Loading evaluation data…
Strengths
The paper documents how noncompete enforcement reshapes internal promotions, fine-grained occupational reallocation, and hierarchy, a margin largely absent from prior NCA research that focused on external mobility and wages.
↳ Introduction pp. 1-5; Sections 5-8
The design requires control workers to be in establishments and parent firms in states with no enforceability change, addressing treatment spillover, and event-study estimates show no differential pre-trends supporting the parallel-trends assumption.
↳ Section 4.1; Figure 1
The firm-level analysis shows increased assets and SG&A per employee but no ROA improvement and a declining Tobin's Q, and the paper reports these null and negative outcomes rather than suppressing them.
↳ Table 11; Section 8
Limitations
The paper invokes Baker et al. (2022) and Goodman-Bacon (2021) but retains two-way fixed effects as its baseline without showing modern staggered-DiD estimators, the number and timing of treatment events, or cohort-specific effects.
↳ Section 4.2 p. 14; Appendix Table A6
Core findings rest on proprietary Revelio Labs/LinkedIn data that cannot be independently inspected, with no code or data released and limited validation of the title-standardization and seniority coding, which constrains verifiability.
↳ Sections 3.1-3.2
Non-random selection into LinkedIn data toward white-collar workers and the possibility that enforcement events coincide with other state-level policy changes are not formally addressed beyond the pre-trends test.
↳ Sections 3.1-3.3; Section 4.2
Evaluations converged on a clear, well-organized paper that opens a genuinely understudied question about how noncompete enforcement reshapes firms internally, supported by a thoughtful clean-control design and clean event-study pre-trends. The widest divergence concerned execution: the analysis retains a two-way fixed-effects baseline without alternative staggered-DiD estimators or cohort-specific effects despite explicitly citing the relevant bias, and it does not address non-random selection into the underlying LinkedIn data or concurrent state-level policy confounders. The honest reporting of null firm-value outcomes (no ROA gain, declining Tobin's Q in Table 11) strengthens credibility, while the proprietary single-source data and absence of any code or data release limit independent verification. The contribution and clarity are solid; the binding constraints are on the causal interpretation and the reproducibility of the evidence base.
Moderate3.6/5.0
The score reflects a genuine and well-motivated extension of the NCA literature from external labor-market effects to firms' internal allocation of human capital, using granular worker-level data and a coherent two-channel framework. It is held below the top band because the substitution and hold-up channels cannot be cleanly separated empirically and adjacent internal-labor-market work makes the advance partly incremental.
NCAs shape the organizational structures through which firms allocate and develop human capital
The staggered state-level design with a clean control group, employee-firm fixed effects, and clean event-study pre-trends is competent, but the score reflects the retention of TWFE as the baseline without alternative estimators or cohort-specific effects despite the paper invoking Baker et al. (2022), plus unaddressed non-random LinkedIn selection and no formal sensitivity analyses for unmeasured confounding. Reliability is Amber, so no Red cap applies.
Standard TWFE estimates can be biased if already-treated observations are used as controls
The score reflects a logically organized paper with a consistent substitution-versus-hold-up narrative spine and explicit variable definitions, where tables align with narrative claims. Minor deductions stem from a roadmap that omits Sections 6-8 and occasional label/sample inconsistencies requiring reader effort.
Section 5 reports the main empirical results. Section 9 concludes
The score reflects substantive positioning across three literature strands supported by a comparative appendix table, going beyond perfunctory citation. It is held at mid-range because the limitations discussion is less developed than the literature engagement, particularly on platform-data representativeness, concurrent state policy confounders, and the absence of a declared reporting standard.
this literature focuses almost entirely on demand-side drivers
Lower confidence on Methodological Rigour, Positioning — domain match limited.
Watch-outs4 of 4 checks
Core results are consistently described across abstract, results, and conclusion, but several minor issues affect interpretability: a rounding discrepancy between the introduction (0.11 pp) and Table 3 (0.104 pp), a roadmap that omits Sections 6-8, variable-label inconsistencies across tables and appendix, and a tension between excluding financial firms in sample construction yet classifying finance and insurance as NCA-sensitive in firm-level analysis. Causal phrasing in framing passages occasionally outruns what an observational design supports.
Core findings rest on proprietary Revelio Labs/LinkedIn data that cannot be independently inspected, with no code or data released and no documented validation of the title-standardization and seniority coding. No ethics, conflict-of-interest, preregistration, data-availability, or code-availability statements are present. These are structural features and reporting gaps rather than indicators of misconduct; ethics approval is not typically required for commercially licensed observational employment data.
Flags: 0 declared / 5 total
All 0 cited references resolved to real publications via PubMed, Crossref, or OpenAlex.
No retraction notice found in Retraction Watch.
Sources: Retraction Watch ✓PubPeer (coming soon)
Medium3.6/5.0
The score reflects direct relevance to an active policy debate over noncompete regulation with identifiable stakeholders including regulators, labor economists, and large employers. It stops short of the top band because the paper offers no concrete implementation guidance for any single actor.
the FTC's proposed noncompete rule
The score reflects findings estimated from real organizational behavior across a large multi-year panel of actual employment histories, which sits above proof-of-concept. It is held at mid-range because the evidence is not deployment-ready operational guidance and is bounded by sample restrictions.
The score reflects limits on generalizability from the public-firm restriction, LinkedIn coverage skewed toward visible white-collar workers, and a US-specific legal setting. Heterogeneity analyses by branch size, function, and worker type provide some boundary-condition evidence but also show effects are non-uniform.
The score reflects a cumulative contribution that fills a specific gap on within-firm adjustment and connects to an active NCA evidence base, plausibly informing future syntheses and policy debate. It is held below the top band because observational evidence of this type shapes rather than settles the debate.
Lower confidence on Readiness, Generalizability — domain match limited.
Nabu’s assessment, alongside the field’s view.
Evaluation Metadata
AI-powered, human-governed. Something look off? Contact us to request a review.