Who Really Benefits When Workers Sign Noncompete Agreements?

Starr (2026) examines whether noncompete clauses benefit or harm workers, firms, and markets. He synthesizes evidence from surveys, administrative datasets, natural experiments, and field experiments, including nationally representative worker surveys and state-level policy changes. The central question is whether noncompetes function as efficient contracts or anticompetitive restraints. He finds that noncompetes are widespread across occupations and typically reduce worker mobility and wages. Causal studies show wage gains of about 3–4% following bans, while a field experiment finds compensation reductions of 4–12%. He concludes that noncompetes broadly suppress wages, innovation, and entrepreneurship while benefiting firms through reduced competition.

Why This Article Was Selected for The Policy Scientist

Noncompete clauses raise a central policy issue concerning how labor markets balance worker mobility with firm investment. Starr (2026), who has contributed extensively to this literature, offers a timely synthesis as the scrutiny of noncompete clauses intensifies at both state and federal levels. The article is important because it situates noncompetes within broader questions of wage determination, innovation, and market structure, rather than treating them as isolated contractual tools  . It extends prior theoretical work by incorporating recent empirical evidence. The underlying data are diverse and generally high quality, drawing from surveys, administrative sources, and policy variation. The strongest findings rely on causal inference and field experiments.

Full Citation and Link to Article

Starr, E. (2026). The economics of noncompete clauses. Journal of Economic Perspectives, 40(1), 139–166.

Central Research Question

Starr (2026) examines a longstanding and increasingly salient policy question: do noncompete clauses function as efficiency-enhancing contractual tools that support firm investment, or do they operate primarily as anticompetitive restraints that suppress labor market competition? More specifically, the article evaluates who signs noncompetes, how they affect worker outcomes such as wages and mobility, whether firms rely on them out of necessity or convenience, and how their use shapes broader economic outcomes including innovation, firm entry, and consumer welfare. The inquiry is not confined to partial equilibrium effects on individual workers or firms, but instead situates noncompetes within a general equilibrium framework in which labor market frictions, firm incentives, and regulatory variation interact. The central objective is to synthesize recent theoretical and empirical evidence to determine whether the prevailing legal and policy regime governing noncompetes is economically justified.

Previous Literature

The article builds on two dominant theoretical frameworks that have structured the literature for decades. The first, often termed the efficient contracting view, posits that noncompetes mitigate hold-up problems by allowing firms to recoup investments in worker training and proprietary knowledge. In this framework, restrictions on worker mobility are compensated through higher wages or enhanced training opportunities, leading to mutual gains for both parties. The second framework, rooted in labor market monopsony and search frictions, views noncompetes as mechanisms that increase employer market power by restricting worker outside options, thereby suppressing wages and reducing allocative efficiency.

Empirically, earlier studies produced mixed findings, particularly those relying on cross-sectional correlations between noncompete use and wages. These studies often suggested that workers subject to noncompetes earned more, though such results were confounded by selection into higher-paying occupations and the simultaneous use of complementary contractual restrictions. More recent contributions have increasingly emphasized causal inference strategies, exploiting variation in state-level enforceability or implementing field experiments. Starr’s article synthesizes this newer body of evidence, while also engaging with adjacent literatures on innovation, entrepreneurship, and industrial organization.

Data

The article draws on a wide range of data sources, reflecting the interdisciplinary nature of the question. These include nationally representative worker surveys, employer surveys, administrative datasets such as the National Longitudinal Survey of Youth, and matched employer-employee data. In addition, the article incorporates evidence from policy changes across states, including bans or restrictions on noncompetes in jurisdictions such as Oregon, Hawaii, and Washington. These natural experiments provide quasi-experimental variation in enforceability that can be leveraged to estimate causal effects.

The data also include occupation-specific studies, such as those examining physicians, engineers, and executives, as well as novel field experimental data in which job offers are randomized to include or exclude noncompete clauses. This diversity of data sources enhances the credibility of the findings by allowing for triangulation across methodologies and contexts. However, limitations remain, particularly with respect to measurement error in self-reported survey data and the external validity of field experiments conducted in specific industries.

Methods

Methodologically, the article places considerable emphasis on distinguishing between correlational and causal evidence. Early observational studies are critiqued for their inability to isolate the independent effect of noncompetes due to confounding factors such as occupational sorting and the bundling of contractual restrictions. In contrast, the most persuasive evidence comes from studies employing causal inference techniques, including difference-in-differences designs exploiting state policy changes, event studies, and field experiments.

The field experimental evidence is particularly notable. By randomizing the inclusion and salience of noncompete clauses in job offers, researchers are able to directly estimate their impact on worker behavior and compensation. This approach addresses many of the identification challenges present in observational data and provides a clearer estimate of causal effects. The article also references structural and general equilibrium models that incorporate labor market frictions and firm behavior, though these are secondary to the empirical evidence. Overall, the methodological approach reflects a hierarchy of evidence in which causal inference and experimental designs are given greater weight.

Findings/Size Effects

The weight of the evidence suggests that noncompete clauses have broadly negative effects on workers and markets. In terms of prevalence, noncompetes are widely used across the labor market, including among low-wage workers, which contradicts the prediction that they are confined to high-skill occupations involving trade secrets. Survey data indicate that approximately 18 percent of workers are bound by noncompetes, with nontrivial incidence even among low-wage employees.

With respect to wages, correlational studies often show a positive association between noncompetes and earnings, but causal evidence indicates the opposite. Natural experiments exploiting state-level bans consistently find that removing noncompetes increases wages by approximately 3 to 4 percent on average. Field experimental evidence provides even more direct estimates, showing that the presence of a noncompete can reduce total compensation by between 4.2 percent and 12 percent, depending on the specification. These effects appear to operate through reduced worker mobility, as noncompetes significantly decrease the likelihood of transitioning to competing firms.

The article also finds that noncompetes reduce entrepreneurship and firm entry. Studies using state-level variation show that stronger enforceability is associated with fewer startups and lower job creation rates. Similarly, innovation outcomes are negatively affected. While noncompetes may increase firm-level investment in intangible assets, they reduce patenting activity, particularly for novel and high-impact inventions. Estimates suggest that stricter enforceability leads to measurable declines in both the quantity and quality of innovation, driven by reduced knowledge spillovers and worker mobility.

Spillover effects further amplify these impacts. Even workers not directly bound by noncompetes experience lower wages and reduced mobility in markets where such clauses are widely used. These general equilibrium effects indicate that the consequences of noncompetes extend beyond individual contracts to shape overall labor market dynamics. Finally, evidence on consumer outcomes suggests that noncompetes contribute to higher prices and reduced service availability through increased market concentration and reduced competition.

Conclusion

Starr (2026) concludes that while noncompetes may generate limited benefits in specific contexts, their overall effect is to function as restraints on labor and product market competition. The evidence indicates that they suppress wages, reduce mobility, hinder innovation, and impede firm entry, with corresponding negative implications for consumers. Alternative contractual and legal mechanisms, such as nondisclosure agreements and trade secret protections, appear sufficient to protect legitimate business interests in most cases.

The article’s synthesis of recent causal evidence represents a significant contribution to the literature, particularly in clarifying the divergence between correlational and causal findings. At the same time, the analysis highlights areas for further research, including the need for additional field experiments across diverse industries and institutional settings. Such work would enhance understanding of the external validity of existing findings and inform the design of policy interventions. Overall, the article provides a comprehensive and methodologically rigorous assessment of a policy issue with substantial implications for labor market performance and economic growth.