Securitization, loan sales and distance

This thesis contributes to the empirical literature dealing with the analysis of loan sales and securitization of U.S. commercial banks and thrifts. First, we investigate how to measure securitization using bank-level data from U.S. commercial banks, by investigating whether there is a set of unobserved, common factors that explains the interrelationships among securitization proxies. Our analysis identifies a single securitization factor. Eight out of eleven proxies measure this factor significantly and reliably. Second, we examine the question of what it makes that some banks securitize their assets while others don't. We find that securitization is only attractive for banks with more than $1 bln of total assets, and that banks with non-traditional activities (such as securities brokerage and investment banking) are more likely to start securitizing their assets than banks without these activities. Third, we study to what extent the impact of distance on lenders' loan origination decisions in U.S. residential mortgage lending has decreased due to the adoption of new lending technologies. We find that distance continues to reduce the likelihood of mortgage loan origination, but that the impact of distance has decreased over time. Our results suggest that the use of credit score modeling by lenders is an important driver of this decrease. Last, we assess whether U.S. residential mortgage lenders use loan sales to lend at greater distances. We find that loan sales are associated with a 179% higher expected lending distance, and that loan sales allow lenders to reduce their loan rates.

Dissertation: http://hdl.handle.net/11370/c56a1010-52e3-459a-8285-74df82d78223