Understanding Jamie Dimon’s Testimony: the strange case of CRM June 13, 2012 at 9:46 am

In the theory of programming languages, you learn that parsing is a syntactical operation that doesn’t require any analysis of meaning. Therefore I shouldn’t complain that dealbook’s recent post, Parsing Jamie Dimon’s Testimony, doesn’t inform as, really, it doesn’t promise that it will. It does, though, set up enough clues that you can guess a little more of the JPMorgan story.

Here are the key pieces.

  • Dimon said:

    In December 2011, as part of a firmwide effort in anticipation of new Basel capital requirements, we instructed CIO to reduce risk-weighted assets and associated risk. To achieve this in the synthetic credit portfolio, the CIO could have simply reduced its existing positions; instead, starting in mid-January, it embarked on a complex strategy that entailed adding positions that it believed would offset the existing ones. This strategy, however, ended up creating a portfolio that was larger and ultimately resulted in even more complex and hard-to-manage risks.

  • The new Basel capital requirements will require a bank like JPM to calculate capital for the correlation trading portfolio using a new type of internal model, a CRM or comprehensive risk model.
  • CRM models operate on a portfolio basis, and will recognise partial risk hedging. Therefore if you have a position and want to reduce capital somewhat but keep some of the risk, you can do that by ‘adding positions that [you believe]would offset the existing ones’.
  • CRM models do not include investment positions, so if the broad theory that JPM was long deposits, long corporate credit risk to invest then, then using synthetic credit positions to protect the crash risk of the bonds is correct, the CRM model would only have included the last of these positions. Thus JPM would have had both an accounting and a capital mismatch: depos and bonds accural accounted and capitalized in the banking book; protection fair valued and CRM-modelled in the trading book.
  • It seems a reasonable theory then that JPM was trying to address this mismatch by modifying its positions to reduce future regulatory capital (and accounting volatility) while still keeping their essential nature as crash hedges. The modifications introduced extra risk which caused the $2B hole.

That’s my current best guess; I await Jamie’s congressional testimony with interest.

4 Responses to “Understanding Jamie Dimon’s Testimony: the strange case of CRM”

  1. I’ve been wondering if CRM is part of this story but still feel like I’m missing something here. CRM is for capital not VaR (I think?). JP specifically said they changed their ‘VaR model’ and this is what has been widely cited, and which they say they have since reverted.

    But ok, assuming the VaR change was just some change that mirrors CRM treatment, is your theory something like the following –

    call X = long portfolio (accrual accounting/etc), Y = CIO ‘hedge’ position
    clearly the correlation portion of Y was CRM’able, while X wasn’t
    So (the theory is) switching Y to standalone CRM led to an overhang of standalone (long) RWA usage by X that Y was no longer mitigating in the (approx, but somewhat helpful) way it did under old model?

    I guess my only confusion is, wouldn’t the capital/risk of X treated differently for VaR/RWA purposes all along since it was/is banking book, i.e. Y never mitigated X in that way in the first place? This would mean that X and Y have always been treated as standalone for RWA, and thus that the switch to CRM (on Y) could only help things, and there would be no added mismatch to address – only benefit. That’s why I had been rejecting CRM as an obvious culprit here, I couldn’t make it fit in my head with what was known.

    What am I missing..? Best,

  2. SC – almost. Before B2.5, there was a split in capital between X and Y, with X going through the banking book rules and Y through VAR. But the VAR charge wasn’t big. Now, however CRM is being used on Y, and that charge is a lot bigger, so the impact of the mismatch is larger.

  3. http://ftalphaville.ft.com/blog/2012/06/13/1043791/copula-culture/

  4. […] slightly expanded version of this post on the JPMorgan losses, accounting, and CRM models is up on FT alphaville […]