Yesterday’s column at American Banker digs into the accounting for JP Morgan’s reported “hedge”. I was shocked – OK, not really – that no main stream media outlet had explained the stunning announcement made by Jamie Dimon last Thursday of a $2 billion loss on a series of trades made by the Chief Investment Office in accounting terms. CIO is the group purportedly managing the investment of the bank’s excess deposits. In London. As in, the low risk sweep function. Uh-huh.
There’s lots of speculation about the nature of the trade itself. The best I’ve seen is the ongoing coverage at FT Alphaville by Lisa Pollack, in particular.
The gist of all the stories is that the CIO was selling protection on the CDX.NA.IG.9 (going long) to balance out the tranches on the high yield index that they’d bought (going short, which turned out to be profitable when Dynegy and AMR Corp defaulted).
In this way the trade would be both a curve play and across indices — one high yield, one investment grade, with the high yield play levered further because it was a tranche. The long on the IG.9 also would have helped to fund the rather expensive short on the high yield tranches.
If you are an expert in this stuff, please get in touch. I’d buy a big steak for someone who can walk me through it, maybe at a quiet table at Gene & Georgetti’s.
I took a long look at the 10K and 10Q and the first clue was the pretty stark statement, all over the place that the credit derivative number, “Represents the net notional amount of protection purchased and sold of single-name and portfolio credit derivatives used to manage both performing and nonperforming credit exposures; these derivatives do not qualify for hedge accounting under U.S. GAAP.”
So from a financial reporting perspective, all the media and trader chatter about “hedge or bet?” is moot.
The bank may now be calling the positions an “economic hedge” but, in hindsight, they look to me like a series of trades designed to generate income that spiraled out of control on incorrect or ignored risk information and lack of control over traders.
“It was there to deliver a positive result in a quite stressed environment,” Dimon said on the May 10 emergency conference call, “and we feel we can do that and make some net income.”
The rest of the American Banker column provides the details.
I left out a discussion of PwC and VAR and risk management and Sarbanes-Oxley because of space limitations at American Banker. I have been getting a lot of questions about whether or not PwC “audited” the Value-at-Risk or VAR number.
Dimon also admitted on the conference call on May 10 that the bank’s primary risk management tool, Value at Risk or VAR, had failed to signal the magnitude of the looming losses. VAR is a statistical risk measure used to estimate the potential daily loss from adverse market moves. The results are reported to senior management and regulators and they are utilized in regulatory capital calculations.
The first quarter press release reported an average VAR of 67 for the CIO. According to Dimon, CIO implemented a new VAR model on its own recently, which it, “now deemed inadequate.” So CIO went back to the old one, which it “deemed to be more adequate.” (Dimon probably feeling less than adequate about now.) The actual VAR number for CIO for the period was 129. There are now reports that the CIO used a different VAR model than the rest of the bank. That sort of negates the argument that CIO was aggregating risks via VAR from the rest of the bank and “hedging” that risk at a portfolio level. There have also been reports that another area of the bank was making the opposite trades that the CIO was making.
As a non-GAAP metric, JP Morgan’s financial statement opinion provided by auditor PwC doesn’t cover the VAR number or the model that produces it. (The audit opinion doesn’t cover press releases or MD&A either although the auditor has an obligation to review disclosures to make sure they are not misleading or incomplete.) However, as the bank’s primary risk management tool, the VAR model and the policies and procedures around it are certainly reviewed as part of the internal controls over financial reporting. VAR should be on the auditor’s radar as a key indicator of the quality of the bank’s risk management policies and processes. Added May 25: It may or may not be applicable to CIO as some have noted. Changing the model for CIO after several years of “adequate” performance for the bank as a whole should raise red flags. Was the model change properly executed, reviewed and approved at the highest levels?
Added March 18, 2013: Appendix E to SEC Rule 15c3-1 The VaR model must be reviewed both periodically and annually. The periodic review may be conducted by the broker’s or dealer’s internal audit staff, but the annual review must be conducted by a registered public accounting firm.
For more about how models can be used to perpetrate fraud or serve as an excuse for reckless negligence, check out the presentation I made earlier this year at the invitation of Emanuel Derman for Columbia University’s Masters in Quantitative Finance Seminar: Modeling For Fraud, Models Behaving Nefariously.