Do Noncompete Agreements Reduce Output and Worker Earnings?

Chang et al. (2026) examine whether stronger enforcement of noncompete agreements affects worker productivity and economic output. They use a new state-level manufacturing dataset covering all U.S. states from 1987 to 2021, supplemented with broader state-level economic data. Using a stacked difference-in-differences design, they find that stronger noncompete enforcement reduces manufacturing productivity and output. A 10% increase in enforceability is associated with a 15.0% decline in manufacturing value added and an 8.1% decline in value added per worker. Earnings fall about 0.6%, while employment and labor’s share of income show little measurable change.

Why This Article Was Selected for The Policy Scientist

This article addresses an important labor-market policy question because noncompete agreements affect not only worker mobility and wages, but also productivity, firm investment, innovation, and the allocation of labor across the economy. The issue is especially timely given recent federal and state efforts to reconsider the enforceability of these agreements. Several of the authors, particularly Johnson, Lavetti, and Lipsitz, have already contributed substantially to this literature. The article advances that work by introducing a strong new state-level manufacturing dataset and focusing directly on productivity. Its stacked difference-in-differences design provides a credible causal framework, which I view as a major methodological strength. The dataset is unusually broad, although the large manufacturing effects and weaker economy-wide effects suggest caution in generalizing across sectors or jurisdictions. Publication in AEA Papers and Proceedings, a prominent American Economic Association outlet, increases the article’s visibility and importance within economics.

Full Citation and Link to Article

Chang, K., Johnson, M., Lavetti, K., Lipsitz, M., & Raval, D. (2026). The effect of noncompete enforceability on productivity: Evidence from a new state-level manufacturing dataset. AEA Papers and Proceedings, 116, 256–261. https://www.aeaweb.org/articles?id=10.1257%2Fpandp.20261075

Central Research Question

The central question is whether stronger legal enforcement of noncompete agreements affects economic productivity. Noncompete agreements restrict workers from accepting jobs with competing firms after leaving an employer, but states differ considerably in how willing courts are to enforce these contracts. The authors focus particularly on whether greater enforceability changes manufacturing output, productivity, earnings, employment, capital investment, and labor’s share of income. This distinction is important because prior research has consistently found that stronger noncompete enforcement reduces worker earnings, but lower earnings can arise through at least two different mechanisms. Firms may become less productive and therefore generate less economic surplus to distribute to workers, or workers may simply possess less bargaining power because noncompete agreements restrict their outside employment options. The study attempts to distinguish between these explanations by examining productivity directly.

Previous Literature

The article builds on a substantial literature examining how noncompete agreements influence labor markets, investment, innovation, and worker mobility. Johnson, Lavetti, and Lipsitz (2025) provide particularly important prior evidence, finding that stronger restrictions on worker mobility reduce earnings. Lipsitz and Starr (2022) similarly show that greater enforceability of noncompete agreements adversely affects low-wage workers, while Balasubramanian et al. (2022) examine the career consequences of enforceability among high-technology workers. Together, these studies establish that noncompete policy can have measurable effects on worker compensation and mobility.

The theoretical literature provides competing predictions about productivity. Garmaise (2011) argues that noncompete enforcement can reduce worker incentives to exert effort or invest in their own skills. Greater enforceability may also interfere with efficient matches between workers and firms by making it more difficult for employees to move to employers where their skills are more productive. Johnson, Lavetti, and Lipsitz (2025) and Gottfries and Jarosch (2023) develop related arguments about worker mobility and labor-market matching.

Other research emphasizes innovation and knowledge transmission. Marx (2011) documents how noncompete agreements restrict the mobility of technical professionals. Johnson, Lipsitz, and Pei (2023), Lipsitz and Tremblay (2024), and Reinmuth and Rockall (2023) examine mechanisms through which worker mobility can contribute to innovation, entrepreneurship, and knowledge spillovers.

The literature also identifies a potential productivity benefit. Stronger enforcement may encourage firms to invest more heavily in worker training, human capital, and other intangible assets because employees are less able to leave after receiving those investments. Starr (2019), Shi (2023), and Jeffers (2024) provide important contributions to this line of research. The empirical effect of noncompete enforcement on productivity therefore remains theoretically ambiguous, motivating the present analysis.

Data

The authors construct a new state-year panel of manufacturing production covering all U.S. states from 1987 through 2021. The dataset combines historical records from the Annual Survey of Manufactures and Census of Manufactures. For 1987 through 1996, detailed production information is available using three-digit Standard Industrial Classification categories, while the data from 1997 onward use four-digit North American Industry Classification System categories.

The manufacturing records contain several measures of economic activity. Output is measured through sales and value added. Inputs include capital investment, payroll, and employment, with employment and payroll separately identified for production and nonproduction workers. The authors also construct estimates of capital stocks using the perpetual inventory method employed by Chirinko and Wilson (2009). Monetary measures are converted into constant 1997 dollars.

To determine whether the manufacturing findings extend to the wider economy, the analysis supplements these records with state-level data from the Bureau of Economic Analysis and Bureau of Labor Statistics. These data cover GDP, employment, and earnings across broad industrial sectors from 1975 through 2023. Because the broader data lack comparable measures of capital inputs, the economy-wide analysis concentrates on output, productivity, employment, earnings, and labor shares.

The principal explanatory variable is an index of state noncompete enforceability developed by Johnson, Lavetti, and Lipsitz (2025). The index measures seven dimensions of state law and judicial treatment of noncompete agreements from 1991 through 2014. The authors rescale the measure from zero, representing unenforceable noncompetes, to one, representing the strongest enforcement.

Methods

The study uses a stacked difference-in-differences research design based on the approach developed by Cengiz et al. (2019). This method exploits changes in state noncompete law and judicial precedent over time. Each change in enforceability is treated as a separate subexperiment. A state experiencing a change is compared with states that did not experience changes in enforceability during the relevant period.

To reduce contamination from closely spaced policy changes, the authors restrict the analysis to events in which the affected state experiences no other enforceability change during the four years before or four years after the event. This structure permits comparison of outcomes before and after a change while providing a relatively clean control group.

The manufacturing analysis is conducted at the state-year-industry-subexperiment level and incorporates several sets of fixed effects. These controls account for persistent differences among states, common changes affecting broad industries over time, and stable differences among detailed industries. Standard errors are clustered at the state-subexperiment level, and observations are weighted using employment before the policy change.

Most outcomes are estimated using stacked Poisson pseudo-maximum likelihood models. Labor’s share of income is estimated using ordinary least squares. The difference-in-differences structure is intended to provide causal evidence by using within-state legal changes rather than simple cross-sectional comparisons among states with different noncompete laws.

Findings/Size Effects

The strongest results appear in manufacturing productivity and output. A rise in noncompete enforceability equal to 10 percent of the observed variation in the enforcement index is associated with approximately a 15.0 percent decline in manufacturing value added. The same change is associated with an 8.1 percent reduction in value added per worker, providing direct evidence that stronger enforcement is associated with lower manufacturing productivity.

Manufacturing sales also decline. In contrast, the estimated effects on capital and capital per worker are imprecise, offering little evidence that reduced capital investment explains the productivity decline. Employment also shows no major statistically reliable response in the detailed manufacturing analysis.

Average worker earnings decline as enforceability increases. A 10 percent-of-observed-variation increase in enforceability corresponds to approximately a 0.6 percent decline in manufacturing earnings, although this estimate is not statistically significant. The magnitude is smaller than the approximately 1.2 percent economy-wide earnings reduction previously reported by Johnson, Lavetti, and Lipsitz (2025) for an equivalent change in enforceability.

The authors find essentially no effect on labor’s share of income. This result is significant for interpreting the mechanism. If noncompete enforcement primarily reduced wages by weakening worker bargaining power while leaving productivity unchanged, labor’s share of output might be expected to decline. Instead, the combination of lower output, lower productivity, and an unchanged labor share is more consistent with productivity reductions playing an important role in manufacturing.

Results for the broader economy are considerably more modest. Using sector-level BEA and BLS data, stronger enforceability remains associated with lower average earnings. However, the large reductions in value added found in manufacturing do not extend clearly to the economy as a whole. The estimated effect on total value added across industries is essentially zero, while the productivity effect is substantially smaller. Employment and labor-share estimates are also statistically imprecise.

Conclusion

The study provides evidence that stronger enforcement of noncompete agreements can substantially reduce productivity and economic output within manufacturing. The estimated effects on manufacturing value added and value added per worker are considerably larger than the associated decline in earnings, while labor’s share of income remains largely unchanged. These findings suggest that reduced manufacturing earnings may reflect lower productivity rather than simply a redistribution of income from workers to firms.

At the same time, the authors emphasize important limits to interpretation. The estimated manufacturing effects are large, state-level production can shift across jurisdictions, and changes in industrial classification may complicate comparisons over time. Aggregate productivity changes could also result from firm entry and exit, worker sorting, geographic relocation, or changes within existing plants. Plant-level data would be necessary to distinguish among these mechanisms more precisely.

The broader-sector results also demonstrate that the manufacturing findings should not automatically be generalized to the entire economy. Stronger noncompete enforcement produces much smaller estimated productivity effects outside manufacturing. The principal contribution is therefore evidence that noncompete enforceability can influence economic production itself, not merely wages or worker mobility, while also showing that the magnitude of these effects differs substantially across sectors.