Are All Credit Default Swap Databases Equal?

Abstract

The presence of different prices in different databases for the same securities can impair the comparability of research efforts and seriously damage the management decisions based upon such research. In this study we compare the six major sources of corporate Credit Default Swap prices: GFI, Fenics, Reuters EOD, CMA, Markit and JP Morgan, using the most liquid single name 5-year CDS of the components of the leading market indexes, iTraxx (European firms) and CDX (US firms) for the period from 2004 to 2010. We find systematic differences between the data sets implying that deviations from the common trend among prices in the different databases are not purely random but are explained by idiosyncratic factors as well as liquidity, global risk and other trading factors. The lower is the amount of transaction prices available the higher is the deviation among databases. Our results suggest that the CMA database quotes lead the price discovery process in comparison with the quotes provided by other databases. Several robustness tests confirm these results.

Detecting abnormal changes in credit default swap spreads using matching-portfolio models

“…These differences in data source can have a material impact on spread measurement. Qiu and Yu (2012) , for example, show that the number of sell-side quote providers for a single-name CDS in Markit is positively related to the amount of informed trading; while Mayordomo et al (2013) provide evidence that systematic differences exist between spreads obtained from different data providers. As a robustness check, we therefore also use CMA data for our analyses.…”

“…CDS spreads reported by different data sources can differ substantially and systematically ( Mayordomo et al, 2013 ). It is therefore important to address the extent to which the conclusions reached based on Markit data can be extended to different databases.…”

“…For this robustness check we use CMA data (retrieved from Datastream), which is the second most commonly used source for CDS-based event studies (see Table 1 ). Mayordomo et al (2013) find that CMA data lead the price discovery process, making them a particularly appealing alternative to Markit data for event studies. CMA data are available via Datastream for the period from January 2004 to September 2010.…”

“…To better appreciate differences in the size and power of statistical tests before and during the crisis, we split the dataset into Pre-crisis (408,411 observations) and Crisis (457,387 observations) samples. In line with Mayordomo et al (2013) , we set the beginning of the crisis on August 1, 2007. 12 In general, we expect a significant deterioration in the size and power of statistical tests during the crisis, as volatility, kurtosis, and (absolute) skewness of daily spread changes increase.…”

“…Mayordomo et al (2013) set the beginning of the crisis using aBai and Perron (2003) test to identify a structural break in the time series of corporate CDS spreads.…”