2.1 The Fairness of NCCs: The Standard Account

Two features characterize the standard approach to NCCs. The first, which I have already mentioned, is the intuition that NCCs only directly serve the employer. The second is a narrow focus on the parties to the contract. The standard account is narrow both in only taking these two parties into account, and by focusing on their relationship at a specific stage in the employment process, often at the pre-employment stage, or the post-employment stage (see, for example: Aydinliyim Reference Aydinliyim2022).Footnote ¹ Thus, the standard account is built on an analysis of a single relationship between two parties just while they are playing certain roles. This is an intuitive starting point, but this narrowness obscures important features and effects of NCCs, as I will show later.

Standard views on NCCs differ in two important ways. First, they may or may not take the benefit to employers as justified. Committing to the standard account does not commit one to either side of the permissibility debate. Second, they may differ in strength. Strong views that take the advantage to employers as unjustified might be fully abolitionist. Alternatively, a strong view that takes the advantage to be justified or in principle unobjectionable might be thoroughly libertarian, and permissive within the limits of what the market itself will tolerate. Most positions land somewhere in between.

While fully libertarian positions are exceedingly rare, fully abolitionist views are not. The FTC rule embodies this kind of position. Abolitionists typically argue that NCCs represent a serious unfairness to the employee by providing employers with yet another advantage in an already lopsided arms race. For example, Elizabeth Anderson compares NCCs’ limits on exit to slavery, stating that “while employers can no longer hold workers in bondage, they can imprison workers’ human capital”, with the use of NCCs (Reference Anderson2019: 66). Given that states such as California seem to enjoy thriving and dynamic economies without the use of NCCs (Reference Anderson2019: 141), Anderson argues that the costs NCCs incur on employees cannot be justified.

If this account seems strong, you might still think that employees are unfairly and significantly disadvantaged by NCCs. Perhaps NCCs limit employee mobility to a degree that is economically undesirable for the employee. Erik Stam outlines some reasons for thinking that NCCs restrict employee mobility with significant economic effect (Stam Reference Stam2019). It is also possible NCCs disproportionately hurt already-disadvantaged groups. For example, Matt Marx notes that NCCs can have a disproportionately large effect on the mobility of female workers (Marx Reference Marx2022). Similarly, NCCs might be a minor inconvenience for high-income professionals that can bear the cost, but severely restrictive for low-paid wage workers. So even if NCCs aren’t a case of legalized indentured servitude, they might track and exacerbate existing injustices and reinforce harmful inequalities.

Both views take NCCs as primarily tools for employers. Restrictive and abolitionist views hold that the advantage to employers is unwarranted and worsens certain disparities. But what if this were not the case? If NCCs did not create or worsen (or better yet, if they improved) such inequalities, a case could be made for NCCs on behalf of employees. So, from the standard account emerges an important condition here for any proper application of NCCs: NCCs ought to reduce rather than expand gaps in bargaining position. This is true even on the strong abolitionist view, where the market is already dominated by the employers, or by the capitalists, and skewed by forces that drive employers to abuse and extort employees.

NCCs may satisfy this first condition if it can be shown that such contracts ever serve employees. Frye advances an account of NCCs as a signalling device that attempts to do exactly this. If he succeeds, then it is possible to defuse the abolitionist worry at least in principle, even if typical labour markets are dominated by employers. If that were true then, when NCCs are impermissible, the problem originates elsewhere. Frye’s account improves on the standard position by more completely considering the utility of NCCs. However, Frye mistakenly commits an error of scope that he shares with the standard positions.

2.2 Frye’s Account: Non-Competes as a Costly Signal

Where the standard account emphasizes the utility NCCs have for employers, Frye argues that NCCs can, in fact, serve the interests of employees in two ways. First, NCCs serve as a credible, costly signal to employers that a prospective employee is committed for the long haul (Frye Reference Frye2020: 235). Second, by consequence, NCCs allow employees to secure training, investment and commitment from employers. For a classic example of costly signals, imagine the ornate colouration of male birds’ feathers, or the massive antlers of large ungulates, or the creative nests of some birds, which males use to signal their health, strength, dominance and sexual viability. All of these take enormous amounts of energy, time and resources, and generate real liabilities for the animal in the form of exposure to predators, loss of mobility and camouflage, and so on. That means that there is a cost, a sacrifice, that the male makes in order to send a signal to prospective mates. A signal which she may well still reject. You can imagine the costly ways that humans signal viability to prospective partners. Perhaps NCCs can work in a similar way, not in attracting a mate but in communicating one’s superiority as a professional partner. By giving up the option of pursuing other opportunities, employees give employers reason to trust their commitment.

There are two difficult requirements for the costly signal account. On the one hand, there must be a sufficient probability that better employment opportunities exist for the employee. Otherwise, the signal is not, so to speak, a costly one. There would be something less remarkable about a peacock’s tailfeathers if it came at no cost or risk to him. We would not be impressed by prospective partners who had not really invested anything meaningful into signalling their superiority. On the other hand, the employee must think that the present employment opportunity is superior, all-things-considered, in order for it to outweigh this sacrifice. That is, the cost cannot significantly outweigh the long-term benefit. Otherwise, the peacock would be exposing himself for nothing. If the employee exposes himself to long-term loss by signing an NCC, he cannot possibly be acting rationally. NCCs in that case would in fact be forcing the employee to act irrationally, against his best interests. So, the cost must be high in the short run, but it cannot be too high in the long run.

This brings us to the second benefit NCCs give to employees. Employees can, by sending this signal, reliably secure further returns in the form of training. By credibly signalling the employee’s commitment, the NCC gives the employer reason to invest in training and advancement because the expected value of this investment (the probability of gain times the value of gain, minus the probability of loss times the value of loss) increases (Frye Reference Frye2020: 237). By encouraging employers to expect a higher return, employees can accrue higher benefits to themselves in the form of benefits, training, advancement, that might be wasted on an employee whose commitment was less certain. So, perhaps NCCs allow an employee to effectively trade significant short-term gains for even more significant long-term gains.

But the cost, again, is real. The ability to leave a trade relationship, the exit option, provides important leverage in negotiation. If you are worried that your customers will take their business elsewhere, you are more motivated to improve your product. Employees can similarly use the exit option to advocate for improvements in benefits, pay, working environment and conditions, and so on. Or to pursue better options when employers are unresponsive. If exit were the only feasible means of accomplishing these goals, it would be hard to see how employees come out ahead by signing NCCs. Fortunately, there is another option. Frye notes that, in the absence of robust exit options, employees must rely on a different form of leverage: voice (Frye Reference Frye2020: 241). Employees can take “political” routes to express their dissatisfaction with diminishing quality of employment. They may not be able to secure a better outcome for themselves, but they can express their dissatisfaction by, for example, striking or unionizing or otherwise establishing democratic input into the operation of the firm.

Albert Hirschman’s seminal study, Exit, Voice and Loyalty explored the relationship between exit and voice as alternative institutional feedback mechanisms (Hirschman Reference Hirschman2004). When you are dissatisfied with the quality of a repeatedly offered good, you can exit the relationship by buying elsewhere. Mass exit signals to the seller that their product is overpriced with respect to its quality. Alternatively, one can complain to the provider or voice her opinion about the product in the hopes of inspiring change. On its face, voice may seem like a less effective tool than exit; if I complain to my local brewer incessantly about the decline in the quality of his beer, but I continue to show up on the weekend, he is unlikely to give my complaints much credibility. If many of us voice a concern and then leave when our concern is not addressed, that lends credibility to our complaints. It means our words are worth something. Voice must be backed somehow. A credible threat of exit is one way, legal recourse and collective action such as striking and unionizing are others. Exit has the advantage of being easily accessible and allowing the employee to voice her own demands. I propose that this puts one constraint on NCCs; they cannot limit exit to such a degree that voice is made effectively impotent.

Frye admits not only that his positive picture may not obtain, but that it often won’t (Reference Frye2020: 242). Applying NCCs as an effective tool depends upon certain background conditions. I pointed out before that certain injustices might be exacerbated by NCCs. Frye also notes that NCCs are likely unjust when combined with other institutions such as at-will employment (Reference Frye2020: 241). In that case, the exit options of employers are bolstered while those of employees are severely constrained. Frye is therefore not arguing that NCCs are in principle beneficial to employees, but merely that they are in principle permissible on the presumption that they can be instrumentally useful to the employee, and that they are impermissible when other conditions cause them to be an unjust constraint.

The debate is therefore not about whether NCCs are in principle helpful or harmful. It is instead about whether our default position should be permissive or not, and when policy ought to depart from whatever the default position might be. For now, Frye has given us good reason to believe that NCCs can be a valuable tool for employees, and therefore reason to think we should be wary of blanket prohibitions. Restrictions are appropriate when they prevent one party from gaining or increasing an unfair advantage. But notice that the parties of interest are still just the employer and the employee. Frye’s contribution is a significant improvement on the standard account, because it both gives more credit to the employee in the bargaining position, and it gives proper attention to the possible uses of NCCs in bargaining situations. But it is still too narrow in its focus on employee and employer, and thus, still only a step away from the standard position. In the next section, I argue that this narrow focus is a mistake. But before moving on, I want to consider what can be gleaned from the back and forth up to this point. Under what conditions, if any, could NCCs be permissible?

2.3 Criteria for NCCs

Frye’s account provides new criteria for the justified use of NCCs. One is that NCCs must be enforceable, disclosed prior to employment, and employees must know (or be able to know) that they are enforceable, what the costs are, and what the consequences will be for violating the agreement (Frye Reference Frye2020: 243–244). Enforceability and transparency are not unique to Frye’s argument but take special significance when the argument in favour of NCCs depends on employees effectively using such contracts as bargaining chips. Think of this as a criterion of informed consent; I know what I am trading, and what for, and what the fallout will be for violating my contractual obligations, and I know this before I have agreed or put myself in a position which might force me to agree.

If acting as a credible signal is a necessary condition for proper use of an NCC, then it matters under what conditions a signal, credible or not, can actually be sent. I propose that NCCs can only be credible signals when they are clear signals. Clear in the sense of being discernible against background noise, and also clear in being understandable. Credibility is a necessary condition for permissibility, per Frye, and clarity is a necessary condition for credibility. Thus, if NCCs do not permit clear and discernible signalling, or if they in fact generate noise that makes it harder to send clear signals, then they are not permissible. This requires that whatever measure we employ either provides tools for identifying or amplifying credible signals in noisy markets, or at least does not amplify the ambient noise. In a noisy environment, signals are harder to make clear; think of a conversation in a loud conference reception. In the next section, I will argue that NCCs fail to reduce noise and, therefore, do not allow much credibility when broadly permitted.

It is worth noting quickly that often it is impossible for all members of a population to send the same costly signal. Not all peacocks will produce the same tail feathers. Finite supplies of food, exposure to predators, age, genetics and other factors mean that they will not all be the most impressive. There are resource limitations that ensure that just a few can signal above the noise. So sometimes signalling mechanisms have a control mechanism built in, so to speak, that ensures that different players don’t enter an endless arms race. This is not true of NCCs; everyone can choose to not take other opportunities.

Finally, Frye is interested, as are others, in ensuring that employers know with greater certainty that investments in employees will pay for themselves by providing sufficient returns. This is done on Frye’s account because of the signalling effect of NCCs, which demonstrate a reliable commitment on the part of prospective employees. But, as I’ve discussed, it is also important that employees have an assurance that this sacrifice has a reasonable promise of return. As Frye notes, this is often a condition on the legal enforcement of NCCs (Frye Reference Frye2020: 231). This condition, that there be consideration in return for sacrificed opportunity, will be weighed on its own in court, but here I absorb it under the requirement that there be assurance – a show of confidence – in both directions.

Combining these new conditions with the old, I propose that NCCs are justified and permissible only if they:

1. 1. At least do not expand gaps in bargaining position; call this Proportionate Leverage.

2. 2. Ensure employees are adequately and accurately informed about the tradeoffs, and only when the NCC is enforceable; Informed Consent.

3. 3. Work to the advantage of employees by allowing them to send costly signals that are both clear (in the sense of being identifiable in a sea of signals) and credible; Signalling.

4. 4. Provide Assurance to both parties that their tradeoff has a sufficient expected return.

5. 5. The NCC does not cause disruptions in the market that reduce long-term value for the employee, introduce serious asymmetry of information, or create inefficient non-innovative markets; Market Failure.

In the next section, I argue that NCCs fail on all counts. NCCs can generate market failures that might induce greater use of NCCs; they generate asymmetries of information and uncertainty that harm employees; they hurt “other actors” in the market – although it will also turn out that “other actors” includes the employee signing the contract, but when she enters a different role, a fact obscured by the double narrowness of the standard account; and, finally, they generate noise such that clear signalling (and, by consequence, credible signalling) becomes impossible.

3.1 Information Asymmetry

I begin with a market failure that distorts bargaining positions. Broad use of NCCs generates an information asymmetry. Starr et al. (Reference Starr, Prescott and Bishara2021) and Lipsitz and Starr (Reference Lipsitz and Starr2022) note that employees are unlikely to negotiate NCCs during the hiring process, are often asked to sign NCCs after being hired, often don’t know whether the NCCs they sign are enforceable, may not realize or exercise their ability to seek counsel from friends or lawyers before signing, and so on. This, combined with Marx’s observation that female workers are likely to feel more constrained by NCCs by default (Reference Lipsitz and Starr2022), suggests that not only will there be an information asymmetry when NCCs are employed, but the strength of this asymmetry will track pre-existing gender disparities, and potentially other pre-existing power disparities. And Frye is careful to note this, arguing that among the conditions are things like actual enforceability, refraining from targeting low-income workers, etc. Yet the important thing to note here is that NCCs bring about important information asymmetries that may still be justified by the signalling approach.

Another asymmetry emerges because employees cannot identify future options or explore their value in the market. The former is true given that widespread use of NCCs dampen innovation and entrepreneurship (discussed below) and can therefore dramatically change what future options emerge for the employee (or other job seekers). The latter comes just from the fact that NCCs limit the ability of employees to pursue other professional relationships, though that this is not necessarily the case for employers. A non-compete is a promise, on the part of the employee, not to participate in the labour market. It does not entail any restriction on the part of employers, though employers may be limited by budget constraints or by the legality of at-will employment. The latter has been dealt with by Frye, who emphasizes that employers cannot maintain the right to unilaterally end the employment relationship while denying exit to employees (Frye Reference Frye2020: 241). But this doesn’t solve the problem, it just slightly dampens the informational advantage available to employers.

This asymmetry is not entirely limited to the participants to the contract. By effectively blocking certain members of the market from participating (and especially those that are developing important knowledge and skills while actively employed) valuable information may just be removed from the market, especially if participation is sufficiently constrained. This is because workers can only bring their labour to market periodically, as determined by the NCC. Employers can hire despite NCCs, but will still face a smaller pool of talent, and employees. Employees cannot test their own value on the market easily, and the group of peers to whom they might look for indicators has been substantially shrunk. This means that other job seekers are engaged in a market where important information may either be unavailable or lag significantly. One would expect, under these conditions, for the information-deprived side of the bargain to lose some ground. This kind of system-wide loss of information, combined with the reduction to employees’ bargaining power, may be visible in the substantial work on the wage-suppressing effects of NCCs (Marx and Fleming Reference Marx and Fleming2012; Starr et al. Reference Starr, Prescott and Bishara2021; Lipsitz and Starr Reference Lipsitz and Starr2022; Marx Reference Marx2022).

It would be difficult to tease apart the impact of information loss from the impact of reductions in bargaining power, but perhaps an example of how this might look would help. However, Michael Lipsitz and Evan Starr (Reference Lipsitz and Starr2022: 162–163) show that the state of Oregon’s NCC ban increased wages more broadly by 2–3% and increased monthly worker mobility between jobs by 17%, even though only a portion of the workforce was bound by such contracts. They suggest that the real effect just on workers under such contracts could therefore be as high as 14–21%. Now, if employers are at a bargaining disadvantage without NCCs, this might be fine. But assume briefly that employers are not disadvantaged without NCCs. In that case, if Lipsitz and Starr are correct, a worker looking at the labour market broadly in Oregon before the ban on NCCs would be off by 2–3% on the fair value of his labour, even if he is not bound by an NCC.

George Akerlof observes how markets gradually deteriorate under the weight of market failures in the presence of information asymmetries (Akerlof Reference Akerlof1970). This motivates the introduction of different mechanisms to “respond”, including bigger and better signals of assurance, and government interventions, none of which are likely to work where information asymmetries and dishonesty are pushing out honest competition (Akerlof Reference Akerlof1970: 495–497). Because, in this case, the use of an assurance mechanism (NCCs) creates the need for further assurance mechanisms, I will call this phenomenon signal inflation, and will return to it later. While, as I’ve suggested, there are some ways in which it is hard to look at the labour market in all of the same terms as other markets,Footnote ² it is very clear how a labour market can become a “market for lemons”, and how this kind of breakdown can harm workers’ future prospects, in ways that they probably cannot predict.

3.2 Reducing Innovation and Dynamism

One way in which NCCs can reduce long-term opportunities for workers without their knowledge is by dampening innovation, entrepreneurship and market growth. Beyond creating information asymmetries and their corresponding market failures and inefficiencies, NCCs have a well-documented effect of stifling economic development compared with investment. For example, Samila and Sorenson note a negative relationship between NCC enforcement and the effects of increased venture capital on new patents, startups and employment growth (Samila and Sorenson Reference Samila and Sorenson2011).

Marx and Fleming similarly note that NCCs limit individual career mobility, protect established firms at the expense of smaller startups (keep in mind that it takes corporate resources to pursue the enforcement of NCCs, as well), and have adverse regional impacts on entrepreneurship and innovation (Marx and Fleming Reference Marx and Fleming2012). This is important. As I observed before, NCCs often require employees to leave either an industry (or part of the industry) or geographic area. This means that the limiting effects of NCCs will be most pronounced in the industry and the community in which they are enforced. This can serve to entrench any firm that can establish itself in a given community, creating barriers to entry for those that follow. Not only do NCCs impose costs of departure on employees, but they impose costs on the local community and economy by pushing potential innovators elsewhere.

3.3 Uncertainty

Because employees will likely face diminishing opportunities as a result of the innovation-dampening effects of NCCs, the tradeoff that was essential to the costly signal argument loses its rational appeal. Remember that a vital part of the picture I have been drawing (and the one which Frye is drawing) is that NCCs are permissible to the degree that they provide both parties with assurances of investment; that is, to the degree to which they reduce uncertainty for both parties. This is because for both employers and employees, greater probability drives up the expected value of investment, defined as the sum of the values of each possible outcome multiplied by their probability.

The expected value is the probability of a gain times the value of that gain minus the probability of a loss times the value of that loss. Recalling that the probabilities of all possible outcomes must sum to 1, expected value can be increased by raising the value of the reward, reducing the value of the loss, or by adjusting probability to favour gains. This third option is what NCCs do for Frye: they allow employees to signal, or reassure, employers that the probability that they represent a positive return is much higher. This is better than option two, reducing cost by withholding investments in employee development. Increasing the value of potential return is not strictly up to employers. By increasing the perceived probability of return, employees can use NCCs to promise increased expected value for employers. Because the expected value of the gain increases for the employer, they will increase the value of the return for the employee (through training, advancements, benefits, and so on). Both parties improve the expected value of the long-term option compared with the short-term options the employee is giving up.

But if NCCs dampen innovation, reduce options and limit employee position in the market outside of their present firm, this reduces the probability that, in the long run, employees will see a positive return on their initial sacrifice. Remember that to be a costly signal the price of the signal must be sufficiently high. The employee is now paying a high price for a future that is increasingly improbable the more NCCs are employed in the market. As the practice becomes more widespread, the probability of gain goes down and, as a consequence, the probability of loss goes up. Employers could reposition themselves by expanding their offerings, but this reduces the cost of the signal (the cost, after all, being comparative). NCCs will begin to lose credibility, unless they can otherwise be bolstered.

This process continues as both parties try and give a better offer back. Employers must now reassure employees that paying a significant price in the short term still has a sufficiently high expected value. If this seems hyperbolic, remember that this is the kind of phenomenon that Akerlof already observed in markets with information asymmetries (Akerlof Reference Akerlof1970). Increasingly trying to improve one’s signal in relative credibility or visibility, calling on increasing levels of intervention to counteract the distorting effects of market failure, and establishing authority all follow from the problems discussed so far, and contribute to the next major problem, which cuts at the core of Frye’s argument. Because of this effect, which I’m calling signal inflation, NCCs generate noise, drowning out what could be clear signals and making them harder to send, identify or trust.

3.4 Cloudy Signals in Loud Spaces

We have now seen several interconnected ways that NCCs generate noise rather than clarity, especially if broadly used. First, broad use of NCCs introduced information asymmetries. Second, NCCs harm future prospects for employees. Third, because of these, NCCs reduce the value of NCCs, making them costlier when compared with (increasingly uncertain) returns. This creates an incentive to out-do other players, generating noise that makes the increasingly costly signalling device increasingly ineffective. It then becomes harder and harder to identify an employee signing an NCC as sending a credible costly signal, as opposed to just being another desperate player in the game.

Imagine the din of family shouting over dinner, or students talking over one another, or any other setting where everyone seems desperate to be heard and unable to make themselves heard. This is the picture that has emerged when NCCs are widely used. An ever-escalating din of voices. This serves neither the employee nor the employer, it does not reduce uncertainty, and it does not allow either party to more easily identify or send a signal. If the major strength of an NCC comes from the ability of the employee to use it as a credible and costly signal, this is a major problem. The ability to send a recognizable signal depends upon the ability to set oneself apart clearly; that is, sending a clear signal is a necessary but not sufficient condition for sending a credible one.

The upshot of all of this is that NCCs fail to meet any of the criteria previously set out, at least when broadly permitted. They fail (5) Market Failure, by both creating information asymmetries and disrupting the labour market, as well as diminishing future options for the employee. They fail (1) Proportionate Leverage, (2) Informed Consent and (4) Assurance by first reducing employee certainty and future options, and second, by generating noise in the market that makes clear signals hard to send. Because of this, NCCs fail (3) Signalling, by making it hard to identify credible signals, and by making it hard for employees to make their signals clear to begin with.

Frye acknowledges that he is fencing off questions about the impacts of NCCs outside the employee-employer relationship (Frye Reference Frye2020: 246). Given this scope of analysis, it is easy to conclude that NCCs can send clear and credible signals, and that it is up to the employee to decide whether to take on the associated risks (and Frye never claims that there aren’t risks to signing an NCC; recall that risk is a part of what makes the signal costly to begin with). What I have offered is a broader, system level examination, to see if the same critiques and defences succeed when one thinks beyond the employee-employer relationship. This section of the paper has shown that very simple and well-documented effects of NCCs put us in a bit of a bind: NCCs can only work as a credible, costly signal when they are not widely adopted. When widely adopted they generate market failure, information asymmetry, and generate noise that makes clear (and, by consequence, credible) signals impossible to send or receive. This suggests that NCCs should not be accepted as a matter of course.

It also means that NCCs can hurt the very people Frye suggests they can serve. Remember the double narrowness of the conventional analysis. NCCs are often considered solely in the context of the employer-employee relationship and only within a narrow window where the two parties are playing specific roles. It considers the employee typically at one stage in the employment cycle, but not others. It is focused, in the terms I used earlier, not just on the two specific parties, but on those parties when they are filling certain roles. The justifications of NCCs, including Frye’s, depend largely on thinking of employees narrowly in their roles as applicants considering the utility of “the next step”. That is, in terms of what advantages accrue over the course of this employment opportunity. This kind of approach largely ignores that the role of applicant, or active seller in the labour market, is a role to which the employee will return. And, as I hope this section has shown, the standard accounts largely ignore that the marketplace to which the employee returns is a dynamic environment that is significantly impacted by things like NCCs. It is, in fact, impacted in ways that undermine the benefits supposedly accrued to the employee.

Imagine that in a sector of the labour market NCCs are ubiquitous, so that the threat of exit is no longer credible. Each employee has guaranteed his services to an employer, and his options are self-limited enough to credibly signal commitment to that employer. Under such conditions, with no credible threat of exit, the employee can voice his concerns until he is blue in the face, but there is no other option to fall back on if the employer is not receptive. The options left to back the use of voice would be other routes of defection (slacking, so-called “quiet quitting”, and so on), some legal action, or collective action such as striking. The first can be worse than having an employee depart in that you now run the risk of paying an employee that is not a net positive contributor even in the short run. The second and third options, legal and collective action, are often inefficient and also absorb the employee into a class rather than letting her negotiate for herself.

Just like a defence of defecting in the prisoner’s dilemma works only if the recommended strategy is not widely used, defending NCCs as a signalling device is persuasive just on the condition that the tool is not widely used and enforced. The insight of the prisoner’s dilemma is that, supposing you have cooperative people to play against, you should see gains by defecting. But when everyone does what it seems they rationally ought to do, everyone is made worse off. You end up with the second worst outcome for the individual, and the worst outcome for society. The paradox for both cases is that the best option on a narrow analysis is increasingly untenable the more you expand your gaze.

Still, it seems possible that NCCs might serve as helpful signalling mechanisms under different conditions. Very significantly, this seems to depend on NCCs not being widely employed. My contention, because it is unlikely that NCCs are never justified, is this: we have good reason to doubt that NCCs are in principle permissible, and, more importantly, we have prima facie reason to treat NCCs as impermissible, but overriding concerns or other market failures or disruptions might justify their use as an emergency intervention in exceptional circumstances.

In the next section, I lay some groundwork for the acceptable use of NCCs as emergency interventions in exceptional circumstances. But first it may be helpful to quickly distinguish my claims here about market failures and distortions from another view. My approach here depends significantly on the causes and effects of market failures. It might be easy to interpret it, for that reason, as a market-failure or Paretian ethical account, along the lines of Joseph Heath (Reference Heath2006). However, I do not depend on an ethical obligation to prevent, diminish or avoid exploiting market failures for the sake of efficiency. Instead, I argue only that such market failures cause harm to the parties of interest on both narrow and wide conceptions, as well as obscuring these harms. The expanded analysis shows how even the parties to the contract are negatively affected, as well as giving reason to think there are harms, which we should take seriously, to other parties. With that clarified, consider how NCCs might be permissibly deployed as a targeted intervention.

4.1 No Internal Controls

One might argue that even given the multitude of distortions discussed above, parties in the market should be allowed to contract voluntarily, even at substantial cost to themselves, and even when their choices stretch the limits of rationality. A strong version of this kind of position is advanced by Matt Zwolinski in the context of price gouging (Zwolinski Reference Zwolinski2008), and could easily be adapted for non-competes. The intuition is that if a practice is not inherently harmful, then it might be permissible for voluntarily contracting parties to engage in the practice even in specific cases where it is harmful. This seems sensible enough, but note that in the case of NCCs this is made harder by the fact that information asymmetry, uncertainty and signal inflation may make it difficult or impossible for employees to give genuine informed consent. This leaves the voluntariness of contract in the case of NCCs in significant doubt.

There is also a distinction between the two cases that highlights my recommendation in this section. Discussions of price gouging have, built into them, the assumption that the conditions under which price gouging occurs are abnormal. That is, price gouging itself is a response to exceptional circumstances, potentially where the absence of price gouging would generate even greater inefficiency. The assumption is not that price gouging restores efficiency but that it might get you through, so to speak, until supply chains or power or other infrastructure are restored. In the absence of market failure, “price gouging” is not typically possible.

I would like to argue something similar for the permissible use of NCCs, but this case is more difficult because NCCs are neither possible nor enforceable only in cases where there are serious market inefficiencies. I can only charge you a 400% markup on ice in the wake of a serious disaster or collapse in the market. Your employer, however, can stop you from moving to a competitor even when the market seems to be working fine. Similarly, competitors can stop their employees from moving to my company. The constraints internal to price gouging, courtesy of normal market forces, do not, as I have shown, emerge for NCCs. Quite the opposite, in fact. Such constraints, then, must be imposed from the outside.

I also want to note that this counterargument would concede much of what I’ve advanced here, arguing that it is within both parties’ rights to do themselves some harm in the process of freely engaging in the market. In one respect it would be less an objection to the argument I have advanced than an objection to more basic principles. The problem is that even if one grants the importance of voluntary contract, which I in fact do, the environment I have explored here does not lend itself to such voluntary contract (for example, because the criterion of informed consent cannot be satisfied when NCCs are widely used and enforced). Nevertheless, this does provide one feature to look for in a theory of permissible deployment of NCCs as a targeted intervention; external constraints must be brought to bear to ensure the intervention does not outlive its usefulness.

4.2 Solving Information Asymmetries

But suppose I am right that non-competes, when broadly employed and enforced, have these negative effects. One might still argue that they are permissible granting certain other enabling institutions. For one such solution, think of the analogy I made to the market for lemons. Perhaps information asymmetries are generated by non-competes on their own, and this can reduce the certainty that employees have in their future options, but this can be addressed by adding other practices to reduce the information asymmetry.

My case was that by reducing the freedom of workers to engage fully in the marketplace, workers’ information about the value of the good they are selling is reduced. Suppose that, to avoid this, a policy of disclosure is introduced, similar to the practice of disclosure in real estate. The “price” of each contract for labour could be reported, so that employees have better knowledge of their position. There are two reasons why this does not address the problem at hand. First, the problem in the case of NCCs is that supply is artificially suppressed while the price mechanism is also controlled. Imagine that, in the housing market, a house could not be sold again for a period of time after a sale. The market would no longer dynamically update the cost of housing. The second problem is that, even given the information about the going rate for labour, workers can’t do much because they are limited in options they can pursue. They cannot, for example, negotiate a raise because company B would pay them better.

But this does highlight again the use that NCCs might have under exceptional circumstances. Consider the house flipping trend of the 2000s or the short-term rental boom of the COVID-19 and post-COVID-19 era. Both of these had significant adverse effects on local housing markets, both in terms of housing availability for residents, housing prices and other negative externalities for local communities. In both cases, some areas enforced strict regulations on how property could be used or traded (Kim et al. Reference Kim, Leung and Wagman2017; Agarwal et al. Reference Agarwal, Chau, Hu and Wan2022; Furukawa and Onuki Reference Furukawa and Onuki2022) though assessments of these measures’ effectiveness vary. Most recently, some areas have enforced limits on the number of short-term lease properties within a neighbourhood or on a city block. Something similar might be true for labour under some circumstances.

But note that in the two housing cases, one market has been pushed out by another as a result of some other disruption. One might easily say that the long-term housing market, which serves residents, has been pushed out by the tourist economy and this should be solved by innovation rather than intervention. Similarly, one might argue that NCCs should nevertheless not be used as an intervention. Instead, whatever market failures seem to call for such an intervention should be addressed by innovation and entrepreneurship. I do not take a position on whether NCCs should be employed in any situation. I am instead allowing for the possibility that NCCs could be effectively used under some circumstances, provided they are treated like targeted and temporary interventions to counteract failures greater than the ones they cause.

4.3 No Clear Message

Non-competes are not, then, always and obviously a credible or reliable signal of loyalty. For one thing, NCCs may not signal loyalty for two reasons. First, if all employees (at least in an area) in a firm are required to sign an NCC, and especially if use of NCCs is broad outside the firm, then signing an NCC signals nothing at all. It is simply what one must do. Second, as Hirschman observes, genuine loyalty must come with both (a) the expectation that one’s voice, in part because of their loyalty to the firm, will be heard, and (b) the expectation that improvement of the good (employment) is not only possible but feasible (Hirschman Reference Hirschman2004: 78). The former expectation is not reasonable without a credible threat of exit, and the second expectation is not feasible without the first. However, given these challenges, surely NCCs can serve this function sometimes. It is also possible that signing an NCC can signal things other than loyalty. What I’m more concerned with here is identifying the conditions under which NCCs can actually fill the role Frye suggests.

As I emphasized before, one precondition for the signalling effect of NCCs is that they are not broadly employed and enforced. NCCs can be profitably employed, then, only when some disruption calls for a significant deviation from the status quo. This might be the case when NCCs are used as a targeted intervention, under exceptional circumstances, to avoid even greater market failures. Under what kinds of circumstances would NCCs provide a feasible solution?

Imagine that there is a serious deficit in labour supply, meaning that employment opportunities far exceed available labour and employees can leverage their exit options for excessive gain. In such an environment it might become common to see workers hop from job to job, collecting skills at the expense of employers, and driving up wages. NCCs might profitably be employed to curb rampant exit and counteract the existing inefficiency. But the NCCs should be structured in such a way that they address the dysfunctional conditions without limiting exit in the long run.

I won’t offer a comprehensive account of when NCCs can be permissibly deployed. But the case I have presented here might be satisfied by something like the following. When demand so exceeds supply that workers can seriously exploit the inefficiency, NCCs might satisfy the anti-inegalitarian condition; buyers (employers) are at a disadvantage and their position might be improved by restricting employees. However, if NCCs are permitted indefinitely then the problems listed above arise. To solve this, temporarily introduce NCCs with both a strict time limit, and shift the start of the NCC from the end of employment to the beginning. An NCC extends long enough to guarantee a reasonable return on investments from the employer, incentivizes commitment in the mid-term, but expires before the employee departs, thus dampening mobility slightly without stagnating the labour market. The employee can make a better calculation about whether the tradeoff of short-term opportunity for long-term gain is in their interest. They can also be confident that the negative effects of temporary use of NCCs will not reduce future opportunity more than highly inefficient markets. So long as the NCCs are applied in order to dampen the destabilizing effects of the deficit in supply, and do not outlive the target problem, then such contracts could plausibly serve as an effective, targeted intervention.

Finally, note again that much of my argument here is conceded by this response. It allows that NCCs are not obviously in principle permissible, nor beneficial. Instead, they require other policies to mitigate their harmful effects. So, the important question becomes whether NCCs, with all their baggage, can be regulated to the point where they provide more benefit than harm. I think that they in fact can be, but that this demands, first of all, that such contracts be treated as emergency interventions rather than broadly permissible.

4.4 A Delicate Balance

One last thing that emerges from Hirschman’s analysis and from this project is that voice and loyalty are not replacements for exit. Instead, exit and voice are essential options that support loyalty in market relationships. They must be kept within a balance (though this gives broad margin) in order for voice or exit to give anything like credible, discernible signals.

My claim is not that NCCs or other limits on exit cannot be justified. Instead, I am arguing that they ought not to be the default. They are justified just when easy recourse to exit, to borrow Hirschman’s language, makes use of voice sufficiently unlikely, and are justified only to the degree that they restore both exit and voice to proper strength. It may be that in some sectors of the labour market exit is so ubiquitous that constraints on exit are necessary. But the above example should draw out that this is a case of market inefficiency that is not made better by broad, long-term use of NCCs. The conditions that call for NCCs are not, then, the product of efficient markets, but they can be the product of improper limitations on exit, like excessive use of NCCs.