Streetwise Professor

February 18, 2023

Non-compete Agreements Face the Wrath of Khan

Filed under: Economics,Politics,Regulation — cpirrong @ 4:25 pm

Nina Khan’s FTC has indicated that it will outlaw non-compete clauses in labor contracts. Be skeptical. Be very skeptical.

My priors are that such restrictions are unwarranted, especially when originating in the Khan FTC.

For one thing, this brings to mind Coase’s dictum that the knee jerk response to any non-standard contracting practice is that it is the result of monopoly (or in this instance monopsony) power. Coase argued that instead most such practices economize on transactions costs. Research dating back more than a half century reveals that many practices frequently branded as monopolistic, such as resale price maintenance, exclusive dealing, and exclusive territories, reduce transaction costs associated with free riding on information. Non-compete clauses can obviously serve this function by preventing an employee who obtains proprietary information at company X from supplying that information to company Y. Eliminating this protection on property rights to information will result in less investment in this information.

Furthermore, although the monopoly/monopsony explanations that Coase belittled at least involve the application of economics, the Khan FTC has basically thrown economics out the window. (It has definitely thrown Chicago School economics–which provided efficiency rationales for RPM, etc.–out the window.) Instead, her FTC has embraced Brandeisian big-is-bad, unconventional-is-bad antitrust. The Khan Theory of Antitrust Enforcement is akin to Potter Stewart’s theory of pornography: she knows an antitrust violation when she sees it. It is a very subjective approach completely contingent on her non-economics-based subjectivity.

Moreover, the process at the Khan FTC is highly dubious. The Chairwoman is alleged to run roughshod over staff and other commissioners–as epitomized by the resignation of the FTC’s Republican Custer, Christine Wilson. This furthers suspicions that this proposed rule is purely the product of Khan’s uneconomic, idiosyncratic mind and prejudices, rather than on sober analysis.

So my priors are that anything that comes out of the Khan FTC is likely bilge.

But priors can be modified by evidence! So let’s look at the economics of a ban.

So as not to keep you in suspense: this analysis provides no reason for me to change my priors. To the contrary.

Evidently the rationale for this ban is that non-competes are a means of exercising monopsony power. Where to begin?

First, given the plethora of participants on both sides of labor markets–including markets for fairly specialized labor–there is considerable room to be skeptical that monopsony power is severe. This is particularly true for information workers for whom non-competes are most likely used.

This is especially the case today, when the reduced cost of remote work sharply mitigates one of the factors that could conceivably create a quasi rent that an employer could extract, the cost of moving.

Not that this would be a major issue for a lot of information workers anyways. Due to well-understood agglomeration effects (that have been operating since, say, the heyday of New England cotton milling in the antebellum US!) firms of a particular type, and hence employees of this type of firm, tend to be geographically concentrated. So an employee with valuable information can find work at another firm that can use that information without having to move, or move far.

Indeed, this can explain why non-competes exist in the first place. Employers in industries with a high degree of spatial agglomeration cannot rely on the costs of relocation to prevent competitors from poaching employees. Thus, I would predict that non-competes are more likely in industries where agglomeration effects operate, whereas the monopsony power theory would predict the opposite.

To make it more specific: I predict non-competes would be more prevalent in places like Silicon Valley, whereas the monopsony theory would predict that they would be less prevalent there (because employers would have less market power when relocation costs and hence quasi rents are low).

Second, although non-competes can potentially create quasi rents ex post, ex ante potential employees can negotiate contract terms (or at least reject terms offered by a potential employer). Employers can attract employees only if the other terms of the contract (including pay, but also other terms) compensate for the restriction on the employees ex post options. As long as the ex ante employment market is largely competitive, contractual terms will compensate (or more than compensate) for the employee’s ex post loss of freedom. If (due, for instance, to information disclosure problem) non-competes are wealth enhancing, the Coase Theorem comes into play. Employers and employees will negotiate other terms that make both better off.

Third, even granting arguendo that non-competes are a way of exercising market power ex ante (a dubious proposition), eliminating non-competes does not eliminate the market power, so it is by no means clear that eliminating non-competes will make employees better off. Rapacious, monopsonistic employers will just find some other way to exploit their market power. And that could make employees worse off. Eliminating an efficient way of exercising market power makes the pie smaller, and it is likely that both employers and employees would both suffer as a result.

This raises the issue of why monopsonists would prefer to use non-competes to exercise (ex ante) market power, rather than just paying lower wages/salaries, or offering fewer benefits. Perhaps you could view it as a form of price discrimination where employers exploit differences among employees by offering a menu of contracts with different pay and non-pay terms (including non-competes, or not) to extract rents. Even if that’s what’s going on, it is well known that the welfare effects of price discrimination–and hence of its restriction–are highly ambiguous.

That is, even if non-competes are a way of exercising market power, eliminating this particular way of doing so does not necessarily mean that market power will decrease: firms may instead utilize less efficient ways of exercising it.

One Chicago School staple is that a monopolist has no incentive to engage in things like exclusive dealing in order to extend market power: it can extract the maximum profit by just charging the monopoly price. Hence, there is likely some non-monopoly leveraging explanation for such contracting practices. Similar logic applies to exercise of monopsony power.

The pushback against Chicago resulted in the creation of a lot of highly stylized, not to say contrived, models in which such monopoly leveraging can occur. Nice models! Never seen one that applies to a real world situation.

I therefore surmise that even if employers have ex ante market power, it is unlikely that non-competes enhance that market power, and that eliminating them will make employees better off–because the elimination does not reduce the ex ante market power.

The empirical evidence that supposedly supports the ban is that elimination of non-competes is associated with higher wages. Even putting aside my skepticism that wages are a sufficient statistic for employee welfare (since so many other terms of employment can be varied), it must be recognized that there are reasons why elimination of non-competes in competitive labor markets could boost wages for some.

Elimination of non-competes essentially gives employees property rights in valuable information. Absent non-competes they can effectively sell this information, whereas they cannot do so (or are limited in their ability to do so) when non-competes are in place. Since the information cannot be sold separately, it is bundled with the other services supplied by the employee. All else equal, employers (including an individuals current employer and potential employers) are willing to pay more for the information-service bundle than the service alone. That means higher compensation.

As Coase and subsequent scholars (including prominently my thesis advisor Lester Telser) noted, contractual restrictions can be ways of creating or protecting property rights in information. It is highly likely that’s the role that non-competes play, especially given their widespread use in tech and other information industries. Therefore, be skeptical indeed about the FTC proposal. Indeed, even if you believe that instead non-competes are in part a means of exercising market power, you should not conclude that their elimination will make employees better off because the market power remains even after one means of exercising it is constrained.

We should all fear the Wrath of Khan generally, given her economic ignorance, and indeed her hostility to economic reasoning. We should especially fear it in her latest attempt to reshape private contracting in what is arguably the largest market in the United States–the market for labor, and for information intensive labor in particular.

Print Friendly, PDF & Email


  1. WKPD insists her name is Lina Khan.

    Anyhoo, call me prejudiced but I suspect that lawyers play far too large a part in American life.

    Comment by dearieme — February 18, 2023 @ 7:20 pm

  2. Restrictions on non-compete contracts remind me of a much more familiar example: minimum wage laws. Like so much interventionist policy, the benefits are immediate and easily identified, while the costs more indirect and delayed, just the thing for those whose policy power exceeds their competence.

    Comment by M. Rad. — February 19, 2023 @ 7:34 am

  3. “I predict non-competes would be more prevalent in places like Silicon Valley, whereas the monopsony theory would predict that they would be less prevalent there”

    I fear Chicagoan theory has blinded you, Prof, to the messy reality of the world

    Silicon Valley doesn’t have non-competes because the great State of California banned them long ago.

    Frankly, tales of non-competes being used against fast-food burger-flippers and primary care docs makes all the arguments about “property rights in valuable information” absurd. Come on! Pull the other one; it’s got bells on! 🙂

    That said I don’t think banning non-competes will do great things on pay etc. for the mass of workers; however, it will allow the poor and underprivileged to escape the grip of abusive employers. So it’s good for humane reasons.

    Comment by Simple Simon — February 19, 2023 @ 11:17 am

  4. There is another reason to use non-competes, aside from protecting proprietary information. In many industries there is a public goods problem with industry-specific but not firm-specific screening and training, e.g. finding the sandwich shop worker and teaching him how to make sandwiches without screwing up the order, cutting himself on the slicer, showing up when scheduled, etc. Until that screening and training is finished, the typical worker has negative expected marginal productivity; once screened and trained, he has positive marginal productivity. Poaching of sandwich shop workers reduces the incentive to hire and train, whereas a non-compete enhances efficiency.

    Comment by SRP — February 27, 2023 @ 8:40 pm

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress