Working Paper: CEPR ID: DP11145
Authors: Tobias Broer
Abstract: The early 2000s have seen an enormous boom and bust in structured financial products, such as residential mortgage-backed securities (RMBSs) or collateralised debt obligations (CDOs). The standard 'Gaussian Copula' model used to quantify their credit risk was highly dependent on the choice of a single default correlation parameter that often required subjective judgement, as underlying assets were not standardised or only had a short history. This paper shows how moderate disagreement about default correlation increases the market value of the structured collateral considerably above that of its total cash-flow, as investors self-select into buying tranches they value more highly than others. The implied 'return to tranching' is sizeable for a typical RMBS, and an order of magnitude larger for CDOs backed by RMBS-tranches, whose cash-flow distribution is not bounded by a minimum recovery value and thus more sensitive to heterogeneous default correlations. In contrast, disagreement about average default probabilities, or recovery values, does not imply a large return to tranching.
Keywords: CDO; RMBS; Disagreement; Default Correlation; Credit Risk; Great Recession; Housing Bubble
JEL Codes: D82; D83; E44; G12; G14
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
Cause | Effect |
---|---|
disagreement about default correlations (C10) | market value of structured finance products (G19) |
disagreement about default correlations (C10) | prices of structured products (G19) |
perceived low default correlation (G33) | valuations of junior tranches (G19) |
perceived high default correlation (G33) | valuations of senior tranches (G12) |
disagreement about default correlations (C10) | returns to tranching (C24) |
disagreement about default correlations (C10) | prices rise (E31) |