Information in Financial Contracts : Evidence from CMBS Pooling and Servicing Agreements

We introduce a novel application of machine learning to compare the Pooling and Servicing Agreements (PSA) that govern asset-backed securities. The PSA is often viewed as mostly boilerplate legal text and thus may appear similar across deals despite heterogeneity in the underlying collateral and deal structures. Our empirical setting is PSAs governing commercial mortgage-backed securities (CMBS). After documenting variation in PSAs, we find that underwriters tailor the PSAs to reflect the differences in the collateral pool as well as the macroeconomic environment prevailing at security origination