I know there are very few new things under the sun . . . but although I’ve read a great deal about clearing, I don’t think that I’ve seen anyone point this out: The main benefit commonly attributed to a central counterparty (CCP)–netting–effectively gives derivatives counterparties a priority in bankruptcy that they would not otherwise possess. As a result of the effective change in priority, the initiation of clearing transfers wealth from CCP members’ other creditors to their derivatives counterparties (who include CCP members.) That is, in dollar terms, formation of a CCP is not a positive sum game (although it can be from the perspective of the CCP members), but a zero sum game. CCP members win at the expense of the other creditors of their members.
This has important implications. Advocates of centralized clearing argue that netting reduces systemic risk by reducing the cost of replacing defaulted derivatives positions. In a bilateral market, counterparties must replace all of the defaulter’s out-of-the money positions; with a CCP, they replace only the net positions. Since replacement is costly (as the counterparties lose in-the-money positions and replace them with new ones of zero value), netting reduces the losses counterparties incur, thereby reducing the likelihood that the counterparties will in turn default; that is, netting reduces the susceptibility of the market to a “daisy chain” of defaults.
Or does it? What about the other creditors? Netting eliminates an asset: the dealer’s in-the-money derivatives positions that are net against some or all of its losing positions. Without netting, both derivatives counterparties and other creditors would have a claim on this asset. With netting, the asset disappears, meaning that the other creditors cannot use it to satisfy their claims against the bankrupt dealer. Thus, netting favors derivatives counterparties at the expense of other creditors.
This means that holding the dealer’s position constant, netting engineers a transfer of default loss, not a reduction of total default loss. It only does so when one considers only derivatives counterparties, and ignores the defaulter’s other creditors.
This has implications for systemic risks. Perhaps netting reduces the possibility that a dealer’s default will cause derivatives counterparties to default, but it increases the likelihood that its other creditors will face financial distress, and perhaps bankruptcy. The defaulter’s other lenders may include banks, hedge funds, money market funds, commercial paper investors, and so on. These lenders may also be systemically important, in the sense that their financial distress can have ripple effects, or trigger a cascade of subsequent defaults.
It is therefore by no means obvious that the ultimate effect of the formation of a CCP is a reduction in systemic risk exposure. Netting redistributes the wealth effect of a default. Unless derivatives counterparties are uniquely systemically important, netting can actually worsen systemic risks.
These points can best be illustrated by an example. Large dealers typically buy and sell the same contract in large quantities. For instance, a dealer may buy 1000 contracts and sell 500 of the same contract to different counterparties. This dealer has a net exposure of 500 long contracts.
In a bilateral market, the offsetting 500 contracts are not eliminated by netting. This has several implications. First, a firm with offsetting positions often has to post collateral on the both the purchased and sold contracts, even though some of those contracts offset (thereby resulting in no market risk exposure.) Since collateral is costly (because it requires the dealer to hold liquid instruments with lower yields than alternative investments), the firm incurs a cost on the offsetting positions that could be avoided if these positions were netted out. Therefore, given the positions held by market participants, the formation of a CCP that nets positions economizes on collateral.
Second, as shown by Jackson and Manning (2006), the lack of netting increases the costs of replacing defaulted positions. Consider a dealer in a bilateral market who is long 1000 and short 500 contracts. Assume that the dealer defaults when the long positions are a liability to him, and the corresponding 1000 short contracts are an asset to his counterparties, and his 500 short positions are an asset to the dealer. The counterparties must replace the contracts at the prevailing market price (at which the short positions they enter have zero values, and hence are neither asset nor liability.) This is an economic loss to the victims of the default, which is reduced by whatever they realize from their claim on the assets of the defaulter, which will be less than what they are owed. It is important to recognize that these replacement costs are incurred on the entire gross position of 1000 contracts.
In contrast, in a cleared market, replacement costs are incurred only on the 500 net positions. Thus, clearing and netting reduces the losses that the defaulter’s counterparties incur to replace the defaulted positions because fewer positions (and sometimes no positions) must be replaced.
It should be noted that this “benefit” from netting is in fact a transfer, rather than a true cost savings. In the absence of netting, the non-offset in the money positions (in the example, the dealer’s 500 short contracts) are an asset to the defaulting firm. This asset can be used to pay the defaulter’s creditors, including derivatives counterparties and others. Netting reduces assets and liabilities by an equal amount; netting reduces the claims against the defaulter’s assets, but reduces the assets by an identical amount. Thus, the total loss suffered by creditors does not change, although the allocation of the loss is likely to change.
In the context of the example, without netting shorts holding 1000 contracts must share with all the other creditors the defaulter’s remaining assets, including the defaulter’s 500 short positions. With netting, in contrast, shorts holding only 500 contracts must share with all the other creditors the defaulter’s remaining assets, which do not include the 500 netted short positions. Thus, in effect, netting gives derivatives counterparties a priority claim to one of the defaulter’s assets–his winning derivatives positions–assets that they would have to share with other creditors absent netting.
An extension of the example illustrates these points. Assume that in a default, the 1000 contracts cost the counterparties $1 billion to replace, and hence the defaulter’s 500 short contracts are worth $500 million to the defaulter in the absence of netting. The defaulter has other assets of $700 million, and other liabilities of $1 billion. In a bilateral setting, the defaulter has assets of $1.2 billion, and liabilities of $2 billion. If the defaulter’s assets are split pro rata among the derivatives counterparties and the other creditors, the victims of the default receive $600 million, and hence lose $400 million. Other creditors also receive $600 million and lose $400 million.
With netting, the derivatives counterparties have a claim of $500 million ($1million on each of the 500 contracts that do not net out), and the defaulter’s assets are $700 million because netting reduces both assets and liabilities by $500 million. Pro rata division of the defaulter’s assets results in a payment of $233.3 million to the derivatives counterparties, meaning they lose $266.7 million. The other creditors receive only $466.7 million and lose $533.3 million. Thus, although netting makes the derivatives counterparties better off, it makes the other creditors worse off by the exact same amount.
Thus, netting effectively transfers wealth in a default from a defaulter’s other creditors to its derivatives counterparties. In the example, netting redistributes $133.3 million from other creditors to the derivatives counterparties. Reductions in replacement costs therefore do not reduce social costs, unless some frictions, related perhaps to systemic risks, imply that giving one set of claimants priority enhances wealth. I discuss this issue further below.
In sum, netting by a CCP benefits its members by (a) reducing collateral, and (b) reducing replacement costs. The important part of that last sentence is “its members.” It is important for two reasons. First, this is a private benefit that these firms will presumably take into consideration when evaluating the net benefits to them of cooperating to create a CCP. Second, since one of the benefits to a CCP’s members is actually a transfer from other creditors, dealers may have an excessive incentive to form a clearinghouse. This makes the absence of a CCP all the more peculiar–and supports the possibility that other costs, such as those arising from asymmetric information, exceed the private benefits of netting received by CCP members.
Netting can also affect systemic risk by affecting the scale of dealer trading activities. To the extent that that the introduction of a CCP reduces the costs dealers incur at the margin to engage in derivatives trades, dealers will trade more. By accumulating bigger positions, they can pose a greater systemic risk.
Netting can affect marginal cost in at least two ways.
First, netting reduces costly collateral. Lower collateral costs induce greater trading activity. (In the 1920s, opponents of central clearing at the CBT argued that bilateral trading encouraged “conservatism”–i.e., less trading–because it imposed higher collateral costs which restrained trading activity.)
Second, reductions in replacement costs could have a similar effect. Here the analysis is more complicated, however. Holding the terms of the dealers’ other liabilities constant, a reduction in replacement costs due to netting encourages dealers to trade more. However, the dealer’s creditors will certainly respond to the implicit reduction in their seniority resulting from clearing by adjusting the terms on which they supply capital to the dealers. Increasing the scale of trading activity increases the default costs other creditors incur. If creditors price capital appropriately, and charge the firm an additional dollar for each additional dollar of default cost they absorb as a result of an expansion in dealers’ trading activity, the dealers will internalize the cost of the transfer, and reductions in replacement costs will not distort their incentives to trade. Given the complexity of dealer balance sheets and the difficulty in evaluating total derivatives exposure, however, it is possible that other creditors will not fully recognize the implications of an expansion of a dealer’s trading activity, and underprice capital as a result. This would encourage the dealer to expand trading activity excessively, as the costs of this expansion are borne by other creditors.
To the extent that lower collateral requirements and replacement costs encourage dealers to trade more, reductions in dealer default exposures that result from clearing will be smaller than one would estimate holding positions constant. Larger positions increase default losses, offsetting in part or in whole the effect of clearing on default losses per trade.
In sum, it is clear that the distributive effects of clearing have not been widely appreciated–if they have been recognized at all. Netting via clearing effectively gives derivatives counterparties a priority claim on some of a dealer’s assets–the portion of a dealer’s positions that are offset by other positions. In a bilateral market, all creditors, not just derivatives counterparties, would have a claim on that asset. This redistributes the burden of a default from derivatives counterparties to a defaulter’s other creditors. Netting is therefore effectively a zero sum game, not a positive sum game. Moreover, the losers may not be members of the clearinghouse, which could distort the incentives to create a CCP. Furthermore, the systemic effects of this redistribution are ambiguous. Imposing additional losses on other creditors could put them into financial distress, even while it eases financial distress from derivatives counterparties. Financial distress by these creditors could also pose a systemic risk.
Thus, the debate over the formation of a CCP should be framed very differently than it has been heretofore. The appropriate question is: Should derivatives counterparties receive a priority claim on a defaulter’s assets? That is, should derivatives counterparties receive priority over a defaulter’s other creditors? The answers to these questions are not immediately obvious.