Researcher ORCID Identifier

0009-0009-9970-0763

Graduation Year

2026

Date of Submission

4-2026

Document Type

Open Access Senior Thesis

Degree Name

Bachelor of Arts

Department

Economics

Reader 1

Eric Helland

Rights Information

2026 Josephine V Aspromonte

Abstract

This paper estimates the effect of legal representation on case outcomes using 531,959 limited-jurisdiction debt-collection cases filed in the Los Angeles Superior Court between 2012 and 2022. Defendants are classified into three groups: those who defaulted without responding, those who responded pro se, and those who retained counsel. Because defendants are not randomly assigned to representation, three estimation strategies address selection bias: a binary logit, a multinomial logit with text-mined docket covariates, and propensity score matching on pre-treatment characteristics.Three findings emerge. First, the binary logit associates representation with a 10 to 13 percentage point reduction in default probability, but matching reduces that estimate to 4.7 percentage points. Raw default rates for pro se and represented defendants are nearly identical (26.52 and 25.72 percent), indicating that the decisive margin is participation, not representation. Second, matched represented defendants pay approximately $28 more than comparable pro se defendants, driven by higher interest, filing fees, and attorney fees. Third, a random forest analysis shows that the features most predictive of representation—longer dockets, more motions, more continuances—are post-treatment, generated by the lawyer rather than predicting which cases retain one. The primary barrier defendants face is participation rather than representation. Effective policy reform is more likely to come from reducing barriers to responding, such as high answer filing fees, procedural complexity, and inadequate notice, than from expanding access to counsel alone.

Download

Included in

Civil Law Commons

Share

COinS