Overexposed?
Just before Christmas, the Basel Committee on Banking Supervision released its “Consultative Document on Capitalisation of bank exposures to central counterparties.” In plain English, it sets out principles on how to determine the amount of capital banks must put aside to protect against the risk of a clearinghouse default, or the risk that the margins a clearinghouse collects are inadequate to cover the loss of a defaulting trader or traders.
A couple of days back I commented on the token 2 percent capital charge against trade exposures. Here I’ll take up the more complicated issue of the capital charges for the banks that are clearinghouse members that must stump up funds to cover the loss (over margin) arising from the default of any other member firm.
In a nutshell: the rules illustrate the fundamental problems with Basel in all its incarnations.
The proposal sets different levels of capital charges, depending on whether the amount of funds precommitted to the clearinghouse (its own resources, or its own monies plus any monies invested by member firms) are larger or smaller than the CCP’s “hypothetical capital.” If the CCP’s own resources (e.g., contributed capital, retained earnings) exceed the hypothetical capital, the amount of capital a bank exposed to the default by another member must hold is 1.6 percent of its default fund contribution. If the hypothetical capital exceeds the CCP’s own resources, but is less than the sum of those resources and precommitted default fund contributions, the capital charge is proportional to 1.6 percent of the difference between the total prefunding and hypothetical capital plus 100 percent of the difference between the hypothetical capital and the CCP’s own resources; this is effectively a 100 percent capital charge on the marginal dollar in the default fund. When the hypothetical capital exceeds the sum of the CCP’s resources and its members’ prefunded contributions to the default fund, the capital charge is proportional to the sum of 100 percent of the members’ prefunded contribution to the default fund and 1.92 percent of the difference between hypothetical capital and the total prefunded contributions. (These calculations assume that the CCP meets to-be-released CPSS-IOSCO standards. The capital charge for non-complying CCPs is based on a risk weight of 1250 percent.)
Using the “waterfall” analogy, the capital charge on default fund contributions is 1.6 percent if the first element of the waterfall–the CCP’s own financial resources–exceeds hypothetical capital. If the hypothetical capital exceeds the resources in the first element of the waterfall, the capital charge depends on how much of a call the hypothetical exposure represents on the banks’ default fund contributions. If the hypothetical capital exceeds the resources in the second element of the waterfall, the capital charge depends on how much of a call the hypothetical exposure represents on the bank’s obligation to commit additional funds in the event of an exhaustion of the CCP’s resources and default fund contributions.
The key variable here is the hypothetical capital. This is calculated using the “Current Exposure Method,” which is defined here. The current exposure is the sum of the mark-to-market value of a position (net of collateral) and “amount for potential future credit exposure calculated on the basis of the total notional principal amount.” This notional principal-based amount depends on the instrument in question, with the exposure of interest rate derivatives being a small proportion (0 to 1.5 percent, depending on maturity) of notional, and at the other extreme, commodity exposure ranging between 10 and 15 percent of notional, again depending on maturity. Hypothetical capital for a given notional amount held by a particular clearing member times the weight is equal to that product, multiplied by 1.6 percent (a 20 percent risk weight times an 8 percent capital ratio).
Given that CCPs mark positions to market frequently, and require posting of initial margin, the mark-to-market portion of the hypothetical capital is/will be zero, or quite close to it. This means that hypothetical capital for CCPs will be, almost always, a multiple of notional amount.
However, notional amount is a poor–very poor–proxy for the risk of underlying positions. The default risk posed by a given notional amount can vary substantially even within the five categories set out in the Basel rules. For instance, different currencies have different risks, and those risks differ than the risks in gold; different equity products (e.g., different indices) have different risks; and different commodities can have substantially different risks (compare, for instance, natural gas or electricity with oil or corn). But the rules treat everything within a given type-maturity bucket as equally risky. Moreover, the weights across categories are highly dubious. Is three month silver seven times riskier than three month gold? Is three month oil ten times riskier than three month gold?
Furthermore, the risk posed to a CCP’s capital and the capital of its members also depends on the riskiness of individual clearing members (via their balance sheets). The hypothetical capital calculation does not take this into account, although in a footnote (note 21) the document does suggest the possibility that the risk weight in the hypothetical capital calculation can be adjusted to reflect for the rating of CCP members: the document says bank supervisors “may increase the risk weight . . . if . . . the clearing members of the CCP are not highly rated.” As written, this seems like a blanket adjustment based on the collective rating strength of the members, rather than an adjustment that reflects differences in ratings among members. And I shall pass over without comment the reliance on ratings as a measure of credit quality.
Lastly, the hypothetical capital merely sums these (imperfect) exposure amounts across the members of a CCP. This does not reflect the correlation of exposures across members; arguably the 20 percent risk weight is a correlation adjustment, but a very rough one that does not vary in a discriminating fashion with the way that individual exposures contribute to a CCP’s total exposure. (As one example: a CCP clearing credit derivatives on financial institutions has a very different structure of exposures, and interrelationships between them, than a CCP clearing commodity products.)
In sum, capital charges depend on hypothetical capital, and hypothetical capital does not vary in an economically sensible way with the risks that CCP exposures pose to member banks.
Given the substantial increase in capital requirements for CCP members of clearinghouses with resources less than hypothetical capital, the rules will provide a strong incentive to make CCP capital (retained earnings plus contributed capital) somewhat greater than this hypothetical capital. But given that this hypothetical capital is only tenuously related to CCP risk (for the reasons discussed above), this provides little assurance that CCP capital is adequate to absorb the actual potential default losses.
This further means that the capital charge of a member bank attributable to its participation in a default fund bears little association with the true risk to the bank’s capital arising from its CCP membership. A bank that is a member of a CCP that has capital in excess of hypothetical capital may still be very much at risk to suffer losses via its default fund contributions, and its commitment to replenish default funds. Its capital charge will not reflect this exposure, however. This will permit banks to take higher exposures to other risks. This, in turn, will have perverse feedback effects because the true exposure of members to each other depends crucially on the riskiness of member balance sheets. Capital charges that do not reflect true CCP exposure risk frees up capital that can be used to take on other risks that inflate the riskiness of the CCP.
As I’ve noted, and as Deus Ex Macchiato emphasizes, based on a directive from the G-20, the new capital regime is intended in part to shove derivatives risks from bilateral deals to central clearing. That’s troublesome given that, if my analysis is correct, the CCP capital requirements don’t adequately reflect actual risks. It is by no means clear that the new rules won’t actually reduce the amount of capital held against derivatives risks. And remember, blather about CCPs reducing interconnectedness aside, this risk taken off of balance sheets by shifting bilateral exposures to CCPs goes right back to CCP member banks via their capital and default fund contributions. It would be nice to have some confidence that this risk recycling actually results in an increase in the capitalization of derivatives risks. If it doesn’t, the supposed systemic risk-reducing benefits of clearing will prove chimerical–or worse. It could indeed lead to an increase in the scale of derivatives trading, and the amount of derivatives risk that banks bear.
It’s also worthwhile to contemplate the implications of these proposed rules for incentives on various decision margins. For instance, with marking to market and some initial margining, capital charges will not vary with initial margin levels. Thus, even though reducing initial margins would increase the exposure of CCP capital and member default fund obligations (prefunded and contingent), a CCP could reduce initial margins without increasing the capital charges its members incur for their CCP exposures. Given that members may benefit from lower initial margins (margins are costly, and lower margins may lead to greater trading activity and higher revenues for member banks), this would tend to provide an incentive to reduce initial margins.
Against this, one could make a couple of arguments. One is that the CCP members will recognize that reducing margins reduces the buffers protecting the other stages of the waterfall and thereby raises their risks. This will lead them to resist such efforts.
But if you believe that bank self-interest is sufficient to ensure that they choose the right level of collateralization, you should also believe that capital rules are altogether superfluous. After all, the ostensible rationale for these rules is that financial institutions don’t internalize the costs of the risks that they bear, and hence when left to their own devices will take on risks that are outsized relative to their capitals. So if you believe that capital requirements are essential due to some mispricing of risk, you also have to believe that undermargining is a serious concern if capital requirements do not properly account for the relation between collateral levels and the amount of risk that CCP members bear.
You could also argue that regulation of margins directly, via CPSS-IOSCO standards and prudential regulation of CCPs, will overcome this problem. Given the complexities of margin, the information advantage that CCP members have over prudential regulators, and the vulnerability of prudential regulators to capture and pressure, this is a very thin reed to lean on.
One last thing. Beating me to the punch, David at DEM compares the default fund to the mezzanine tranche of a CDO. He further notes that “mezz tranches in Basel are punitively treated.” He interprets the Basel proposal thusly:
The Committee tentatively proposes to use this idea, and figure out whether the CCP’s first line of defence – margin – is enough to cover the capital that would be required for its portfolio of exposures. If it is, then default fund contributions aren’t that risky; if not, then they are acting in part like capital (i.e. margin is too low), and thus they should be treated as pretty risky.
The idea is a reasonable one. But the details are delicate. I haven’t run any numbers yet, and for once I’m excited about seeing how they come out. Will the major CCPs turn out to have enough margin based on the Basel methodology?
My analysis above suggests that the answer to this question is no. That in reality, the proposal will have no bearing on whether CCP margins are accurate or adequate. Or put differently, capital charges will depend tenuously–very tenuously, at best–with the adequacy of margins.
And taking the CDO analogy further, we know that there is a good reason to treat mezzanine exposures punitively. They are very risky–tail risky. In particular, as the whole subprime fiasco demonstrated, mezz tranche risk is systematic risk: they perform very badly precisely in those states of the world in which returns on financial assets generally are low. They are like short, out of the money puts on the market portfolio. Put differently, default funds take a hit precisely in those states of the world in which banks in general are likely to be in bad shape. (And as puts, these exposures have embedded leverage.)
I fail to see how the proposed CCP capital rules really take this systematic tail risk into account. If you believe that some other market or regulatory failure makes capital requirements necessary, banks by themselves won’t take the systematic tail risk into account voluntarily. So if the capital requirements don’t get them to internalize that risk, they’ll take on too much of it.
Which means that the new rules will not induce the proper scaling of derivatives trading activity. The markets will be too big.
In sum, IMO the Basel proposal does not give incentives to (a) choose margins appropriately, (b) control the risks on bank balance sheets (which feedback to CCP risks), (c) rightsize the derivatives markets, or (d) allocate trading activity efficiently across firms and products. All of which is worrisome, and reinforces my concern that clearing mandates have, at best, exchanged new risks for old, with no confidence that the exchange ratio is one-for-one.
