I am delighted to announce that this month’s guest article is authored by Richard Epstein, Laurence A. Tisch Professor of Law at NYU, and Director of the Classical Liberal Institute. Richard argues that antitrust law shall not be extended to labor markets. I am confident that you will enjoy reading it as much as I did. Richard, thank you very much! All the best, Thibault Schrepel
The Unwise Extension of Antitrust Law to Labor Markets 1I should like to thank Jeremy Brown, Samuel Milner and Mitchell Pallaki for their superb research assistance.
During this past year, the Biden Administration has been intent on expanding the scope of antitrust law through an Executive Order whose ostensible purpose is to improve competition in product and labor markets. In the abstract, this aim is laudable, but, as I have argued previously both here and here, the means chosen by the administration are not. It is widely agreed that monopoly (or monopsony) power reduces overall output and is thus a source of social loss. It has also been widely agreed that collusive behavior is a live possibility in the markets for products and services. The total sales of such common products as potash, concrete, and gasoline are huge, the number of producers is often relatively few, and the barriers to entry are often large, which makes it relatively easy to apply Chicago-like antitrust principles to attack various unlawful horizontal arrangements such as price-fixing and territorial division.
More doubtfully, Biden administration has taken the view that the same Chicago-like logic applies with equal force to labor markets:
When there are only a few employers in town, workers have less opportunity to bargain for a higher wage and to demand dignity and respect in the workplace. In fact, research shows that industry consolidation is decreasing advertised wages by as much 17 percent. Tens of millions of Americans—including those working in construction and retail—are required to sign non-compete agreements as a condition of getting a job, which makes it harder for them to switch to better-paying options.
Fortunately, this gloomy assessment is wholly off point because it misses all the relevant structural differences that distinguish labor markets from product markets. There are two major components of the administration’s program touching on the general question of industry consolidation and the specific field of non-compete agreements. I shall critique these points in order.
The fundamental error that pervades the Biden critique of industry mergers is that it posits a false equivalence between product and labor markets. Historically, labor markets have been considered sufficiently competitive such that no antitrust oversight was required even if other forms of regulation were necessary to deal with issues such as health and safety. That traditional approach was—and is—correct. The firms that make common products often use very different means of production, which means that they need not draw on the same pools of labor talent in order to achieve their ends. In addition, there is no reason to think that workers are bound to apply their skills only within a given class of productive activities, defined by industry classifications. For high-skilled workers, their grasp of abstract and general principles makes it highly likely that they can take key management and data skills with them to other industries. For low-skilled workers who provide janitorial, catering, or clerical services, for example, movement across market sectors is a relatively easy matter because the current skills are easily transferrable. In both these occupational groupings, it is implausible—if for somewhat different reasons—to say that there are “only a few employers in town.”
Situations where multiple employers compete for a given worker are very much the rule, not the exception. Today, there are acute labor shortages across multiple fields, a development that has occurred for good reasons. These shortages give employers strong incentives to reach out to workers in different fields in order to bring them into their own. In many cases, the needed skills are easily transferrable from one line of business to another. And in those cases where additional skills are needed, firms are already accustomed to giving on-site training to familiarize new hires with peculiar firm practices, precisely to increase the pool of available workers. In current markets, common perks also include signing bonuses and additional workplace amenities.
Where those internal firm solutions do not make sense, many large and small firms contract out large portions of their work to other firms, many of which operate in small, localized niches. By way of an example, a nationwide oil company may hire one or more smaller local or regional specialty firms to perform maintenance, repair, and delivery work. For all these reasons, labor markets rarely track product markets and, as a result, the Herfindahl-Hirschman index (HHI) that works tolerably well to measure concentration in stable product markets (although cannot track well the impact of innovative new firms) sputters from the outset in labor markets.
There is also nothing in the record that supports the notion that applying antitrust laws to labor markets will—or even could—improve levels of “dignity and respect in the workplace.” This last phrase is a clear indication of Biden’s strong pro-union stance. His official position paper wrongly treats the purported dominance on the employer side of labor markets as the cause of the difficulty for workers to bargain—a supposition that is an indirect appeal to strengthen labor unions to operate as a counterweight to the purported level of employer dominance. But that argument also fails. There is no evidence that unions will take root only in firms that purport to have monopoly power. They can spring up anywhere and thus introduce monopoly elements in otherwise competitive markets. Yet ironically, in markets that have high levels of labor mobility, unions find it much more difficult to gain a foothold. Why will workers want to take jobs that cut them off from close, direct contact with the management whose ranks they may hope to join? Why will workers want to undertake union jobs or pay substantial dues to a firm that they may quickly leave for better opportunities? Why will workers want to sign up with firms that could be pushed over the edge by high product prices and unproductive labor negotiations? Is there any surprise that the United Auto Workers has not been able to gain foothold with Japanese and German automakers, when it has inflicted huge losses on traditional American makers? For some workers, these concerns may not dominate. But for others, they become ever more salient. Unions currently are exempt from the antitrust laws, but even that protective boost in the private sector has not prevented them from losing additional market share—in 2022, they barely constitute six percent of the current workforce. And, why would anyone who cares about maintaining competition ever want to boost unions if, in fact, they only survive because they are given an explicit exemption from the antitrust laws?
The Biden labor monopsony story, moreover, cannot account for today’s massive movements of labor under what is called the “great resignation,” which involves extensive shifts in at least three relevant dimensions: movement across occupations; movement across geographical territories; and movements from employee status to entrepreneur status, as evidenced by the record number of start-ups that have taken place in the midst of the pandemic, which has already created a boon for online work, a boon that (like Zoom meetings) is likely to survive even after the pandemic is over.
This evermore common pattern of behavior is inconsistent with the implicit assumption in this field, advanced by such writers as Jose Azar, Ioana Marinescu, Eric Posner, and Glen Weyl, who start from a quasi-Marxist point of view that the monopsony power of these large firms is sufficient to create a pool of “excess labor” that allows them to keep wages low. That point seems utterly improbable when signs like “employees wanted” sprout up everywhere. Indeed, the real problem in many cases is a mismatch between the skills that workers have and the jobs available.
The increased presence of licensing laws in many professions and occupations, including brokers, cosmeticians, doctors, lawyers, nurses, and pilots, can explain a large part of that mismatch. In some of these cases, it might be argued that the licenses are necessary to protect the public from incompetent and fraudulent practices. But it appears far more likely that they will further reduce labor supply rather than reduce worker power, which should only exacerbate current labor shortages.
There are ways to mitigate the effects of these licensing laws. The first is to switch from a system of licensing to certification, which gives consumers valuable information without imposing a formal barrier to entry. The second is to allow persons of good standing who are licensed in one state to practice their trade in another. Today, most licensing requirements are tied not to competence but territory so that a license to practice a given trade or business in one state does not carry with it the right to practice that same profession in another. Antitrust law is a poor tool to deal with these (government-imposed) restrictions. What is needed is a direct attack on the restrictions in question so that insurance companies can sell their products nationwide, and not only in states where they are licensed. For similar reasons, telemedicine is a far more effective tool if physicians licensed in state A can treat patients from their home office, even if the patient is located in state B.
The removal of these barriers to entry imposes no additional administrative costs while increasing allocative efficiency. That makes this direct attack on licensing a far more efficient means to address labor market concentration than a cumbersome antitrust attack, which would require answering difficult questions about the scope of the geographical and occupational markets in an effort to discover the effective scope of a firm’s monopoly power. Moreover, figuring out antitrust damages and fines is a fine art. Breakups after mergers are also subject to serious risks that they will undercut firms that achieved their dominant position by skill or luck, rather than by illicit combination or some other illegal practice.
Using the standard antitrust analysis, only very few markets are even susceptible to monopsony power. In a recent work, Elana Prager and Matt Schmitt offer a nuanced empirical analysis of the hospital market that yields a picture that is basically consistent with theory. Their basic strategy is to examine the effect of hospital mergers on wages by breaking out workers with industry-specific skills, such as separating nurses from those without industry-specific skills, such as cafeteria and janitorial workers. For the former group, the combination of union power and licensing requirements creates something of a bilateral monopoly situation in relation to hospital employers, where entry is limited by certificate-of-need requirements that limit the construction of new facilities. As a result, the monopoly power of each side tends to blunt that of the other, leading to only a modest contraction in wages. But for occupations hired by hospitals without industry-specific skills, the authors find little or no effect.
That result is a far cry from the claim that pervasive monopoly power can reduce (advertised) wages by up to 17 percent. The Biden administration cited for that claim an article by José Azar, Ioana Marinescu & Marshall Steinbaum, who treat advertised wages for “posted vacancies as a proxy for local demand.” From that (dubious) premise, they observe that the HHI indices for various industries run from a low 2,500 to, for a quarter of the firms, an astronomical 7,500. The high HHI stems from the wholly incorrect assumption that the markets for labor and products are identical, without asking how many applicants applied for jobs in firms in different industry categories. And, posted vacancies cannot provide a consistent indicator of market share—a new, small start-up could easily be hiring more workers than larger established firms, notwithstanding its small market share. What is needed is a reliable accounting of total compensation, which is exceedingly difficult to derive because it includes not just wages, but also fringe benefits, job security (which may be greater in established firms and may in turn partially depress wages), working conditions, and future prospects (which may be better in smaller firms). This complexity in labor markets makes an accurate determination far harder than measuring than the price of a gallon of gasoline, a fungible product with none of these attributes.
The Biden program also takes dead aim at covenants not-to-compete—signed by literally tens of millions of workers in such trades as retail and construction—arguing that they impose obstacles to workers’ ability to move to higher-paid jobs. But that point is seriously overstated. Legally, these covenants not-to-compete are subject to a rule of reason test, which constrains enforceable covenants in three key dimensions: typically, these covenants cannot run for more than one year; they apply only to existing lines of businesses and not new ones that the employer may wish to establish, and they only apply within the geographical market in which the employer currently does business. These limits are imposed to make sure that the former employee is still able to become a competitor to the former boss. It takes powerful reasons to obtain enforcement of more restrictive covenants. But, on the other side of the rule of reason calculus is the awareness that the former employee should not be allowed to use confidential information—customer lists, marketing plans, and trade secrets—free of charge against the employer who generated these benefits at high investment costs. Competition is never fair if the new entrant receives a hefty in-kind subsidy at the expense of the established player.
The three conditions mentioned above are a sensible effort to give clear guidelines to the relevant tradeoffs. For high-paying jobs, it is a close call whether these covenants should be enforced. Thus, there is a tendency to favor the free mobility of labor within, say, the high-tech field, in those cases where protected trade secret interests are not clearly invoked. At the bottom end of the market, the situation is quite different. With rare exceptions of doubtful legality, the standard covenants prevent the worker from taking a new job, but only at another franchisee of the same franchisor. And these covenants make perfect sense. If a company such as McDonald’s owned all of its outlets, it could control workers’ movement from one outlet to another to keep its costs down. Once the company uses franchisees, it has the same interest in preventing its independent franchisees from bidding up the labor cost in ways that could make the franchisor’s overall products less competitive. But, these workers are free to go to the employees of franchisees of different franchisors (e.g., Burger King) even if their departure causes the original firm to raise wages in order to attract substitutes. There is little risk, moreover, that employers will seek to intimidate naïve workers with overly broad covenants when the new employer can assure them that they are not enforceable. The competition, therefore, remains intense, which is why media stories recount on a daily basis the “fierce competition” for workers. Indeed, Biden has been chastised for citing McDonald’s to claim that these covenants are more pervasive than they really are. McDonald’s never imposed restrictions on movement to competitors and removed its limitations on movement among McDonald’s franchisees in 2017.
In sum, both parts of the Biden program for heightened antitrust enforcement in labor markets fail. The tragedy here involves, at the very least, high administrative and compliance costs to deal with a largely nonexistent threat. Thus, on the labor monopsony theory, it would be the rare case in which high concentrations in limited labor markets should be relevant to efforts to block mergers where product-market concentration levels are already high. And in certain hospital markets, high labor concentrations are an added fillip given that these local concentrations in service markets enhance the ability to examine such mergers. Needless to say, it would be the height of madness for the Department of Justice and Federal Trade Commission to look at supposed monopsony power in large firms wholly outside the merger context. The modern antitrust crusade in labor markets has the real potential to chill innovative business activities.
The time to abandon the new clarion call is now. The traditional forms of antitrust analysis can perfectly deal with labor markets without succumbing to Biden’s ill-conceived notions of how best to promote competition. What is truly needed is a close examination of the other impediments to labor, including the full range of anticompetitive laws dealing with minimum wage, overtime, family leave, anti-discrimination, and the panoply of labor union protections, where the gains to deregulation should be both immediate and large. Indeed, the best feature about the Biden Executive Order is that it does direct the FTC to address these licensing and regulatory barriers to entry. Once again, the anti-monopsony approach comes out second-best in labor markets, such that the rest of his initiative should be scrapped.