Title
The Information Content of Market Prices: the Case of Mortgage-Backed Securities
Abstract
Because diversification is the key tenet of securitization, loan default correlation is a key risk to the holder of a security subject to default risk. Using loan performance data to infer beliefs about the underlying probability of default, loss given default and prepayment speed, this study estimates a single factor Gaussian copula model to elicit beliefs about default correlations from the market price of residential mortgage-backed securities (RMBS). Default correlations increase over the crisis, but the increase is moderate compared to what the literature records for collateralized debt obligation (CDO) markets.
Through the implied correlations we assess how much information is contained in prices regarding future bond outcomes. While confirming earlier claims that bond prices exhibit private information regarding future downgrades, the results suggest differential knowledge is less between AAA investors and the rest than between alt-A investors and others (e.g. prime and subprime). Using a loan documentation completeness index aggregated at deal level, we confirm that the difference is explained by loan opacity rather than other differing aspects of alt-A bonds.
Using a novel dataset containing the panel of trading prices and agency ratings on private label residential mortgage-backed securities, this study tracks the presence of private information over the life of the bond. It finds that investors continue to produce private information over time, the exception being alt-A investors for whom, again, no evidence of learning is found. Overall, the results suggest due diligence about the underlying assets is an important friction in RMBS markets regardless of investor sophistication.