Search

This paper contributes to the general understanding of how sovereign CDS prices are formed by studying the information content of pricing errors generated by a non-arbitrage model. We implement a price-discrepancy measure in the spirit of the noise measure introduced by Hu et al. (2013) in the Treasury Bond market, and analyze its main determinants in panel data analysis. The main results show that sovereign CDS pricing errors are systematically related to higher bid-ask spreads. The evidence in this paper also suggests that exits of capital arbitrage during distressed periods, as measured by changes in net offsetting, can be associated to larger pricing errors in sovereign CDS from advanced economies, thereby supporting the main claims of the limit-to-arbitrage theories. These findings are robust for the most common CDS pricing models employed in the industry and different estimation techniques. Antonio Rubia acknowledges financial support from the Spanish Ministry of Economy and Competitiveness (MEC) grant ECO2012-33619 and ECO2014-58434-P. Pedro Serrano acknowledges financial support from MEC grant ECO2012-34268 and from Junta de Andalucía project P12-SEJ-1733.