The Asian Institute of Chartered Bankers has just published “A Best Practice Approach to Modeling Sovereign Defaults” in the December 2014 issue of its Banking Insight magazine. Since the article, which I co-authored with my colleagues Suresh Sankaran and Dr. Clement Ooi, was targetted toward the Asia market, it is helpful to emphasize some of the most important points in modeling sovereign default risk from a world-wide perspective. There are three key points in modeling sovereign default risk that we explain in the rest of this article:
- The credit default swap market is a very problematic source of credit information and, at best, it is reliable only for a short list of reference names.
- The conflict of interest faced by legacy rating agencies is even more extreme in the sovereign case than it is in the well-documented corporate and structured products markets.
- Modern statistical modeling techniques are best practice and the only realistic alternative to the credit default swap market and legacy credit ratings.
We outline the reasons for these assertions in the rest of this note.
The Single Name Credit Default Swap Market on Sovereign Reference Names
Kamakura Corporation has been publishing a detailed analysis of trading volume in the single name credit default swap market for sovereign reference names for a number of years. The most recent report covered the period from the week ended July 16, 2010 through the week ended June 27, 2014. While trading in sovereigns is the strength of the single name credit default swap market, that market is an insufficient source of credit information on the majority of sovereigns for four reasons. First, trading volume is high only on a very short list of sovereigns. The chart below shows the non-dealer trading volume in the period from July 16, 2010 through December 26, 2014:
Only four reference names average more than 10 non-dealer trades per day. Only 12 reference names average more than 5 non-dealer trades per day. Only 60 sovereign reference names have traded since the week ended July 16, 2010. Assessing the credit risk of the remaining sovereigns using the CDS market is impossible when there are zero trades in those reference names.
Secondly, there is an active concern about manipulation in the credit default swap market, in part because of the thinness of trading. Third, implying default probabilities from credit default swap spreads is fraught with difficulties as explained in a recent article by my colleague Prof. Robert Jarrow. He compares using credit default swap spreads to infer default risk to estimating your own mortality from the price you are quoted for life insurance from an agent who may be your brother-in-law. Fourth, the number of formal sovereign CDS “credit events” associated with sovereign reference names is much too small for statistical analysis to have any validity. The list of formal credit default swap credit events upon which ISDA has rendered an opinion is listed on the ISDA website. Events for Europe, the Middle East and Africa are listed here. The number of sovereigns is far too small to build a valid statistical link between the credit default swap spread and actual default events. Recent names upon which ISDA has opined include Greece, Ecuador, Argentina, and Ireland (ruled to be a non-default). Most default modelers would be concerned if the number of defaulters in a model (in this case, the number of sovereign defaults upon which CDS were traded at the time) is less than 50. We are very far short of that modeling threshhold. The simple rule that the credit spread equals (one minus the recovery rate) times default probability, often referred to in the popular press, is simply false as documented here .
Legacy Rating Agencies and Sovereign Default Risk
Legacy rating agencies are even more problematic as sources of credit quality information on sovereigns. The first concern is a lack of accuracy, as documented by Hilscher and Wilson (2013) in the context of corporate bond issuers. A second major concern is the conflict of interest inherent in the “issuer pays” business model of the legacy credit ratings, documented in more than 150 pages by the U.S. Senate in the aftermath of the credit crisis. This conflict is even more complex due to the ability of a sovereign to punish a rating agency for a rating that the sovereign disagrees with. Even Standard & Poor’s has made this allegation after the firm was sued by the U.S. government not long after S&P downgraded the United States to AA+. The other problems with ratings are well-known: lack of an explicit maturity, lack of timely ratings changes, lack of an explicit default probability, instability of historic default rates, and crudeness of a 20 grade rating system instead of the 10,000 grades in a default probability that runs from 0.00% to 100.00%. The problems with ratings are so severe that the Bank for International Settlements recently proposed reducing the use of ratings in the BIS capital regulations.
A Modern Statistical Approach to Assessing Sovereign Default
Kamakura Corporation launched the world’s first quantitative default probability model for sovereigns in 2008 under the sponsorship of one of the world’s largest insurance companies. The Kamakura Risk Information Services Sovereign Default Service covers all of the world’s important sovereigns. Default probabilities are updated daily and include an annualized and cumulative default probability term structure out to 5 years. The paper by van Deventer, Sankaran and Ooi gives a number of examples of how the Kamakura Risk Information Service default probabilities for sovereigns are used in practice.
The KRIS Sovereign Default Service uses modern reduced form default probabilities like those in Kamakura’s well-respected public firm models. For background on the role of macro-economic factors in sovereign credit risk, see Kamakura’s senior research fellow Prof. Jens Hilscher’s paper with Yves Nosbusch. Kamakura Corporation’s management team has published an introduction to sovereign risk assessment and comparisons with corporate default probabilities . For more information on Kamakura’s Sovereign Default Service, please contact firstname.lastname@example.org.
Copyright ©2015 Donald van Deventer