On January 4, we showed that on average 81.68% of trades in credit default swaps in the trade warehouse of the Depository Trust & Clearing Corporation were trades between dealers. In today’s blog, the second in the series, we look at weekly CDS trade data since the week ended July 16, 2010 for 1,090 reference names and find minimal “end user” trade volume for the overwhelming majority of those reference names.
The first blog in this series analyzed the degree to which trading in single name credit default swaps was trading between two dealers as defined by the Depository Trust & Clearing Corporation.
van Deventer, Donald R. “Collusion and CDS Dealer Volume,” Kamakura blog, www.kamakuraco.com, January 4, 2012.
The data showed that, for 77 weeks beginning with the week ended July 16, 2010 through December 30, 2011, on average 81.68% of the “live” single name credit default swaps in the DTCC trade warehouse were trades between two dealers. We concluded that market participants who assumed there was no collusion in the market for both traded single name CDS and quoted single name CDS were at great risk.
In this blog, we analyze CDS trading volume for the 1,090 reference names for which CDS trades were reported by DTCC during the same 77 week period. The weekly trade information is from the Section IV reports from DTCC. The data is described this way in the DTCC document “Explanation of Trade Information Warehouse Data” (May, 2011):
“Section IV (Weekly Transaction Activity) provides weekly activity where market participants were engaging in market risk transfer activity. The transaction types include new trades between two parties, a termination of an existing transaction, or the assignment of an existing transaction to a third party. Section IV excludes transactions which did not result in a change in the market risk position of the market participants, and are not market activity. For example, central counterparty clearing, and portfolio compression both terminate existing transactions and re-book new transactions or amend existing transactions. These transactions still maintain the same risk profile and consequently are not included as ‘market risk transfer activity.’”
Our emphasis is not on gross trading volume. As mentioned above, dealer-dealer volume is 81.68% in the single name credit default swap market and it would be nearly costless for dealers to inflate gross trading volume by trading among themselves. Instead, we focus on “end user” trading where at least one of the parties to a trade is not a dealer. Accordingly, we make the following adjustments to the weekly number of trades reported by DTCC for each reference name:
- We divide each weekly number of trades by 5 to convert volume to an average daily volume for that week
- From that gross daily average number of trades, we classify 81.68% of trades as “dealer-dealer” trades, using the average “dealer-dealer” share of trades in the DTCC trade warehouse during the 77 week period studied.
- The remaining 18.32% is classified as daily average “non-dealer” volume, the focus of the reporting below.
Daily Non-Dealer Trading Volume for 1,090 Reference Names
Before reporting on the 1,090 reference names in the DTCC warehouse, it is important to note a very important fact. Kamakura Risk Information Services’ default probability service produces daily default probabilities on 30,764 public firms around the world, including 5,222 in the United States. The DTCC weekly reports were designed to report the top 1,000 reference names ranked by daily trading volume. In none of the 77 weeks, however, were there as many as 1,000 reference names reported by DTCC. In essence, the weekly reports contain every reference name in which there was non-dealer trading volume in the DTCC. For all other reference names, then, trading volume was zero. This means that more than 29,600 of the public firms in KRIS had zero credit default swap trading volume over the entire 77 week period studied. We now turn to the 1,090 reference names for which there was at least 1 credit default swap trade during the 77 week period.
We first analyzed daily average non-dealer CDS trading volume by calculating the daily average, as described above, for all 1,090 reference names for the entire 77 week period. The distribution of the reference names by non-dealer daily average trading volume contains some very surprising results:
Among the surprising conclusions are these facts:
- 76.2% of the reference names had 1 or fewer non-dealer contracts traded per day.
- 93.6% of reference names had 2 or fewer non-dealer contracts traded per day.
- 99.0% of reference names had 4 or fewer non-dealer contracts traded per day.
- Only 5 reference names averaged more than 5 non-dealer contracts traded per day.
- Only 1 reference name averaged more than 10 non-dealer contracts traded per day.
These trading volumes are miniscule to say the least, and it is astonishing that financial journalists have not made the appropriate disclaimers when quoting “CDS prices” (many of which are quotes, not trades) for specific reference names.
In order to better understand the data, we next analyzed each week of data for all 1,090 reference names. In aggregate, there should be 77 x 1,090 = 83,930 “reference name-weeks” of data reported, but there was no data reported on 14,632 occasions because there were no trades in that week for that reference name. Dividing these 14,632 observations by 1,090 reference names shows that, on average over the 77 week period, there were 13 weeks for each reference name in which there were zero CDS trades. The daily average non-dealer trade volumes for these 83,930 observations have the following distribution:
Again, this analysis allows us to draw some more surprising conclusions about all weekly trading in single name CDS for all 1,090 reference names, a total of 83,930 observations:
- 17.43% of the observations showed zero non-dealer daily average CDS trading
- 77.48% of the observations showed less than 1 daily average non-dealer CDS contract traded
- 90.64% of the observations showed less than 2 daily average non-dealer CDS contracts traded
- 99.17% of the observations showed less than 6 daily average non-dealer CDS contracts traded
- There were only 10 observations of the 83,930 where more than 28 daily average non-dealer CDS contracts traded
“You agree to treat any Report containing data on a specific entity, rather than aggregate position or transaction activity or other aggregate data, as confidential, or, if you are a regulator or governmental entity, in accordance with any statutory confidentiality requirements applicable to you.”
After reviewing the non-dealer trading volume reference name by reference name, we repeat the warning from our January 4, 2012 blog: a sophisticated observer should assume that both traded CDS spreads and quoted spreads are highly likely to have been affected by collusion. Any other assumption could be hazardous to your wallet.
Donald R. van Deventer
Honolulu, January 9, 2012