The Rigged Labor Market
alan krueger, a professor of economics at Princeton, served as chairman of President Obama’s Council of Economic Advisers and a member of his cabinet from 2011 to 2013. Previously, he was assistant secretary for economic policy and chief economist at the Treasury Department (2009-10) and chief economist at the Department of Labor (1994-95).
Published April 28, 2017
The opinions expressed are solely those of the authors and do not necessarily represent the views of the Institute.
More than two centuries ago, Adam Smith, now widely regarded as a cheerleader for free market capitalism, raged that the market for labor was rigged. He argued that self-interested employers manipulated the labor market to drive workers’ wages below their competitive level. Smith warned that employers “are always and everywhere in a sort of tacit, but constant and uniform combination, not to raise the wages of labor above their actual rate.” And he ridiculed naysayers who denied that employers colluded to press their advantage against workers “as ignorant of the world as of the subject.” He further noted that “we seldom, indeed, hear of this combination, because it is the usual, and one may say, the natural state of things, which nobody ever hears of.”
The opinions expressed are solely those of the authors and do not necessarily represent the views of the Institute.
As in many other areas, Adam Smith’s insights were spot on and prescient. The conspiracy that he warned us about is alive and well in the 21st century, and it still receives little attention from policymakers, economists or the media.
Much research in labor economics over the past quarter century has confirmed Adam Smith’s fear that employers routinely use anticompetitive practices to reduce pay and curtail worker mobility. Research has further expanded our understanding of how “frictions” in the job market, such as imperfect information or costs associated with changing jobs, can give employers anticompetitive advantages even in situations where many employers are vying independently to hire workers. New practices have emerged to facilitate employer collusion, such as noncompete clauses and no-raid pacts, but the basic insights are the same: employers often implicitly, and sometimes explicitly, act to prevent the forces of competition from enabling workers to earn what a competitive market would dictate, and from working where they would prefer to work.
This was clear in professional sports before free agency enabled athletes to earn what the market would bear. And it remains the case in many other less glamorous and lower-paying industries. Even professional economists have faced a rigged labor market. The heads of several leading U.S. economics departments used to regularly confer at the Annual Meeting of the American Economic Association to jointly agree upon starting pay and teaching requirements for assistant professors until the Justice Department started nosing around and raising concerns about the legality of the practice.
In one sense, this is encouraging news. It suggests that relatively simple corrective measures would improve the lot of many workers without undermining productivity, growth or job creation. Indeed, countering anticompetitive employer behavior could be a win-win for the economy, raising both pay and employment. But it would take a fundamental shift in perceptions of how labor markets work — one that is a hard sell in an America seemingly disinclined to look closely at how everyday labor practices undermine workers’ prospects for moving up the income ladder.
The opinions expressed are solely those of the authors and do not necessarily represent the views of the Institute.
Let’s Make a Deal
Adam Smith could not possibly have anticipated Buzz Lightyear, Nemo or Dory, but the animated movie industry provides concrete evidence of the conspiracy that he warned about. In early 2017, the Walt Disney Company and its subsidiaries Pixar, Lucasfilm and ImageMovers, became the last of the major film companies to reach a settlement in an antitrust suit brought on behalf of movie animators. Along with Sony, Blue Sky and DreamWorks, they agreed to pay $169 million to settle charges that they conspired to suppress compensation by agreeing not to solicit each other’s employees, to take special procedures when contacted by each other’s employees, and to coordinate compensation policies through direct, collusive communications.
The evidence against the film studios, captured in emails, internal documents and sworn testimony, suggests that it was business as usual to conspire to avoid bidding away employees from competitors and to coordinate on pay setting to keep a lid on compensation costs. The conspiracy apparently began with a “gentleman’s agreement” between Pixar and Lucasfilm to avoid bidding wars over employees. George Lucas testified that, as a matter of policy, Lucasfilm “would not actively go out and recruit from other companies.” Emails from Lucasfilm human resources personnel indicate that the company even withdrew job offers to Pixar employees whom Pixar deemed “essential.”
For its part, Pixar refused to be drawn into bidding wars with its competitor. In an internal document, Pixar stated it would “never counter if the candidate comes back to us with a better offer from Lucasfilm.”
The collusion encompassed other major filmmakers as well. An email from a Pixar executive to Steve Jobs, for example, noted that Pixar had an “agreement with DreamWork[s] to not poach their people.” And a 2006 list of Lucasfilm’s “gentleman’s agreements” stated that Lucasfilm would not “recruit actively or passively from DreamWorks … for any positions.” Emails also documented that Pixar, Disney, Lucasfilm, Sony and Blue Sky had gentleman’s agreements to not recruit each other’s employees.
The court ruling that certified the animation workers’ class action suit against the film companies concluded: “In addition to the documentary evidence that defendants agreed not to recruit from each other, the documentary evidence supports plaintiffs’ allegations that defendants colluded on compensation policies through industry surveys including the Croner Survey, annual closed-door in-person meetings, and emails.” Lucasfilm’s president, Jim Morris, once invited Pixar’s president, Ed Catmull, to participate in a salary survey because he knew Catmull was “adamant about keeping a lid on rising labor costs.” Not unlike the economics department chairs, the film company officials often met during industry conferences over meals to discuss compensation issues. At one such “intimate dinner” in 2006, human resources officials from DreamWorks, Disney, Pixar, Blue Sky, Sony and Lucasfilm collectively agreed on average salary increases. Human resources officials also routinely shared their employees’ salary ranges with their counterparts at other companies.
The antitrust action on behalf of the film animators had its roots in a well-publicized Department of Justice investigation of antipoaching practices used by the tech companies Adobe, Apple, Google, Intel, Intuit and Pixar. That investigation revealed some colorful evidence. For example, after Google’s co-founder Sergey Brin tried to hire a programmer from Apple’s browser team, Steve Jobs wrote in an email, “If you hire a single one of these people, that means war.” When Intel’s chief executive, Paul Otellini, was asked by a recruiter about the no-raiding agreement his company had with Google, he responded in words that would ring familiar to Adam Smith: “We have nothing signed. We have a handshake ‘no recruit’ between Eric [Schmidt of Google] and myself. I would not like this broadly known.” The Justice Department complaint was quickly settled in 2010, with the tech companies agreeing to avoid “pressuring any person in any way to refrain from soliciting, cold calling, recruiting or otherwise competing for employees of the other person.”
Subsequently, a separate class-action civil suit was brought on behalf of more than 64,000 software engineers and other employees of Adobe, Apple, Google, Intel, Intuit, Pixar and Lucasfilm. The evidence that the tech companies colluded to restrict labor market competition was overwhelming. Before going to trial, the various parties settled. In an unusual move, Lucy Koh, the judge hearing the case, ruled that the initial cash settlement was inadequate. The lawsuits were eventually settled for a total of around half a billion dollars in 2015.
High-tech employees are not the only ones to have won civil suits alleging anticompetitive conduct by employers in recent years. Several suits have been successfully brought on behalf of nurses against hospitals, for example. On Sept. 16, 2015, the Detroit Medical Center became the last of eight major Michigan hospital systems to reach a settlement in a suit alleging that the hospitals colluded to reduce their pay. The hospitals apparently endeavored to share information about nurses’ salaries and pay increases. With pay pushed below competitive levels, the hospitals often turned to temporary staffing firms to hire workers (at salaries above those of existing staff) and made do with vacancies.
Similar cases are in various stages of resolution in Albany, Chicago, Memphis, San Antonio and Arizona.
You’re Mine for a Long Time
Another tactic that employers frequently use to tilt the balance of bargaining power in their favor and restrict competition in the labor market is to require employees to sign noncompete agreements that prevent them from working for rivals for an extended period. Note at the start that mutually-agreed-upon noncompete agreements can sometimes be justified as a means to increase labor productivity. They may give employers incentives to share trade secrets with employees, secure in the knowledge that they won’t decamp for a competitor. And noncompetes may give employers and employees incentives to invest in expensive firm-specific training. In both these situations, workers could presumably negotiate for higher compensation in exchange for signing a noncompete agreement.
However, for a vast majority of employees whose work does not entrust them with trade secrets or provide much training, such agreements undoubtedly restrict worker mobility, reduce labor market competition, and suppress pay and productivity by preventing workers from moving to jobs that offer better compensation and working conditions. Nonetheless, according to survey research by Evan Starr of the University of Maryland and J.J. Prescott and Norman Bishara of the University of Michigan, 18 percent of American workers are currently constrained by noncompete clauses. Moreover, nearly 40 percent have signed noncompete agreements at some point in their careers. These figures greatly exceed any plausible estimate of the share of workers with access to trade secrets that could justify noncompete agreements. And the fact that a 2016 Treasury Department report found that the age-earnings profile is steeper for workers in states that do not enforce noncompete agreements than in states that vigorously enforce them strongly suggests that noncompete agreements do not lead to greater job training; instead, they suppress wage growth.
Although noncompete agreements are more common in higher paying jobs, there are plenty of instances in which low-paid service workers are caught in the net. Amazon, for example, requires its warehouse workers, including seasonal hires, to sign an agreement that seems to cover everything but the kitchen sink:
During employment and for 18 months after the separation date, employee will not … engage in or support the development, manufacture, marketing or sale of any product or service that competes or is intended to compete with any product or service sold, offered or otherwise provided by Amazon.
Noncompete agreements have even become common in the fast food industry. For example, the sandwich chain Jimmy John’s, with 2,000 restaurants, used a noncompete clause that prohibited its employees from working at any other restaurant that sells “submarine, hero-type, deli-style, pita and/or wrapped or rolled sandwiches” within two miles of a Jimmy John’s shop while they were employed at a Jimmy John’s and for up to three years afterward. The company agreed to drop this practice in New York and Illinois last year after being challenged by the attorneys general in those states.
While Jimmy John’s prohibited its employees from working for competitors, McDonald’s came at the issue from the other side. The company prohibits its own franchisees from hiring workers away from other McDonald’s restaurants; nor can they hire workers who left another McDonald’s within six months. With approximately 14,000 McDonald’s restaurants employing 420,000 workers across the United States, this restrictive hiring covenant could significantly curtail the freedom of McDonald’s workers to seek better pay and working conditions.
How Labor Markets Work in Practice
In the idealized version of a perfectly competitive labor market, many employers freely compete to hire workers from a large pool. Thus, neither employers nor workers have “market power,” meaning the ability to dictate terms of employment that differ meaningfully from the terms offered (or received) by others to workers with comparable skills. In this world, employees can freely (and costlessly) change jobs when better opportunities arise. For their part, employers are able to seamlessly fill job vacancies by simply offering the going wage rate. This wage is determined by the market — by the intersection of the supply and demand curve for labor; nobody has discretion to set pay. Thus, in econ jargon, the supply curve to any firm is “infinitely elastic,” implying that if it paid just a penny below the going wage it would lose its entire workforce, and if offered just a penny more it would be inundated with able job applicants. An individual employer in a competitive labor market is not subject to the usual law of supply, as there is no upward-sloping supply curve to the firm: paying the going wage attracts all the workers the firm demands.
The opposite of a competitive labor market is one with a single employer. That employer, called a monopsonist, is in a peculiar position because she is subject to the law of supply in the way that an entire industry is in a competitive market. If she wants to hire an extra worker, she must pay a somewhat higher wage since the supply curve she faces is sloped upward. And, as a practical matter, she will need to pay that higher wage to the rest of her workers, too. Thus a monopsonist with, say, 100 employees who finds it necessary to pay an extra dime an hour to hire one more worker, must shell out a total of $10 an hour (100 times 10 cents) more to the existing workforce as part of the price of hiring that additional worker.
Note the critical difference: an employer in a perfectly competitive market, who is one among many, can always hire more workers at the currently competitive wage. So, for this employer, the added cost of one more worker is exactly what she pays that worker. A monopsonist’s marginal cost of hiring a worker, however, is higher — perhaps much higher — than the wage paid to that additional worker.
Thus, paradoxically, while a monopsonist is likely to pay less, on average, for labor, than an employer in a competitive market, the monopsonist’s cost of labor at the margin is likely to be higher. So, in aiming to maximize profits, a monopsonist will hire less labor and make do with vacancies. And an economy full of monopsonists will be less productive because employers will fail to hire workers who could contribute more value to output than they received in wages.
If the government forces a monopsonist to increase the wage that it pays — by, say, imposing a minimum wage that is modestly above the wage it currently pays — the monopsonist’s marginal cost of labor will fall. And this cost at the margin may become low enough to give the monopsonist an incentive to hire more workers. In other words, without a minimum wage the monopsonist operates with vacancies, unwilling to raise the wage it offers to hire additional workers because it would have to pay that higher wage to existing workers as well. However, with a binding minimum wage — that is, a minimum wage above the rate the monopolist was already paying — a monopsonist can fill its vacancies without worrying about having to increase everybody else’s wages, because that was already required by the minimum wage. A monopsonist would not be happy with this situation because the minimum wage would cut its profits. But once there is a minimum wage, the firm would find it possible and in its interest to fill its vacancies, provided the minimum wage wasn’t set too high.
The canonical example of monopsony is a one-company town — say, a remote coal mining town. While few labor markets are characterized by pure monopsony today, if employers collude to suppress competition — either by restricting labor mobility or conspiring to fix pay — they jointly exercise monopsony power. This should be evident because employers are exercising discretion to set pay, and in a perfectly competitive labor market, employers have no discretion whatsoever to set pay.
Although it remains unclear how pervasive or effective employer collusion is in restricting competition, recent economic analysis suggests that employers often have monopsony-like power over workers even when there are many employers who operate independently. In part, that’s because employers have some discretion over what they pay because of labor market frictions, such as job search costs.
For example, in the “wage-posting job search” model pioneered by the economists Kenneth Burdett at Penn and the late Dale Mortensen, and expounded on in Alan Manning’s book Monopsony in Motion, employers consider the tradeoff between their position on the wage ladder and their employee turnover rate. Employers can choose a low-wage strategy and make do with high turnover and chronic vacancies, or a high-wage strategy with low turnover and few vacancies. Requiring employers who choose the low road to increase their pay would raise total employment by making jobs more attractive and reducing turnover.
In another model of the labor market developed by Mortensen, Chris Pissarides of the London School of Economics and Chris Flinn of NYU, firms and workers bargain over the value that each unique worker-firm match creates. Factors that determine their relative bargaining power, such as the ability of workers to take other job opportunities, influence where the bargain ends up.
Yet other models recognize that, even absent search frictions and idiosyncratic value of worker-firm matches, individual workers value nonwage features of jobs differently. For example, some workers may live very near a McDonald’s where they work. Those who live close by would presumably be willing to work for a lower wage at that outlet than those who live far away and must spend time and money on commuting.
It turns out that this very minor and realistic tweak coverts a perfectly competitive labor market to one in which employers have monopsony-like power because firms no longer face a perfectly elastic labor supply curve. If a firm were to cut its wage in this situation, it would find that some workers would leave (namely, those with a long commute), while others would stay because it is more convenient working at this restaurant than another one. To recruit new workers, every firm would have to pay more than the going wage (to attract workers who live farther away), just as in the case of a single monopsonist. And, as in the case of a monopsony, a minimum wage could lead to higher employment. The takeaway: monopsony-like incentives can intrude on what might otherwise look like a competitive labor market, even when there are many employers.
The details of these models are less important than the general result. A perfectly competitive labor market operates on a knife edge, on which employers have no choice but to pay the going market-determined wage. If they collude to restrict competition and suppress pay, as Adam Smith predicted, or if very common features of the labor market give them discretion over what they pay, they will have incentives to pay less and hire fewer workers. Smart interventions to redress the imbalance between companies and workers, such as a prudently set minimum wage or vigorous enforcement of antitrust restrictions on anticompetitive employer behavior or steps to enhance employee bargaining power, can result in more employment as well as higher pay.
The reality that many, if not most, markets for labor have some of the characteristics of monopsony helps explain a persistent puzzle in labor economics: studies have not found strong or even consistent evidence that a higher minimum wage reduces employment, as would be expected to be the case in a perfectly competitive labor market. Indeed, the most comprehensive summary of the recent literature, by Dale Belman of Michigan State and Paul Wolfson of Dartmouth, finds that the average study shows essentially no effect of minimum wage increases on employment. Although some companies may respond to a higher minimum wage by reducing employment, this is apparently offset by others who increase employment — as would be expected if some firms have monopsony-like power.
Other evidence at the national level suggests monopsony-like power is restricting pay increases. For example, job openings and posted vacancies have steadily risen in recent years, but hiring has not kept pace. This pattern is sometimes cited as evidence of an economy-wide shortage of skilled workers. Yet, in a competitive labor market, one would expect the phenomenon to be fleeting: employers would bid up wages for skilled workers until vacancies were filled or firms were discouraged from posting vacancies because of the higher competitive wage. Consider, too, that contrary to what one might expect in the face of skill shortages, wages have not grown faster in sectors with rising job openings. This suggests that companies are resisting raising wages as a means to attract more workers — as one would expect in an imperfectly competitive market.
Labor Economics 2.0
An important first step in rethinking how labor markets work — and how they could be made to work more efficiently and fairly — is to change the default assumption from the perfectly competitive model we all learned in Econ 101 to one in which employers have some discretion to set pay. This discretion can come about because, as Adam Smith predicted, it is the “natural state of things” for employers to collude, or because of frictions in the job market. We can begin to change the default by changing how we teach economics. Introductory economics textbooks rarely discuss the role of monopsony or collusion in the labor market, although market power is standard fare when it comes to product markets.
While it is difficult to determine at this stage if the antitrust cases brought on behalf of animators, software engineers and nurses indicate widespread employer collusion, the fact that the behavior spans diverse industries and employee skills suggests that collusion may be common. At a minimum, the accumulation of cases suggests that more resources should be devoted to enforcing the antitrust laws when it comes to employers’ HR practices.
Toward the end of Obama administration, the Department of Justice and the Federal Trade Commission issued new guidelines for human resources professionals to help identify and report wage fixing and hiring collusion among employers, and created a hotline to report instances of collusive behavior. Moreover, for the first time, the Justice Department threatened to bring criminal (as opposed to civil) cases against individuals and their companies for entering into no-poaching and wage-fixing agreements with competitors. The Obama administration also called on the states to adopt a set of best practices to ensure that noncompete agreements are narrowly targeted and appropriately used.
With the change in presidential administrations, though, the fate of these initiatives is uncertain. While running for president, Donald Trump was fond of declaring, “We will never be able to fix a rigged system by counting on the same people who rigged it in the first place.” He used this line to rail against his opponent, the political system, big donors, bureaucrats and big businesses. Although the fact of his election suggests the political system is not rigged in quite the way he claimed, he undeniably tapped into a significant vein of discontent — and, to mix metaphors, a large grain of truth — in arguing that the economy is “rigged against you, the American people.”
If he is serious about addressing this problem, there would be no better place to start than by unrigging the labor market by (a) raising the federal minimum wage, which has remained at $7.25 an hour since 2009 and which Mr. Trump pledged to raise to $10 an hour during the campaign; (b) reining in the excessive use of noncompete clauses; (c) vigorously enforcing antitrust laws to prohibit employer anticompetitive practices; and (d) increasing worker bargaining power.
If President Trump does not try to reverse the ways in which the labor market is rigged against American workers, there is an alternative: the states could step up. Many have already begun to raise their own minimum wages and to prohibit the use of noncompete agreements for low-wage workers, and their attorneys general have been on the vanguard at challenging excessive use of noncompete clauses. The states could also pursue means to improve worker bargaining power and pass tougher laws to penalize employer collusive behavior that restricts worker mobility and suppresses pay.
After all, workers deserve a break too.