Hiding Risk by Netting Exposures

whistling past the graveyard

Which representation of a bank’s derivative portfolio provides a fairer picture of the risk it presents, the net or gross balances? US banks, operating under US Generally Accepted Accounting Principles (GAAP), report the balance after netting out offsetting exposures with the same counterparty together with collateral. European banks, operating under International Financial Reporting Standards (IFRS), report the balance gross.[1] Consequently, a naïve comparison of banks using total assets as reported under the two different standards gives an erroneous impression that US banks are much smaller relative to their European counterparts. Were the assets reported on a comparable basis, US banks would climb in the rankings. But which comparable basis is the right one? Should the US bank assets be adjusted upward with the netted derivative assets added back, or should the European bank assets be adjusted downward by netting out more of their derivative assets. A number of US banking regulators and experts have recently started calling for putting the gross exposure onto the balance sheet. Not surprisingly, the big US banks and derivative trade associations like the International Swaps and Derivatives Association (ISDA) argue that the net exposure is the right one.

What is at the root of the disagreement?

It’s all about whether we examine the bank before or after it fails. Are we concerned about losses incurred in bankruptcy, or the danger that the bank may be driven to bankruptcy?

The US bankers and ISDA make their case based on how a bank’s derivative assets and liabilities are treated after the bank fails, in bankruptcy. They argue that the Master Netting Agreement negotiated between a bank and its customers governs the settlement in bankruptcy, allowing the offsetting of derivative assets and liabilities. Suppose the bank currently owes a customer $100 on a derivative, but the bank also is currently owed $70 by the same customer on another derivative, then in bankruptcy those two amounts are netted so that the bank only owes the customer $30. Therefore the maximum exposure to the customer of a potential failure by the bank is the $30 net derivative position.

The problem is that this netting only takes place if we arrive at bankruptcy with those same exposures in place. A lot can happen before a bank fails which changes the picture dramatically. More importantly, what we first need to be concerned about are the forces that drive a bank towards failure. Here the gross exposures are important.

For a bank that deals in derivatives, a large fraction of the derivatives should be thought of as short-term positions in the same fashion as demand deposits. Counterparties expect their dealers to stand ready to liquidate or novate positions, just as people expect to be able to withdraw the balances in their checking and savings accounts–on demand. Were a dealer bank to hold up liquidations and/or novations, it would only adds fuel to the fire of suspicion that the dealer bank is insolvent. Consequently, a small net exposure at a dealer bank can suddenly balloon to a large net exposure over the next few days as counterparties liquidate or novate the positions on which the bank owes them money. It is the gross scale of positions that reflects the potential scale of such a run. A regulator that paid attention only to the banks net exposure would not have appreciated the dangers. Moreover, should the bank ultimately arrive in bankruptcy where the derivative assets and liabilities are netted against one another, the scale of actual netting will now be much smaller than originally anticipated due to the intervening bank run. The original net exposure was deceptive in measuring the ultimate losses to those parties left holding the bag.

We need only look back at the two major bank runs of 2008 to see the point—Bear Stearns and Lehman Brothers. The Report of the Financial Crisis Inquiry Commission is especially detailed on how derivative counterparties participated in the run on Bear–see esp. pp. 287-288 & 291. See also their conclusion on Lehman, p. 343. Bryan Burroughs’ account of the run on Bear in Vanity Fair also discusses the role played by derivative counterparties in the run on the bank, as does William Cohen’s book House of Cards. Finally, Darrel Duffie’s article on the Failure Mechanics of Dealer Banks in the Journal of Economic Perspectives makes the same point.

[1] It is a little sloppy to say that US GAAP allows netting while the IFRS do not. Both standards permit netting under certain circumstances. US GAAP allows significantly more netting than the IFRS.


  1. Posted April 4, 2013 at 6:40 am | Permalink

    You make some very important points which dig much deeper into the issue than the usual discussion. I need to take some time to think my way through your points. My main point was that the argument in favor of taking the net exposure at face value ignores the dynamics. It shows a snapshot of exposure assuming the risk referees blow a whistle and declare “game over” and all players simply count up their positive and negative exposures. In reality, there are no referees and no sudden whistle. There are complicated market dynamics, sudden events and rumors, and there are strategic actions taken by all parties which dramatically alter the positive and negative exposures as the game plays out. The final result can be very different from what the net exposure had predicted. Your point, as I understand it, is that collateral works differently in this dynamic game than do offsetting exposures. And that CSA’s shape they dynamics, along with other things. You are basically agreeing that we shouldn’t ignore the dynamics, as a simple netting would. But you are delving deeper into what determines the dynamics. Good point.

  2. Posted April 3, 2013 at 11:36 pm | Permalink

    Dear Prof Parsons

    That is a very interesting view point about why we should consider Gross exposures. If I understand it correctly, it effectively states Netting (which offsets the exposure from one derivative against another with the same counterparty under the same master agreement) should not be considered. It doesn’t say much about collateralisation. Is collateralisation deemed different in this aspect(banks have this dirty jurisdiction problem in recognising collateral posted against derivatives in certain jurisdictions, am not sure if it is related to Netting, since using collateral to offset an exposure uses the same principle of netting). Because if a counterparty is required to post collateral under a CSA arrangement (I am assuming Netting is a prerequisite for this) then as a bank I would be ok with unwinding such exposures as you have stated above as that would require the unwinding party to post significant collateral to keep my net exposure same. I understand that a lot of corporates typically do not enter into CSAs (which is the focus of this blog), but I suspect that would be addressed going forward with the new regulations and as it makes it quite expensive to trade in the absence of CSA. But without such a collaterlisation provision, I agree with you that it can give rise to higher exposures. But if I understand the banking business right, bulk of their derivative exposure comes in dealing with institutional counterparties (including hedge funds) where Netting and CSA are common. I personally don’t think CSA should be ignored (which seems to be the case under IFRS), other than in a dirty jurisdiction case. Otherwise reporting gross exposures does seem to be a useful idea (though I never thought of it that way till I read your blog). Of course, if the bank is stressed can it pay the counterparty the high MTM on unwind even if they post the right collateral? I suspect, with all the windows now available for borrowing against all kinds of collateral to banks, it is possible and typically CSAs only accept high quality collaterals. In fact post-crisis this has become a key issue with the banks, which started looking at the kinds of collateral they have been accepting from the counterparties and in fact the valuation of the derivative is now a function of the type of collateral posted (essentially reflecting its liquidity not just is credit quality) and we have a new thing called Liquidity Value Adjustment (in addition to CVA) to deal with this problem. So a bank that has implemented all this valuation adjustments in their systems and valuing the derivatives accordingly and which has access to the collateralised lending windows in their jurisdictions should be allowed to Net their exposures against such collateral if not against the other derivatives as you have pointed it out.


One Trackback

  1. […] mask the risk of runs and potential insolvency; as my colleague John Parsons has explained, gross exposures can tell us a great deal about vulnerability and therefore systemic […]

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: