Research Highlights Podcast
October 22, 2026
Noncompetes and the spread of ideas
Kate Reinmuth and Emma Rockall discuss the effects of labor mobility restrictions on innovation.
Source: Gajus
In 2024, the Federal Trade Commission (FTC) issued a rule banning most noncompete agreements nationwide, arguing that greater worker mobility would foster new business creation and raise workers' earnings. The rule was later vacated, but it brought renewed attention to a longstanding debate over the economic effects of noncompetes. Supporters of noncompete agreements contend that these restrictions encourage firms to invest in new ideas by preventing employees from taking proprietary knowledge to competitors.
In a paper in the American Economic Journal: Applied Economics, authors Kate Reinmuth and Emma Rockall examine the effect of noncompetes on innovation, using 26 changes in state-level noncompete enforceability between 1991 and 2016. They find that an average-sized increase in enforceability reduced patenting by about 14 percent after five years, with larger declines among the most novel patents and venture-backed firms.
Reinmuth and Rockall recently spoke with Tyler Smith about the arguments for and against noncompetes, how they isolated the causal effect of enforceability on innovation, and what their findings imply for policymakers.
The edited highlights of that conversation are below, and the full interview can be heard using the podcast player.
Tyler Smith: What is a noncompete agreement, and how common are they?
Kate Reinmuth: A noncompete agreement is a provision in an employment contract that prevents a worker from joining or starting a competing firm for some period after they leave their initial employer. In general, these agreements are pretty widespread. Prior estimates suggest that roughly 30 to 45 percent of the US private-sector workforce is subject to one, and they're especially common among the kinds of workers we would expect to be most directly involved in innovation: engineers, architects, scientists, researchers, and so on.
Smith: What is the case for thinking that noncompete agreements encourage innovation?
Emma Rockall: The case for noncompetes is ultimately all about innovation incentives. Just as patents are thought to encourage innovation by allowing firms to better appropriate the returns on their R&D, if noncompetes help firms protect their confidential information, that might incentivize them to produce more of it.
There has historically been a common presumption in policy discussions that this benefit to innovation must be traded off against negative impacts on workers, like reduced wages and mobility. But we argue in our paper that there are also important reasons to think that noncompetes may harm innovation.
One reason is that they create barriers to entry. Noncompetes can make it difficult for a would-be entrepreneur to enter the market if they're subject to a noncompete agreement. But even if they aren't, noncompetes might make it difficult for new firms to hire if many of the workers they would need are themselves subject to noncompetes. So, they can't hire the talent they need to grow.
Noncompetes can also reduce knowledge diffusion. If workers moving between firms is an important mechanism through which innovations diffuse, then limits on employee mobility will hamper the ability of other firms to learn about and improve upon new technologies produced by rivals.
Smith: Naively, you might think that you could just compare states with stronger noncompete laws to those with weaker laws to estimate their impact. Why doesn't this approach work well, and what strategies did you use to get around these problems?
Reinmuth: There are a couple of problems with doing these kinds of simple cross-state comparisons. The first has to do with endogeneity or reverse causation. You might think that innovative firms strategically choose to locate in states where noncompetes are more enforceable, for example.
There are also a lot of other things that differ across states—legally, economically, and politically. People often compare Silicon Valley to Route 128 and point to California's refusal to enforce noncompetes. But there are lots of other things that differ between California and Massachusetts, so isolating the effect of noncompetes specifically has been really hard in the literature.
Rather than doing what we might call a static cross-sectional comparison, we exploit plausibly exogenous changes in state-level enforceability. These let us compare how innovation changes in treated states relative to control states that aren't experiencing policy changes. I think that gives us a pretty reliable estimate of the causal effect of noncompete enforceability on innovation.
Smith: In your paper, you first look at a 2004 Ohio Supreme Court decision, Lake Land v. Columber. What did that ruling change and what happened to innovation in Ohio afterward?
Rockall: In 2004, the Supreme Court of Ohio made noncompete agreements easier to enforce. Prior to the change, if an employee signed a noncompete agreement after starting their job, the employer had to compensate them for doing so, perhaps through a promotion or a pay raise. Otherwise, the agreement wouldn't be enforceable. After the ruling, continued employment was sufficient compensation—so, essentially, "Sign this noncompete agreement or I’m going to terminate you." It meant that Ohio firms could hand out noncompete agreements to their existing employees the next day at no cost, and you might expect that to increase the use of noncompete agreements within Ohio.
Indeed, what we see is that, after the ruling, patenting in Ohio fell significantly relative to a synthetic-control Ohio, which we built from states that were not experiencing changes in noncompete enforceability at the time. And, to preview the mechanisms, we also see a sharp drop in inventors moving between firms within Ohio, which is exactly what you would expect if the ruling was driving the effect.
Overall, I think the evidence suggests that many states may be able to weaken current noncompete law without sacrificing innovation, and that they may actually increase innovation in the process.
Kate Reinmuth
Smith: When we look at your full sample of 26 changes in enforceability, what did you find? How much does patenting decline after an increase in enforceability?
Reinmuth: In our baseline sample from 1991 to 2016, we focus on these 26 changes in enforceability: 17 strengthenings and 9 weakenings. They're clean treatments in the sense that there aren't other changes in enforceability close in time that make it difficult to isolate the effect of a particular change.
Across these changes, we find that stronger noncompete enforceability produces an economically and statistically significant decline in patenting. For example, for an average-sized change in enforceability, which is pretty close to the Lake Land case we were just describing, we estimate that patenting falls by about 14 percent after five years. So, it's a pretty strong effect, but one we think is reasonable in magnitude given the changes we see.
Smith: What are the potential explanations for this fall in innovation?
Rockall: Noncompetes may strengthen incumbents' incentives to innovate, but they may also restrict entry and reduce labor mobility. Since the overall effect is negative, the incentive effect must be outweighed by one or both of the others, and we find that it appears to be mainly mobility.
None of the entry effects we look at are large enough to explain the overall decline we see. And if we focus on incumbent patenting, it falls by almost as much as patenting overall. So, entry alone can't explain the negative effect.
On mobility, though, the evidence is strong. After an average-sized increase in enforceability, the share of inventors changing employers drops by about 31 percent after five years. We see this in knowledge flows too. Firms' share of citations to their own patents increases by 17 percent, suggesting that when inventors can't move, firms build more on their own prior innovations and become more technologically isolated.
Smith: What does this work suggest for policymakers weighing restrictions on noncompete agreements, either at the state level or at the national level?
Reinmuth: I think the main implication is that policymakers shouldn't assume that stronger noncompetes involve a trade-off in which workers lose but innovation gains. That's what noncompete proponents have argued in the past: yes, there are negative effects on worker welfare—wages fall, and so on—but there are all these innovation gains. Our results run opposite to that. We find that stronger enforceability also reduces innovation, and the effects are particularly large for novel technologies and highly innovative firms.
At the same time, in the paper we're careful not to argue that noncompete enforceability should be zero. Our reduced-form estimates are based on the kinds of changes we can observe historically in the data, so they're best suited to evaluating reforms of similar magnitudes. But overall, I think the evidence suggests that many states may be able to weaken current noncompete law without sacrificing innovation, and that they may actually increase innovation in the process.
♦
“Innovation through Inventor Mobility: Evidence from Noncompete Agreements” appears in the October 2026 issue of the American Economic Journal: Applied Economics. Music in the audio is by Sound of Picture.