 14 and was replaced by Matthew Zames, 41, who had headed fixed-income trading. Goldman was stripped of his duties in May, though he remains at the firm, according to a person familiar with the situation.

After Dimon held the May 10 conference call to announce the derivatives-trading losses, Hogan sent an internal memo urging his employees in the risk department to “remain vigilant.”

“Our focus is no surprises,” Hogan wrote. “Remember, as an independent oversight function, it's our responsibility to escalate early and often.”

That oversight extends to JPMorgan's chief investment office, which in 2008 and 2009 started expanding its use of credit derivatives to hedge its holdings against an economic slump. The shift was led by Drew's top deputy in London, Achilles Macris, who in turn oversaw Iksil's boss, Javier Martin-Artajo, people with knowledge of the matter said.

Iksil was assigned to design those hedges and execute trades. Last year, Iksil made a bearish bet on an index of credit derivatives, speculating that one or more companies included in the index would default before trading contracts expired in December, according to market participants at hedge funds and banks. Some hedge funds were taking the opposite view.

Winning Bet

Iksil's bet won out, and the hedge funds faced losses of 25 percent, when American Airlines parent AMR Corp. filed for bankruptcy less than a month before the insurance-like swaps matured, the market participants said. The trades were made in so-called tranches of the index, which concentrates risks on the member companies.

Traders sometimes complain about VaR models that “have adverse effects on the business,” and they're “less likely to challenge an outcome that results in an advantage for them,” according to the OCC supervisory guidance.

The agency said May 14 that it was examining the losses at JPMorgan, including “details related to the specific transactions as well as the surrounding risk-management processes that resulted in this unexpected loss.”

The Federal Reserve, as JPMorgan's holding-company supervisor, is studying organizational issues around the trading loss to assure that they aren't repeated in other areas of the firm, Barbara Hagenbaugh, a spokeswoman for the central bank, said on May 15.

The most common reasons for altering a VaR model include changes in the range of historical pricing data used to estimate the potential swings and revisions in the amount of hedging activity that's allowed as an offset, said Daniels Webster, the ex-JPMorgan risk manager.

Some of the toughest questions about the VaR changes may be reserved for the executives who approved them, she said.

“There is no one VaR model out there that is recognized as the sine qua non,” Daniels Webster said, invoking the Latin for something indispensable or essential. “So we're dealing with judgment.”

Bloomberg News

Copyright 2018 Bloomberg. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.