Modeling US bank CDS spreads during the Global Financial Crisis with a deferred filtration pricing model

This paper uses a simplified version of the Duffie and Lando (2002) deferred filtration model to handle the effect of asymmetric information about US banks asset portfolios during the recent crisis, when banks were reluctant to lend to one another because they were not sure about the balance sheet strength of the counterparty and its ability to repay. The accounting lag in the deferred filtration model gives a very useful way of calibrating this uncertainty and provides convenient closed forms suitable for econometric models. I use these to model the default probabilities implied by CDS rates. Comparing the fit of this model with that of the standard full information structural defaultable debt pricing model strongly supports the hypothesis that investors were wary of the value of accounting information. The performance of the model is comparable to that of a benchmark reduced form hazard rate model, the workforce of the empirical literature to date.