Answers are grounded in this paper and cite their sources.
Mortgage servicers decide who receives loss mitigation. The record of how they decide is rarely observable. This paper uses a servicer's own call-center transcripts, read by natural language processing, to recover what was actually said to 19{,}418 borrowers seeking pandemic relief, and pairs that record with the legal boundary the CARES Act drew through the servicer's book: forbearance was an entitlement for federally backed borrowers on attestation alone, and a discretionary decision for everyone else. Discretion turns out to have operated through documentation, verification, and referral rather than denial. Relief followed borrowers who called in; servicer-initiated calls track the foreclosure moratorium, not forbearance. Statutory eligibility raised forbearance by 3.6 percentage points (s.e.\0.26) across 96{,}244 loan-months --- an estimate of what a documentation requirement costs in take-up, and a quantity directly at issue in current rulemaking to streamline loss mitigation. Racial differences are small and imprecisely signed on both sides of the boundary. A triple-difference estimate rules out differential gaps above roughly one and a half percentage points, though smaller ones cannot be excluded. In this portfolio a documentation requirement was a barrier of uniform height: it suppressed take-up among everyone subject to it.
No takes yet. Share an insight, caveat, or question.
Arka Prava Bandyopadhyay (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: