Energy derivative markets and systemic risk

This intermediary report aims at presenting our researches on systemic risk, from April 2010 to January 2011. Since this work is in the direct prolongation of a previous contract on the same subject (CFE Contract number 58) we decided to integrate it with the previous report, written in March 2010. This results in quite an important document (more than 150 pages) where at least 50 pages are entirely new. Such a choice insures the global coherence of the document: there is no need, for example to present the data and the problematics. This does not mean, however, that we did not enhance our database. In order to give evidence of what was done during the last 10 months, we put the recent work, as often as possible, in chapters that are entirely new. Compared with what was done in the previous report, the reader will find, essentially, three new chapters (i.e. Chapters 3, 5 and 6) as well as a conclusion (Chapter 9) and some supplements in Chapter 8. All these chapters are summarized in the executive summary that follows this addendum. Since April 2010 our attention was focused, firstly on the prolongation of the empirical analysis initiated previously (Chapters 3, 5 and 6), and secondly on a reflection on the way we could built a model of systemic risk in derivative markets (Chapter 8). We first undertook an analysis of the behavior of futures prices in the maturity dimension, especially as regards to extreme deviations. The results of this analysis can be found in Chapter 3. Part of this chapter has just been accepted for publication in the review Physica A and will be presented at an international conference on econophysics in Shanghai, in June 2011. We then examined the causality relationships among derivative markets. In the previous report, an appropriate methodology for understanding systemic risk, graph theory, has indeed been identified. By using this approach, and more specifically by relying on Minimum Spanning Trees, we have been able to show the pathways by which price shock waves can be transmitted. Now, we are looking for a method for directed graphs, which would allow us to study the direction in which price fluctuations are propagated. Two kinds of methods are investigated: the Granger methodology (Chapter 6) and the concept of conditional entropy (Chapter 7). Finally, an important issue of the work undertaken almost two years ago is to allow for the elaboration of a model representing the links connecting the derivative markets. The usefulness of such a model will be to highlight the mechanisms that are responsible for markets integration and their implication in prices moves. Moreover, such a model will allow for the study of diverse shock scenarios for the prices path propagation and for the intensity of this transmission. On the basis of our empirical findings, we progressed towards the construction of such a model, as can be seen in Chapter 8. (authors)