Communications Between Borrowers and Servicers: Evidence from the COVID-19 Mortgage Forbearance Program

Abstract

I utilize proprietary servicer call transcripts between a single servicer and the corresponding borrowers, whose loans they service, to shed light on borrower responses to the mortgage forbearance program contained in the CARES Act. My analysis reveals that borrowers (especially with non-performing loans) did not actively seek out mortgage forbearance (conditional on communication) in response to this policy, which was intended to prevent a pandemic-induced foreclosure. This is an outcome of the servicer's differential treatment between government and private loans, as the CARES Act was designed for government loans only and left scope for servicer discretion for private loans. These results bring into question the effectiveness of ad hoc laws and the implementation thereof during the COVID-19 pandemic.

Keywords: Forbearance, Servicer, Machine Learning, Natural Language Processing

JEL Classification: E60, E65, G18, G21, H3