Socialist Calculation Meets the OTC Markets
| Craig Pirrong |
A new Federal Reserve Bank of NY staff report by Darrell Duffie, Ada Li, and Theo Lubke, “Policy Perspectives on OTC Derivatives Market Infrastructure” has received a lot of attention in the press.
There are some good things in the paper. Notably, it is suitably cautionary about the potential systemic risks posed by central counterparties, and the consequent need for prudential regulation thereof. It also makes a good case for data repositories, and for the role of the Fed and other government agencies in reducing the costs that intermediaries incur to coordinate risk-reducing actions, such as portfolio compression and improvements in the process of confirming deals.
But overall the paper is extremely disappointing. Its tone is Olympian and prescriptive. The word “should” is used 61 times 21 pages of text (that includes several space-eating tables and charts).
This is extremely dangerous because these prescriptions and dictates are not based on a a rigorous analysis of costs and benefits. Most disturbingly, there is virtually no discussion whatsoever of the informational demands inherent in the prescriptions. We’re told that regulators should set the right capital and collateral requirements on non-cleared deals, and that CCPs should maintain “high collateral standards.”
Great. Thanks for the advice.
Now, give me some numbers. What is the right capital requirement? How high is high?
Can’t do that off the top of your heads? Understood. Then tell us the information required to establish these requirements. Tell us the process by which that information will be collected. Tell us about the incentives facing those using that information, so that we can have some confidence that the regulatory process will select appropriate charges.
In essence, Duffie, Li and Lubka (DLL) are recommending a type of financial central planning, a centralized price setting system. Regulators set a vector of prices, in the form of capital requirements and collateral charges (or regulate/oversee the collateral charges and other decisions of CCPs).
The first, second, third, . . . Nth question that should come to mind when hearing such a recommendation is: “How do you address the knowledge problem?” Or, “What information do you need to set these prices?” Relatedly: “What are the costs of getting the prices wrong (as you certainly will)?”
These prices create a high powered incentive system that very smart, very motivated people will respond to. If the prices are set wrong, because the regulators don’t have the information necessary to set them properly, or the incentive to do so, or both, these responses can be individually rational but systemically perverse. Extremely.
This is especially true inasmuch as these are one size fits all charges, so the effects of any mistake are almost by definition systemic.
And it’s not like we don’t have examples of this. The Basel standards created incentives that drove many of the (correlated) risk taking and investment decisions that contributed substantially to the financial crisis. Which goes to show that setting these prices improperly can create the very problems the prices are meant to solve.
The knowledge problem here is extremely complex. To set the relevant prices properly, you have to understand the marginal contribution of any particular transaction to risk. Moreover, if the objective of the exercise is to reduce systemic risk, you have to have some estimate of the cost of systemic risk, and the contribution of particular transactions to this risk. We don’t even have an agreed upon definition of systemic risk, let alone a way to measure it or its costs, or the contribution of particular instruments or trades to it, in the complex system that is the financial market. So just how do we set this price?
DLL don’t even raise any of these questions, let alone answer them. This is a classic example of what Coase called blackboard economics.
Moreover, adverse-selection problems are likely to be rife here. Again, the highly motivated bankers and traders are going to have much better information than the price-setting regulators. They are going to find out what risks are underpriced and what risks are overpriced, and trade accordingly. This is a recipe for trouble.
The fundamental prescriptions regarding incentivizing clearing and raising capital charges on non-cleared trades is also based on a superficial analysis, as is all too common. DLL argue that an externality is pervasive in OTC trading. For instance, they claim that traditional capital charges do not “provide a direct incentive to [a] bank to lower the exposure of its counterparties to the failure of the bank itself, that is, the potential losses of others that are based on the bank’s payables.” They relatedly argue that there is a negative externality in OTC trading, due to the failure of counterparties to take into account the systemic risk inherent in their trades.
I find it hard to believe that DLL take seriously the argument regarding a bank’s failure to consider the effect of its potential default into account when making trading decisions. Yeah, maybe A doesn’t take into account of its failure on B, but B sure the hell does. Therefore the prices at which B is willing to deal with A, and the amount it is willing to trade with A, depend on B’s assessment of A’s default risk. There is no externality here. The risk is priced.
Surely, in a world of imperfect information, the risk is priced imperfectly. But that gets us back to the question of who is in a better position to set the price for this risk, and the relative merits of two mechanisms: one in which a one size fits all price is set, and hence errors are likely to have systemic consequences, or one in which many economic agents utilize dispersed information to make individual judgments about the terms on which they are willing to trade.
DLL also predicate their analysis on an assumption that policy should be devised to reduce the likelihood of “runs” on OTC dealers. This is understandable, and conventional, but it fails to come to grips with recent research on liquidity and financial intermediation (much of it done by Doug Diamond) which shows that runs — and the fragile capital structures that can produce them — might well be a feature, not a bug. That is, the risk of runs is an incentive mechanism that deters opportunistic behavior by intermediaries. This is, to be sure, a somewhat controversial position, but it is sufficiently grounded in logic to at least deserve some attention.
In sum, the DLL analysis raises some important concerns, but overall it cannot provide any real policy guidance because it fails to address the fundamental information problems inherent in its policy prescriptions. It’s like the old Steve Martin routine, “How to Make a Million Dollars Without Paying Taxes”: “First, make a million dollars.” DLL’s version is “How to Price Performance Risks to Save the Financial System”: “First, price the performance risks properly.” Gratuitous advice, and advice that can be very costly indeed if the fundamental informational problems are too daunting, as is quite likely to be the case.