Legal aid eligibility and court outcomes: a design-based double-machine-learning approach

Legal aid eligibility and court outcomes: a design-based double-machine-learning approach

Source: econ.GN updates on arXiv.org — https://arxiv.org/abs/2608.05211 Date read: 2026-09-02 Connected to: L-004, seed-018 Kind: meta Escalation: store-only Escalation rationale:

What this is

An econometric causal inference study using double machine learning to estimate the effect of legal aid eligibility (via means-testing) on court outcomes in New South Wales. The paper treats means-test assignment as a natural experiment to isolate treatment effects of private vs. public legal representation on acquittal/conviction rates.

What I took from it

This is a case study application of causal inference method to a specific policy domain, not a sustained theoretical or empirical argument about how protocols behave under formalization. The means test itself is a proxy measure (ability to pay as surrogate for "legal indigence"), but the paper does not examine what happens when that proxy becomes the optimization target — i.e., whether defendants or courts learn to game the eligibility threshold, whether the formalization of "indigence" shifts behavior upstream, or whether the metric capture dynamics predicted by L-004 appear in this domain.

The paper measures an outcome (acquittal rates by eligibility status) but does not investigate the mechanism of metric capture — whether the legibility of the means test as a decision rule creates incentive distortions, strategic claiming behavior, or institutional gaming. It is descriptive causal inference, not a study of how formalization itself alters protocol behavior.

Research connections

• L-004: The means test is a proxy for legal need, but this paper does not examine whether the proxy becomes an optimization target or whether its use as a computable eligibility rule triggers capture.
• seed-018: Mentioned in triage; likely concerns responsibility implication (who bears cost/blame when means test denies aid?), not mechanism of formalization effects.

Method note

This represents sound econometric practice within its frame: causal inference on policy effects via administrative data linkage. However, it exemplifies a common pattern in policy evaluation research — measuring outcome differences by policy regime rather than studying how the policy rule itself alters agent behavior and institutional structure. For the new nature agenda, we need work that traces the causal pathway through formalization, not just the final outcome gap. This suggests meta-need: policy evaluation papers should be screened for whether they study policy effects (outcome differences) or policy mechanisms (how formalization restructures incentives and eligibility gaming).