Dominic O’Kane, of the EDHEC Business School, has performed an analysis of the relationship between the credit default swaps market and the market for the underlying bonds, with reference specifically to the bonds issued by the Eurozone’s more credit-challenged countries.
The result is a working paper for the EDHEC-Risk Institute that is at least in part a reaction to the European Parliament’s ban on naked CDS’ in October 2011. That Parliament was clearly acting on the hypothesis that speculative trading in CDS’s in 2010-2011 had pushed bond prices down in the eurozone periphery, worsening the fiscal crises there.
EDHEC-Risk, in a news release that accompanied O’Kane’s report, expressed the view that the ban on naked CDS’s may be counterproductive; that is, it may cause harm to the banking system by “removing one sovereign risk mitigation tool.”
Causal Links and Radars
O’Kane’s study itself undermines the naked CDS ban from the other direction: not by pointing to its costs but by questioning the reality of its alleged benefits. The ban does some good only if there is a clear causal relationship between the CDS market and the referenced bonds, with the suspect activity in the former creating distortions in the latter. O’Kane cautions against accepting that causal link. His point turns on the difference between statistical (Granger) causality and physical causality.
There is, he writes, “evidence for a Granger causal relationship with a one day lag for CDS and bonds for Greece and Spain, the reverse relationship for France and Italy, and a feedback relationship for Ireland and Portugal.”
Granger causality is a statistical relationship in which the past value of Y improves an observer’s ability to predict a future value for X. Because of the element of a time-lag, this is something more than mere correlation, but it is still considerably less than what O’Kane calls “pure causation in a physical sense,” which requires a “transmission mechanism.” If there is no Granger causality, then it seems a safe inference there is no physical causation either, but the reverse inference does not hold.
Indeed, the relationship between these two markets that O’Kane posits might almost be taken as a paradigm of the difference between Granger causation and physical causation. Consider the case of two distinct radar systems, one better at longer range detection than the other. The superior radar system will detect an oncoming airplane before the inferior system will. Thus, there will be a relationship of Granger causation between the detection of a particular blip on the better system and its detection on the other system. If we see an incoming blip on the better system we will be able to predict that it will soon show up on the inferior system. It doesn’t follow, though, that the one radar is physically causing anything to happen to the other radar. The oncoming airplane is the physical cause of the blips, and the mechanics of the two radar systems are the physical causes of the time lag. That is the relationship that O’Kane posits between the CDS market and the market for the underlying bonds. The deteriorating credit worthiness of the countries issuing those bonds serves as the oncoming airplane in our analogy.
Let us note explicitly the similarities: the CDS and the bond markets involve the same credit risk. Further, at what O’Kane calls a “micro market-structure level,” most dealers have the same traders make markets in both instruments. Thus, they are sensitive to the same approaching planes: “[The] arrival of new information about the reference entity sovereign … will quickly be reflected in the prices of both.”
But the workings of the two radars differ. On this, O’Kane refers us to a study by Sanjiv Das at al., “Did CDS Trading Improve the Market for Corporate Bonds?” (2011) suggesting (in O’Kane’s paraphrase) that “the greater convenience of the CDS market means that price discovery occurs there before it does in the bond market.” One way in which the CDS market, an over the counter market in synthetic assets, is more convenient is simply that it is equally liquid in both directions, in contrast to the bond markets which are more liquid for buyers than for sellers. Also, it is easier to trade in large sizes through the CDS market and easier to use leverage.
In general, CDS is “the preferred instrument for those seeking to implement a short credit position.” This may well have made it in certain circumstances the better radar, and thus may explain the Granger causality for Greece and Spain.