 |
|
08.01.2009
Elimination of CDS as the necessary prerequisite for recovery
During several months, many reasonable recipes have been suggested to fight the current crisis in general and the global financial contagion in particular. Unfortunately, one of the main reasons (if not the main reason) for the current crisis, namely, a very large volume of credit default swaps (CDSs) on books of many financial institutions, has almost disappeared from pages of newspapers and magazines. The topic was much more popular several months ago but among recent publications we know only of an article by John Dizard in FT on Dec 9. We could not agree more with the suggestion in the title: "Put the credit default swaps market out of its misery" but we believe that the technical reasons for doing so listed by Mr. Dizard should be complemented by a more fundamental one. The CDS market had several peculiar features, which would have lead to a serious crisis even if there had been no subprime loans and all market participants had been honest and qualified, and the same features will prevent a sustainable recovery. First, the mark-to-market procedure could not work for CDOs and CDSs because they mushroomed very fast, and, therefore, a reasonably long period of observation of price dynamics was unavailable. In particular, the volume of information about credit events and reaction of prices of CDOs and CDSs to these events was, essentially, negligible. Thus, the market participants had to rely on in-house models for pricing CDOs and CDSs, and the market prices were determined, mainly, by beliefs in accuracy of the models used. Second, in a financial market, where asset prices may jump, a new derivative security distorts the price dynamics of existing securities even if the underlying sources of risks do not change. One can think about financial innovations as dark matter, which influences the dynamics of classical matter in ways that are impossible to predict with any degree of precision unless the volume of the innovation is small. As a pure economic example, think about a new good introduced into the market: clearly, the whole price system must change, especially if the amount of the new good is large. Specialists in economic theory of financial markets know of distortion of the price system by a new security, but this distortion is ignored in popular models for calculating prices in markets with jumps (the recent favourite being the so-called indifference pricing). To make models simpler, it is presumed that the price dynamics of the existing assets does not change. Hence, the calculated price structure is incorrect. The standard response to this critique is that, if the volume of trades in the new security is small, then it is reasonable to assume that the prices of the existing assets do not change by much and, hence, the calculated price of the new security is close to the correct one. However, the only interesting innovations are those that will be bought in large quantities, as CDOs and CDSs were. If CDOs and CDSs had been introduced on a piece-meal basis, then, after a sufficiently long period of observations, the mark-to-market procedure would have been reasonable; but the volume of CDOs and CDSs grew exponentially. Even sophisticated investors, who used to understand the dynamics of asset prices perfectly, had to rely both on their own beliefs about fundamentals and the beliefs of naïve investors to a much larger degree than usual. The case of CDSs was further complicated by the fact that a large chunk of CDSs were written on CDOs, which themselves were derivatives with low trade volume. We conclude that CDSs should be regarded as IOUs ("I owe you"), whose properties are extensively studied in random matching models in monetary economics. The absence of a centralized exchange exacerbates the problem even further and makes the random matching models perfect tools. One of the standard conclusions of these models is that, typically, the economy can be in several equilibrium states with very different systems of beliefs and prices. But beliefs are fragile, and even a moderate external shock, whose timing is usually unpredictable, can lead to a new equilibrium. If we think of CDSs as IOUs, then we have already observed a similar event ten years ago. After Russia’s default in August 1998, the local prices in Russia remained essentially the same in roubles, but the purchasing power of IOUs of large firms such as Gazprom dropped two-fold – although nobody really believed that Gazprom would default. In financial markets, the external shock is triggered by stubborn Cassandras, who understand that the price system the majority of market participants believes in is incorrect. The first Cassandras, as usual, are ignored by everybody and suffer for being too clever too early (such as George Soros during the dotcom bubble) but, eventually, one or two Cassandras get lucky. Our main conclusions are as follows. First, the current crisis will not abate unless a very large proportion of CDSs is eliminated because their very presence distorts the price system. Buying of CDOs and creation of an exchange for CDSs cannot solve the problem (only new trillions will be lost in vain), because even if the market for CDSs revives for a while, the prices will be subject to changes in beliefs to a great extent; if it does not revive, banks will still hold dozens trillion dollars of illiquid securities with no real price on their books. Spending one-two trillion to buy CDSs in an attempt to facilitate the "price discovery" is hopeless because reasonable market participants must feel that nobody can know the "true price", and the book value of CDSs is much larger than several trillion. Probably, the governments have no real alternative but to cancel most of CDSs (better, all) by force. If a temporary nationalization of the banking system is a necessary prerequisite, then let be it. The monetary easing will not be especially helpful, either. In Russia’s virtual economy ten years ago, there was a huge amount of IOUs in circulation, and standard monetary tools proved to be inefficient for containment of IOUs. For the future, we would like to stress that, each time the "next big thing" grows too large too fast, a crisis is guaranteed because nobody will be able to calculate prices accurately and the price system will depend strongly on the beliefs of the market participants. The learning process will update information about both beliefs and the fundamental sources of risk but the updating will be distorted not only by current but by the past beliefs as well; since the beliefs are inherently unstable, the resultant price system will be unstable as well. The level of unjustified optimism will increase as the time after the last crisis goes by, leading to an increase in the amount of the "next big thing" traded. Eventually, however, the system of beliefs and prices will abruptly change leading to a new crisis. The severity of the crisis will depend on the speed of the growth of the volume of trades in the "next big thing". A natural recommendation for regulators is to regulate the speed of creation of new securities. Finally, as more recent models have recognized, perfect learning is impossible even in the limit, and there is a certain non-negligible lower bound for the degree of ignorance about the market processes. Conversely, the upper bound for the volume of risk-reducing financial innovations is finite and not very large. Thus, instead of blaming the complexity of the models for the current crisis, regulators should strive to make financial markets simpler; whether they ever will really try or be able to do that remains an open question. |  |
