The Economic Basis of the Independent Contractor/Employee Distinction

In recent years, a controversy has erupted over the distinction between employees and independent contractors. Commentators have argued that in the modern gig economy, many people traditionally classified as independent contractors are as vulnerable as employees and should be granted the legal protections that employees alone normally enjoy. However, the distinction between the two categories remains inescapable, and the theoretical basis for it has not been identified. I argue that the distinction is derived from market structure. Employees are workers who, because they must make relationship-specific investments in a single firm, are subject to labor monopsony. Independent contractors do not make such relationship-specific investments, and hence normally operate in a competitive labor market. Employment and labor law may be explained as a method for protecting workers from labor monopsony; because independent contracts are not subject to labor monopsony, they do not require such protection, and in fact may impose costs on others if granted it.

Introduction

The law’s distinction between employees and independent contractors (or, merely “contractors”) has sparked intense debate over the last few years. As a result of advances in technology, some workers who have traditionally been classified as employees are now being treated as contractors: they make a living by undertaking a series of “gigs” for different “labor buyers” (as I will call the firms or households that purchase the services of workers) rather than working for a single employer.^[1] Many commentators worry that these gig contractors are being exploited because they are not entitled to the protections of employment law;^[2] a related view is that workers who are really employees are being deliberately misclassified as contractors by rapacious employers for the same reason.^[3] Commentators have proposed numerous reforms designed to bring contractors, or a subset of them, under the protection of employment law.^[4]

While new technology and employment trends have highlighted these problems, they are not new. The distinction between employees and contractors is deeply entrenched in the law and reflects a basic intuition about the organization of labor markets. Consider, for example, a person trained as an electrician. She might choose to set up her own business. She advertises her services and spends her days working for various homeowners who pay her to repair the fuse box or install new lighting. The electrician seems like a business owner, not an employee, and indeed she would be legally classified as an independent contractor. Or she might go to work for, say, a company that manufactures electric turbines. She shows up at a worksite every day at 9 a.m., leaves at 5 p.m., and draws a salary from a single firm. Here, she would be classified as an employee. As an employee, the electrician would be protected by numerous federal and state laws that control wages, working conditions, and benefits. If instead she works as a contractor, she would enjoy none of these legal protections.

Because the same person doing the same type of work might be self-employed or an employee of someone else, the distinction between employee and contractor can be elusive. The distinction is made more complicated still by the administrative requirements that have grown up around it. Because contractors often charge for a job rather than by the hour, it may be difficult to calculate an hourly wage and, thus, to apply the minimum wage laws to them. Because contractors often work alone, it might seem that the right to organize a union would do them no good. Because contractors often choose their own tools and control working conditions, it would make little sense to compel those who buy their labor to comply with legal requirements for workplace safety. Contractors often work for homeowners and other consumers who lack the legal sophistication and administrative resources for complying with the huge number of legal restrictions that apply to employers, including the obligation to withhold taxes. And, until recently, it was common to think of contractors as highly trained professionals—electricians, plumbers, lawyers, doctors—who were not as vulnerable to mistreatment by buyers of their labor than ordinary employees were. Contractors did not seem to need employment-law protections.

But technology has put pressure on these intuitions. We now see that companies can organize their businesses so that drivers, janitors, and home healthcare workers are classified as contractors rather than employees. Compensation can be structured so that it is hourly or based on the accomplishment of tasks; control over working conditions can be assigned to the worker, retained by an organization, or divided between them; and organizations can match workers with consumers so that consumers, rather than the organization, seem like the employers. Organizations can knit workers together into loose teams, keep them apart from each other, or use them as a conventional workforce; they can assign the price-setting power to workers or keep it for themselves. Or both: Uber normally sets prices for drivers but recently allowed drivers in some cities to set their own prices up to five times a base price.^[5] With the scrambling of categories, the intuitions have lost their force, and we need to look deeper for the policy reasons behind the distinction between employee and contractor.

I argue that contractor or employee status, properly understood, depends on the nature of the interaction among the worker, assets under the worker’s control, and the labor buyer who benefits from the worker’s manipulation of those assets. As a rough approximation, I distinguish between what I will call “discrete” and “relational” work. Work is discrete when its value is best exploited by a worker acting alone (in the sense of not being subject to the control of another), in most cases bringing to bear her labor on an asset or various assets that she owns. Work is relational when its value increases with the degree of coordination between the worker in question and other workers, including peers, subordinates, and superiors, all in relationship with assets that the worker uses but does not own.

Because relational workers coordinate with others in an organization (that is, a firm or business), they normally “invest” (in the sense of incurring costs in time and money) in personal relationships with those other people, and in learning to manipulate the assets controlled by the organization. As they do so, they become more productive for the firm in question relative to other firms, and thus they earn higher wages at that firm than they could at alternative firms. This is what is meant by labor monopsony: because such workers cannot earn comparable wages from competing employers, those workers lack a credible threat to quit if their wages stagnate and conditions worsen at the margin. The person or firm who buys the labor obtains an advantage as the only buyer of the worker’s services. Employment and labor law step in to offer these workers protection: they are effectively a form of price (wage) and conditions regulation that counters the downward pressure of monopsony. By contrast, discrete workers are not subject to labor monopsony (except when the market is concentrated, as discussed below). If one buyer of their work refuses to pay them their market price, those workers can exit and find another buyer who will. Employment and labor law will either not help these workers or will help them only by enabling them to extract rents from others—as I will explain in due course. Thus, classification law performs the important task of segregating workers into the class that should receive the protection of employment and labor law, and those who should not. The first group are classified as employees and the second group are classified as contractors.

The legal tests for distinguishing employees and contractors are famously ambiguous. They involve many factors that often point in different directions and that firms can manipulate. The main problem is not the factors themselves but the failure of courts and regulators to understand how those factors relate to the goal of labor market regulation. The discrete/relational distinction provides guidance for identifying workers as contractors (discrete) or employees (relational).

Consider, for example, the controversy over Uber’s treatment of its drivers. Uber claims that the drivers are contractors and that Uber merely matches them with customers, like a dating service.^[6] Many drivers argue that they are employees, and thus should be protected by minimum wage and related law, and should be allowed to organize a union.^[7] Litigation and commentary have focused on whether drivers “control” their worksite, but because the concept of control can be understood only in relation to a normative goal, gallons of ink have been spilled on an essentially unanswerable question.^[8] I argue that the answer depends on whether Uber enjoys a monopsony over drivers because they engage in relational work for it. I will discuss this issue in subpart II(B).

An enormous body of academic literature has developed in the wake of the controversies over Uber and other gig-economy companies. Most commentators argue that these companies have sought to avoid employment law protections and advocate an expanded definition of employment so as to sweep in gig-economy workers, or a third category such as the “dependent contractor,” who would be entitled to certain protections.^[9] However, no one has provided a satisfactory theoretical argument for the distinction, and without explaining in the first place why the distinction between contractor and employee, multiplying categories will only multiply confusion. By rooting employment law protections in the problem of monopsony, I can offer a test that is (by the standards of law) clear and stable.^[10]

Thus, my thesis. In the absence of law other than basic contract and property law, (1) some workers will end up in competitive labor markets and others will end up in monopsonized labor markets, based on the type of work they do (discrete or relational). (2) The workers in competitive labor markets should not be protected by employment and labor law; the workers in monopsonized labor markets should be. (3) Firms that misclassify employees as contractors cause social harm by evading restrictions on labor monopsony power. Their behavior should be seen as arbitrage, a way to evade laws that have been developed to reduce the harms from monopsonistic labor market competition.^[11] (4) The classification test should be based either on whether workers are relational or not, or, more broadly, the extent to which they are subject to labor monopsony. I will discuss this choice in Part II. To be clear, my contribution is not the insight that employment and labor law may counter labor monopsony. This idea has a long history. The purpose of this Essay is to bring to bear the literature on labor monopsony on the misclassification debate, and in particular, to criticize and reform the classification test.

My argument can be contrasted with the dominant view about the misclassification controversy in the law review literature. That view sees employment and labor law mainly as devices for protecting low-income workers and argues that these protections should be extended to contractors who are similarly vulnerable.^[12] Some commentators thus argue that the labor exemption in antitrust law—which allows employees to organize—should be extended to contractors.^[13] I argue that employment and labor law are not oriented toward wealth distribution but rather toward efficiency—by countering labor monopsony. These laws thus should not apply to competitive labor markets, and the purpose of classification law is to prevent them from doing so.

I. Discrete and Relational Work

A. Employee and Contractor

At least as far back as Ronald Coase’s paper on the “Nature of the Firm,” published in 1937, economists have identified the distinguishing feature of employment as the employer’s control over the worker.^[14] An employee is “in” the firm because the managers of the firm can issue orders to her and expect her obedience. A contractor is “outside” the firm because the managers of the firm can elicit cooperation only through a negotiated bargain that is embodied in a contract that governs a specific task. The idea of control has also been central to the common-law definition of employment (or master–servant) relationships for centuries.^[15] But the nature of control has turned out to be elusive. Criticizing Coase’s reliance on control for the definition of the employment relationship, economists Armen Alchian and Harold Demsetz pointed out that a customer exerts control over a contractor as well.^[16] For example, if a grocer refuses to do what a customer wants—stock a certain product, for example—the customer may stop patronizing the business. The fear of losing customers puts the grocer under customers’ control—just as the stock clerk’s fear of being fired causes her to obey the grocery store owner’s order to fill the shelves with one product rather than another.

A moment’s reflection, however, reveals the problem with this argument. Suppose a customer tells a grocer to move the candy bars to a shelf where children will not see it, or to sweep the floor because it is too dirty, or to change the window display. Even at risk of losing the customer, the grocer is likely to tell him to get lost. By contrast, a grocer could certainly tell her stock clerks to do any of these things. While the stock clerks could refuse and quit, most likely they would obey the grocer’s directions. The control that an employer exerts over an employee is different from the kind of control that a customer exerts over a contractor.

A line of literature has made progress with the notion of control, rooting it in the idea that contracts—including employment contracts—can never fully specify the optimal actions of the parties, and so unavoidably allocate discretion among them.^[17] As a rough approximation, contracts between labor buyers and workers that allocate key aspects of discretion to the labor buyers create employment relationships, while those that allocate those aspects of discretion to the workers create contractor relationships. In the balance of this section and the next, I draw on this literature. Then I will turn to its implications for the misclassification test.

To understand what “control” means in the employment context, consider a simple example. A person we will call the Rider needs a car and driver in order to travel around town. The Rider may choose between two arrangements. Under the first, the Rider owns a car and contracts with a person, the Driver, to serve as her chauffer for a period of time. Under the second, the Rider does not own the car; the Driver does. The Rider and the Driver enter a contract under which the Driver agrees to drive the Rider around for a period of time.

The two contracts look nearly identical, and it is easy to imagine that, in practice, the “output”—the routes used by the Driver, the amenities of the service, and so on—is identical under both contracts. If the Driver is capital-constrained, the Rider could lend him the money to buy the car, and now the two contracts seem even more similar. We can further imagine that the parties, in each case, enter into highly detailed contracts that specify numerous attributes of the relationship—when the Driver must show up every day, how far the Driver must drive, even the routes and any chitchat that will take place while the Driver and Rider share the car.

Yet, there is an important difference. When the Rider owns the car, the Rider enjoys “residual control”—meaning control over how the car is used where the contract fails to specify the Driver’s obligations. And when the Driver owns the car, the Driver enjoys residual control. As a concrete example, imagine that the parties enter into a 100-page contract that specifies nearly every aspect of the work relationship but omits, say, whether the Driver can talk on the phone with friends while driving the Rider. If the Driver does talk on the phone, and the Rider objects, the assignment of residual control matters. If the Driver owns the car, and the Rider tells him to stop talking to his friends, he can simply refuse because he is not prohibited from doing so under the contract. If the Rider then refuses to pay, she breaches the contract. She could offer to pay him more to stop talking to his friends, and he may agree, but this additional bargain is costly to negotiate, and the Driver has the bargaining power. If the Rider does not have good alternative people to hire as a driver (and is reluctant to approach a stranger who has not acquired what we will call relationship-specific knowledge), the Driver can “hold up” the Rider for a high, additional fee.

If the Rider owns the car and tells the driver to stop talking to his friends, the Driver is required to obey. He does not have residual control over the car and cannot use it in ways prohibited by the owner. The Rider can order the Driver to stop the car and leave. Of course, the Driver could quit. But now, the Rider has the bargaining power. The Driver might prefer to work and might be willing to accept a reduction in wages in return for the right to speak, but now, he has to pay the Rider for this right. If the Driver has made a relationship-specific investment—preferring to drive this Rider rather than other people because he has learned her needs and doesn’t want to have to learn the needs and idiosyncrasies of a new boss—he will need to make concessions.

The parties will assign the property interest (that is, car ownership) to the party whose use of residual control is more likely to maximize the surplus of their interaction. Consider first the Driver. If the Driver owns the car, then he will be given strong incentives to maintain the car. He will drive it carefully, take it to the garage frequently, and so on. He will anticipate that, in the future, the relationship may end, and he will want the car in good working order as he searches for new customers. We will call this behavior “worker care.”

The problem with maximizing worker care is that the worker may have correspondingly weak incentives to follow the Rider’s orders. The Rider might order the Driver to take the fastest route, and the Driver might prefer not to because the fastest route is a potholed road that is hard on the car. The Driver might, therefore, choose another route. The Rider might be willing to tolerate a less-than-perfectly-maintained car in return for greater obedience. We call the Rider’s concern “worker coordination”—by which we mean the value of coordination between the worker’s action and assets owned by, or other workers managed by, the labor buyer, here the Rider. To ensure worker coordination, the Rider requires “managerial direction” (or “managerial discretion” to emphasize the freedom to direct the Driver); specification of optimal Driver behavior by contract is impossible.

Thus, when the parties negotiate the contract, they will trade off these two values. Car maintenance may not be a serious issue because maintenance can be specified by contract, or because the car is well-constructed and poor maintenance will not significantly reduce its value. In that case, the parties will assign the car to the Rider. Otherwise, the parties will assign the car to the Driver. In this economic sense, the Driver is an employee in the first case and a contractor in the second.

A useful, real-world example comes from a paper by economists George Baker and Thomas Hubbard, which examines the shift from the owner-operator model of trucking to the company-driver model.^[18] Owner-operators are independent contractors; they own their trucks and contract with shippers. Company drivers are employees; they work for a single, usually very large, shipper. For truckers, “worker care” is the same as in our example of the Driver-Rider relationship: “Wear and tear on the truck is minimized when drivers drive at a steady and moderate speed, but drivers may prefer to drive fast and then take longer breaks because it allows them to rest longer, visit friends, etc., and still arrive on time.”^[19] Worker coordination is also important because “hauls vary in their desirability to drivers in ways that are not captured in agreements with carriers. Those that take drivers into congested or dangerous areas are less desirable,” as do those that take drivers to remote areas where there are no desirable “backhauls”—that is, return trips with a new load of cargo.^[20] The problem faced by the shipper and the trucker is how to optimize care (minimizing wear and tear) while ensuring that the best routes are taken.

When drivers operate as contractors, they have strong worker care incentives: they drive their truck carefully because they own it. But they are reluctant to take less-desirable routes, and thus dispatchers—who coordinate routes—must negotiate with them case by case, which is disruptive. When drivers operate as company drivers (employees), they have weak worker-care incentives: the company bears the maintenance cost. But the company can simply order drivers to take the route that is optimal for the company. Here, we see the basic tradeoff in action. Baker and Hubbard hypothesize that longer hauls create greater problems for worker effort because there are more opportunities for the driver to engage in the suboptimal speed-and-rest strategy, while the backhaul negotiation problem seems the same for long-haul and short-haul drivers.^[21] The hypothesis finds support in evidence showing that long-haul drivers are more likely to be owner-operators than short-haul drivers are. The authors also examine the effect of the introduction of on-board computers, which monitor driver performance, and thus reduce the worker’s ability to shirk on worker care. The evidence indicates, as one would expect, that the introduction of this monitoring technology substantially reduced the share of owner-operated (contractor) trucks.^[22]

To sum up, a labor buyer and a worker maximize their joint surplus by assigning the relevant property right to the worker when care is more important than other actions that require management direction and coordination, and otherwise to the labor buyer. In the first case, the worker is called a contractor; in the second case, the worker is called an employee. These factors in turn depend on how completely the parties can specify by contract the relevant actions of the worker. Where the worker can be given optimal or very good incentives to engage in care through contractual specification or technology, the labor buyer will employ the worker. Where the worker can be given optimal or very good incentives to engage in coordination with the assets and other workers through contractual specification or technology, the labor buyer will treat the worker as a contractor.

B. Relational Work and Discrete Work

A major difference between an employee (who interacts with assets owned by the firm as well as the firm’s other workers) and a contractor (who interacts with assets owned by the contractor) is that the employee makes a relationship-specific investment in the “firm,” which is to say in the assets owned by the firm (the labor buyer) and the other employees of the firm. The contractor does not make a relationship-specific investment or much more limited investment. This difference is related to two concepts we discussed earlier: contract specificity and worker coordination.

When a contract can be specified in a relatively substantial sense, the underlying reason is that the task in question is standardized—routine, predictable, the same in different contexts. For example, when a homeowner hires a locksmith to fix a lock, the task is fully specified by the customer (though usually on the basis of a custom that the contract references rather than through detailed contractual specification). The locksmith performs the same task from place to place and does not need to learn anything about the individual interests or needs of different homeowners. In other contexts, however, many tasks cannot be specified. An elderly homeowner who hires a personal companion to keep her company, maintain her records, deal with outsiders, and so on, will expect the companion to invest a great deal of time and effort in learning the homeowner’s needs, interests, and ways of doing things. As a result, the companion’s work is “relational,” in the sense that abilities that the companion obtains over time as a result of frequent interaction with the homeowner are not easily transferable to other contexts—so if the companion quits and goes to work for someone else, she will start again from the beginning with a new employer.

Thus, I distinguish between “discrete work” like the locksmith’s and “relational work” like the companion’s. The homeowner’s concern is “worker coordination,” the need to ensure that the companion will do just what the homeowner requires. The homeowner does not need, or needs less, worker coordination with respect to the locksmith, who relies on his own transferable training and experience when he fixes the lock. The locksmith is a contractor; the companion is an employee.^[23]

The nature of work—relational or discrete—does not necessarily depend on the profession. A locksmith might be a contractor or employee in different contexts. Let us consider a more complex example involving two lawyers. Andrew is a solo practitioner who drafts wills for clients. He operates out of an office that he rents from a commercial landlord. Most of Andrew’s clients see him only once or a small number of times over their lives. He has drafted wills for thousands of people. It takes him only an hour or so to draft a will in most cases.

Beth is a tax attorney who works full time in the family office of a noted tycoon. She works nine to five in a suite of offices with several other workers—accountants, investment advisors, and other lawyers. She works closely with this team and consults frequently with the tycoon and various family members. She does not work outside the office, and, like most other workers in this office, she has worked there for many years and is vague about when or even whether she might leave for a different job.

While Andrew and Beth may have attended the same law school and received the same legal training, their skills are by now quite different. Andrew’s body of knowledge—a deep understanding of testamentary law and related areas of law—allows him to offer identical services to numerous different people. His skill (or human capital) is general.^[24] Moreover, while his clients all have different financial resources and needs, along the relative dimension for which they hire Andrew—the disposition of their assets at death—the services that Andrew offers are very similar, one might even say commodified. We can thus think of Andrew’s work as discrete in the following sense: the value of the service he offers is independent of his relationship with the client—whether he has known the client for a long time, for example. Andrew’s work is discrete in another sense: its value to the client is independent of Andrew’s relationship with other workers. He works alone. The nature of Andrew’s skills—his knowledge of testamentary law plus his ability to listen to and understand his clients—enables him to offer discrete services to a range of clients.

Beth, like Andrew, possesses a body of knowledge that she brings to bear as she provides services for her client, the tycoon. When she first goes to work for the family office, she brings the same type of discrete skill to the operations of the office as someone like Andrew would. But after working for a few years for the client, the nature of the skill changes significantly. The hiring launches a relationship, and it is in the context of this relationship that most of Beth’s value is generated. Over time, she acquires a deep understanding of the client’s resources, needs, and idiosyncrasies. She also develops an understanding of the other workers—how they work together, what they need from her, the appropriate way to behave in the office, and so on. Unlike Andrew, Beth engages in primarily relational work in the following sense: the value of the service she offers is dependent on her relationship with the client and with her coworkers (and specific assets owned by the employer such as office equipment or, more plausibly, bespoke software and records). As a rough approximation, the longer and deeper the relationship, the greater the value that Beth confers on the family office. Beth’s major skills are thus relationship specific. Aside from her knowledge of the law, the skills she develops at the family office—including her skills in working with others—generate value.

We can now return to the concepts of contract specificity and managerial direction. Because Andrew’s work produces a relatively standard product, the customers do not need control (that is, managerial direction) over how Andrew does his work. The customer merely supplies some information in advance that Andrew uses to conform the product at the margin to the customer’s needs. In contrast, because Beth’s product is constantly changing in response to changing conditions and the idiosyncratic needs of the firm she works for, someone—the manager or the person who coordinates the joint production of all the workers in the family office—is given managerial direction over Beth’s work.

The distinction between discrete and relational work is an old one. In the Middle Ages, craft workers typically proceeded through two stages—a relational and a discrete stage.^[25] The relational stage was the apprenticeship, during which the worker worked for a single master and became steadily more valuable to the master as he gained experience both at the craft and at serving the master’s interests. Once the apprenticeship was over, the worker could sell his services, now discrete work, to the market by forging swords, cobbling shoes, and building walls—directly for customers. During the Industrial Revolution, some workers worked in teams at factories, while others did piecework at home. The factory workers did relational work—and would be regarded as employees today. The pieceworkers did discrete work, often at home, and resembled contractors. There have been cases in which a single worksite housed workers who worked in teams and workers who worked independently on components; the latter group of workers thought of themselves as craftsmen with higher status than the first group of workers, who resembled today’s employees.^[26] In the pre-internet age of the twentieth century, relational work continued in factories and other workplaces while discrete work was often conducted by skilled professionals—plumbers, doctors, artists, writers, and lawyers. Thus, the notion that some work lends itself to relationships, and other work does not, is not a new one; it is a thread that runs through the history of labor relations.

C. Market Structure

The distinction between relational and discrete work matters for policy because it roughly maps onto a distinction in market structure. Because the seller of discrete work maximizes the value of the work without entering into relationships with buyers, she can sell that work to many different buyers. The “cost of exit”—the cost to the worker if any specific labor buyer stops buying from her—is low because the worker can simply find another labor buyer. (Think of the locksmith who goes from household to household and will not be significantly harmed if one household stops using the locksmith’s services.) The market for discrete work is thus (relatively) competitive. By contrast, the seller of relational work maximizes the value of that work by working for a single labor buyer and remaining with that buyer for an extended period of time. She must make a relationship-specific investment in assets owned by the labor buyer and in other people who work for the labor buyer. This means that the highest-value buyer of the worker’s labor is the person or firm for whom the worker already works and thus that the worker’s cost of exit is high. The market is (relatively) monopsonistic.

In making this argument, I use the term “monopsony” in a specialized way familiar to labor economists but mostly unfamiliar to antitrust lawyers and industrial organization economists. In the antitrust world, monopsony means a single buyer (or sometimes a small number of buyers where technically oligopsony prevails); in the context of labor, this means a single employer.^[27] Thus, a city like New York with dozens of law firms has a competitive labor market for lawyers because the firms compete among each other to hire the best lawyers, while a small town with, say, a single law firm and no other employers of lawyers has a monopsonized market. For labor economists, however, monopsony simply means that the employer can (and thus presumably does) pay the worker below that worker’s marginal revenue product—roughly, the value that the worker generates for the employer—because the worker’s outside options are limited; exit cost is high. The worker’s outside options might be limited because few other employers exist (a traditional monopsony). But the worker’s outside options might be limited for other reasons—for example, it takes a long time to find similar employers who are hiring (called “search costs”) or nominally similar employers are actually quite different in ways that matter for employees, including the location of the worksite (which affects the commute), the workplace culture, and so on (called “job differentiation”).^[28] All of this derives from the relational nature of work.

For our purposes, the distinction is unimportant. By definition, a relational worker who works for an employer is subject to monopsony pressure. Because her services are worth more to the current employer than (in normal cases) to another employer, the current employer can pay her more than any outside employer. But, by the same token, the worker’s outside options are limited. If she earns $100,000 at the current employer, and outside employers will offer only $50,000, then she cannot make a credible threat to leave if the current employer stops giving her raises. The current employer has strong incentives to limit wage increases (or even to reduce wages) or to worsen working conditions (for example, by demanding more work after hours). As labor economists have observed, virtually all employers thus have monopsony power. The usual explanation they give is that search costs and job differentiation give rise to frictions, but the simpler way of putting this is that work for employers is relational. The buyer of discrete work, by contrast, has no such power.

A possible objection to this argument is that people who apply to become relational workers can protect themselves by demanding an employment contract that protects them from such ex post exploitation. When Beth applies for the position with the family office, she is effectively a discrete worker (with a commodified body of knowledge about tax law) who hopes to obtain relationship-specific skills and the higher compensation that comes with them. If she is rational and far-sighted, she knows that as she develops these skills, the family office will be able to use its market power over her to pay her less than her marginal revenue product. To protect herself, she could, in principle, demand that her compensation rise indefinitely as her marginal revenue product rises and insist on a contract to that effect. If the family office breaches, she would be able to sue for damages.

This counterargument brings us back to the issue of contract specificity. The property rights literature on which I draw takes for granted the impossibility of such contracting; that is why residual control is so important. Because the parties cannot specify value-maximizing behavior in advance, they assign residual control to whichever party has the best incentives to use it to generate a surplus.^[29] Empirically, such bargaining is rarely observed, most likely because the employer does not trust a court to determine the marginal revenue product of a particular employee and requires flexibility to adjust wages and working conditions in response to external shocks. It may also be that few employees have the sophistication to demand such contracts—perhaps aside from a handful of highly experienced, highly skilled employees who can hire agents and lawyers to protect them. And if few job applicants are sophisticated and wealthy enough to bargain for wage protection, then employers do better by looking for those who lack sophistication. Thus, the competition for workers that takes place when a soon-to-be-relational worker is hired (say, recent law school graduates) does not translate into market protection for those workers years or decades later in their careers.

As a general proposition, discrete workers are protected by competition for their work; relational workers are not. By “protected,” I mean merely that the workers are paid the efficient wage, their marginal revenue product—the wage necessary to maximize production, rather than a lower wage.

The argument is a generalization; exceptions exist. While relational work unavoidably generates monopsony, markets for discrete work are not always competitive. I will return to this issue in subpart II(B). Moreover, monopsony power can result from other aspects of the employment relationship not captured by relationship-specific investment.

D. Policy Implications

I have argued that work can be divided into discrete and relational types, and that discrete work tends to yield competitive labor markets while relational work tends to yield monopsonized labor markets. Why does this matter? The answer is that the appropriate legal protections are different for workers in monopsonized labor markets and workers in competitive labor markets. Where markets are monopsonized, they fail: they produce inefficiently low output. Legal regulation is called for, and it takes the form of what is conventionally called “employment law” and “labor law.” The label of the first body of law should come as no surprise given my claim that workers who faced monopsonized conditions are properly called “employees.” Where markets are competitive, they do not fail. Legal regulation is not called for—at least not required for the purpose of correcting a market failure.^[30] This is why there is no separate body of “contractor” law that confers protections on contractors. The table below summarizes the argument to this point.

Major Contractor/Employee Distinctions

This is why classification matters. When workers engage in discrete work in competitive labor markets, they should be classified as contractors because competition adequately protects them, while employment and labor law can do little good for them and possibly cause harm to them and society at large. When workers engage in relational work in uncompetitive labor markets, they should be classified as employees because both employment and labor law can help them, and in a socially beneficial way.^[31]

By “employment law,” I mean the range of federal and state protections for workers, including minimum wage laws (which will be my focus), maximum hours laws, laws that regulate workplace safety and other conditions of employment, antidiscrimination laws, and so on. By “labor law,” I mean the National Labor Relations Act and its amendments, which give workers the right to organize unions and strike, along with the labor exemption in antitrust law, which protects workers who engage in such organization from the antitrust laws.

Employment law and labor law reflect different approaches to labor relations, and so I will discuss them separately. Among economists, employment law protections have always been a bit of a puzzle. In a perfectly competitive labor market, employment laws cannot benefit workers and will likely harm them.^[32] Consider the minimum wage. If the labor market is competitive and the market wage exceeds the minimum wage, the minimum wage law does not affect wages. If the market wage is below the minimum wage, then employers will fire workers who generate benefits less than their wage cost. While the wage of other workers may rise, in aggregate workers will be harmed.

This analysis applies to “employment mandates” as well, including family leave policies, for example, and tax advantages for retirement savings. The logic is the same. Suppose, for example, that a family leave policy costs an employer $1,000 per worker. If workers value the family leave policy more than $1,000, say, at $2,000, then the employer will offer it without legal compulsion. The employer can thus save $2,000 in wages by offering a benefit that costs it only $1,000. If workers value the family leave policy at less than $1,000, say, at $500, then legal compulsion can only make the worker worse off—forcing her to accept a benefit of $500 rather than wages of $1,000.

Labor law is also a puzzle from this standpoint, although for a different reason. Labor law authorizes workers to organize a union even though the agreement among workers in a union to strike if they are not paid the same wage, or according to the same compensation schedule, is equivalent to a price-fixing (or actually a wage-fixing) cartel. If the labor market is competitive, then the employer would be required to pay market wages to the workers. If the workers cartelize the labor market through the union, they can insist on an above-market wage, which would reduce both overall employment in the labor market and output, which would probably raise prices for consumers as well. If labor markets are competitive, there would be no reason for the government to encourage or even tolerate unions.

However, as we have seen, labor markets are not competitive. Search costs and related costs ensure that employers enjoy labor monopsony; labor market concentration, which exists in most employment markets, further enhances employers’ market power.^[33] In conditions of labor monopsony, employment law and labor law have stronger justifications.

The best illustration of this claim is the debate about minimum wage laws, which have been subject to extensive empirical research. As noted above, the traditional view of economists assumed competitive labor markets and held that minimum wage laws cannot benefit workers and can only harm them.

Yet after years of controversy, it is becoming increasingly clear that minimum wage laws do not always have this predicted dis-employment effect. They often raise wage levels without reducing employment.^[34] While this result contradicts the assumption of competitive labor markets, it is compatible with labor monopsony. Monopsonistic employers do not fire workers after a compelled wage increase because a monopsonist will make money off workers even when forced to pay above the monopsony wage as long as the minimum wage is not too high. The minimum wage law pushes wages toward the competitive rate that would prevail in a competitive market. And if the minimum wage is closer to the competitive wage than the monopsony wage is, the minimum wage will result in higher employment—since more workers will work for the higher wage.^[35]

Similarly, labor law is a straightforward legal response to the problem of monopsonized labor markets. If the employer enjoys a monopsony over workers, then it can pay a wage lower than the marginal revenue product. The main reason that employers can do this is that workers rarely have a credible threat to quit if they are paid below the market wage—again because of search costs, employer concentration, and related frictions. Workers can increase their market power by agreeing among themselves to quit en masse (that is, strike) if the employer refuses to pay them a higher wage. As a result, the wage should be pushed toward the competitive level (though unions could demand wages higher than the competitive level if they are powerful enough). Labor law could be justified for its role in preventing employers from using aggressive tactics to defeat organization so as to preserve their market power over wages.

We can now see why the law goes to such trouble to classify workers as “contractors” and “employees,” and why employers try so hard to reclassify employees as contractors. Employees benefit from legal protection because they are subject to labor monopsony and hence are not protected by market competition. Employment law protections prevent employers from using their monopsony power to push down wages and worsen conditions. Labor law enables workers to counter employer monopsony power with their own aggregated bargaining power. Contractors do not benefit from employment law protections because market competition already protects them, while the right to organize would enable them to form cartels that charge above-market prices.^[36] The market protects contractors because their discrete skills are valued similarly by numerous labor buyers. And this is why firms that buy work from relational workers have an incentive to misclassify them as contractors; by doing so, they evade employment and labor laws that restrict their ability to exercise monopsony power over their workers.

It is important to see that this justification for employment and labor law—for limiting the law to employees rather than extending it to contractors—is based on market structure, not on income inequality or poverty, which is the focus of the law review literature. From an empirical standpoint, the market structure theory is clearly superior. Employment and labor law protects wealthy employees as well as poor employees, and contractors are deprived of that protection regardless of whether they are rich or poor. From a normative standpoint, the market structure theory is superior as well. Employment law and labor law counter labor monopsony, and in doing so, should generate wealth. While employment and labor law under this understanding should also redistribute wealth from on-average wealthier investors to on-average poorer employers, it does not do so in a targeted way.^[37] These laws also do nothing for the very poor, who are often out of work. Traditional antipoverty programs and tax-and-transfers are a much more suitable way to address the problems of poverty and inequality.^[38]

E. Should Labor and Employment Law Be Mandatory?

The argument so far assumes that workers are locked into employment relationships once they enter in them, but that they are initially hired in a more or less competitive labor market. This raises a question. If workers are rational and informed, they would demand that the employer commit to any mechanism that would diminish or eliminate the employer’s ex post labor market power over them; the employer would agree to the mechanism in order to avoid losing the hire to a competitor. And even if the parties are unable to enter into a complete contract that includes a compensation schedule that adjusts over time, they should be able to agree to the various employment and labor law protections. This suggests that those protections are unnecessary—the parties would choose them if they are efficient in the absence of a legal mandate—or that they should be merely default rules that the parties can opt for if they wish.^[39]

There are two responses to this argument. First, even if this argument is correct for employment mandates like the minimum wage or workplace safety, it is not correct for labor law. A rational and well-informed worker will demand, say, a safe workplace because she values safety over the wage premium the employer would offer if it was spared the expense of providing safety equipment. But a rational and well-informed worker will demand the right to organize only if other workers have the right to organize as well. A one-person union is not a union at all. A rational employer will exploit this collective action problem, and workers can overcome it only by unionizing before they are hired, which is extremely difficult. The mandatory nature of labor law overcomes this problem.

Second, workers are not necessarily rational and well-informed. Indeed, the evidence suggests that they are poorly informed, and that they rarely bargain over the terms of the employment contract.^[40] Employment and labor law can then be justified as terms workers would bargain for if they were informed about the risks of ex post labor monopsony and possible mechanisms to mitigate it. In principle, the law could solve the information problem by requiring that employers inform potential hires of various employment protection options rather than by mandating them. But disclosure rules work poorly,^[41] and so a mandate may be justified.

II. Legal Implications

A. Misclassification

1. The Traditional Test.—

Because employment and labor law restricts firms’ ability to exploit their labor monopsony power, firms have an incentive to classify workers as contractors, regardless of whether the workers are employees in an economic sense, that is, workers who are subject to managerial direction and face a high cost of exit because of their relationship-specific investments in the firm. Uber’s classification of drivers as contractors has recently attracted public attention, but the problem is a longstanding one.

Courts use various tests to resolve disputes over whether a worker is an employee or an independent contractor. The common-law test turned on “control”: the worker is an employee if the labor buyer “controls” her, and an independent contractor otherwise.^[42] While that test was originally developed to determine whether a labor buyer is liable for torts committed by a worker, it continues to be used for classification issues in the employment context.^[43] For federal legislation that protects the rights of workers, the relevant test is the “economic reality” or “economic dependence” test: a worker is an employee if she is “dependent” on the labor buyer. ^[44] Related state legislation uses similar tests.^[45]

The economic dependence test involves six factors:

1. the nature and degree of the alleged employer’s control as to the manner in which the work is to be performed;
2. the alleged employee’s opportunity for profit or loss depending upon his managerial skill;
3. the alleged employee’s investment in equipment or materials required for his task, or his employment of workers;
4. whether the service rendered requires a special skill;
5. the degree of permanency and duration of the working relationship;
6. the extent to which the service rendered is an integral part of the alleged employer’s business.^[46]

Commentators agree that courts apply this test inconsistently.^[47] Part of the problem is that all these factors are vague, and relationships between labor buyers and workers are extremely diverse. It is easy to think of examples of employees who have a great deal of control over their work (traveling salesmen); who are paid bonuses based on managerial skill (managers, for example); who have special skills (industrial scientist); and so on. But the main problem is that the normative goal of the test has been forgotten—or perhaps never fully understood—and so there is no single principle that disciplines how courts apply the various factors.

2. The Relational Work Test.—

The solution is a test grounded in the normative goal of employment and labor law, which, as I have argued, is to counter labor monopsony where workers make relationship-specific investments and face high exit costs. The two major factors—economic dependence and control—are unified: a worker is economically dependent on an employer because of the high costs of exit, and because of the high cost of exit, the worker is willing to subject herself to the firm’s control. Thus, economic dependence does not mean poverty; it means labor monopsony. Economic dependence and control are not in tension; they are different aspects of the employment relationship.

According to the “relational work” test, as I will call it, a worker is a person who performs activities (“work”) for another person in return for pay. A worker is an employee of a firm (or household) when the worker’s cost of finding alternative work of the same type and at the same level of pay is high (“high exit option”). A worker has a high exit option when the work is “relational,” that is, the work is worth more if performed for a single firm over time than if performed in discrete units. In short, employment is relational work; independent contractor status arises from discrete work.

Seen in this way, the factors used in the economic dependence test can be given greater specificity:

1. Control—When work is relational, the labor buyer retains “control” over the worker in the sense of discretionary authority over the worker’s behavior as the worker performs services. Control in this sense is necessary because the labor buyer must constantly coordinate the worker’s work with the work of others. In contrast, discrete work can be largely specified by contract in advance of the work; thus, the labor buyer does not retain control (that is, discretionary authority) over the worker’s behavior as the worker performs tasks.
2. Opportunity for profit or loss depending upon the workers’ managerial skill—When work is relational, the worker allows herself to be managed by the labor buyer who directly, or through subordinates, coordinates the behavior of multiple workers or adjusts work in response to the buyer’s changing needs. The independent contractor uses managerial skill in the course of contracting herself out to multiple labor buyers, and thus is compensated for that managerial skill as well as taking on the risk of managerial failure that results in the loss of clients and hence profits.
3. Investment in equipment or materials required for his task, or employment of workers—This factor reflects the importance of the assignment of property rights so as to locate discretion in the party that is most likely to use it to maximize the surplus. When the work is discrete, the property right is assigned to the worker. When the work is relational, the property right is assigned to the employer.
4. The service rendered requires a special skill—If we interpret “special” to mean relational, this factor fits the relational test. A plumber’s skill is the same whether the plumber is self-employed or employed by a firm; what is special about the skill in the second case is that a plumber adjusts it, through relationship-specific investment, so it satisfies the unique needs of that employer.
5. The degree of permanency and duration of the working relationship—Because the value of relational work is highest for the labor buyer with whom the worker has a relationship and low for other potential labor buyers, the relational worker will tend to remain with the labor buyer in question. Discrete workers, by contrast, may move from labor buyer to labor buyer. Accordingly, when a contract between a worker and a labor buyer stipulates a long-term relationship and prohibits the worker from working for others, the relationship should presumptively be classified as employment regardless of how the parties characterize it.
6. The extent to which the service rendered is an integral part of the alleged employer’s business—Relational work, unlike discrete work, is integral in the sense that it is valuable to the extent that it is used in the operations of the labor buyer with whom the worker has a relationship.

This gloss of the factors in the economic dependence test shows that they can be integrated into the relational-work framework. My claim is not, however, that the courts consistently interpret the factors in this way. On the contrary, in the hands of the courts, the factors have become unmoored from any plausible goal of employment and labor legislation. The courts often seem more focused on whether workers are low-income than on whether they are economically dependent in the market-structure sense.^[48] But, as noted earlier, low-income status is not a reason for classifying workers as employees rather than as independent contractors.

In 2019, the California state legislature passed a law that expanded and simplified its definition of employee. The earlier test resembled the federal economic dependence test. The new law provides that anyone who works for remuneration is an employee unless the labor buyer shows that:

• A) The person is free from the control and direction of the hiring entity in connection with the performance of the work, both under the contract for the performance of the work and in fact.
• B) The person performs work that is outside the usual course of the hiring entity’s business.
• C) The person is customarily engaged in an independently established trade, occupation, or business of the same nature as that involved in the work performed.^[49]

The first provision repeats the control test, and the third provision reflects the idea that discrete work can, and often will, be performed independently of a single labor buyer. But the second provision is broader than the federal rule and seems questionable as well. Under the second provision, a plumbing contractor who establishes contacts with plumbers and contracts them out to households would be deemed an employer. Yet unless the plumbers work in teams on large projects, it does not seem that their plumbing skills are relational to the operations of the plumbing contractor.

3. The Misclassification Debate.—

Gig-economy workers float somewhere between the traditional employee and the traditional contractor. Take the case of Uber. Drivers control some aspects of their work (they choose their automobile, within constraints; they choose when to work) and not others (choice of routes); they do not manage anyone but they do invest in equipment, including the automobile itself; they do not seem to exercise a special skill (driving) but they are free to switch among other labor buyers including other ride-sharing companies and their own clients. Drivers sometimes seem economically dependent in the traditional sense because most of them earn a paltry sum for their work, but they are not economically dependent in the sense of being dependent on just one company, or even on the occupation of driving. Other gig-economy firms locate their workers in other places along the spectrum from contractor to employee. Some workers are allowed to negotiate with customers, set their prices, and exert greater control over working conditions, while others are given less control, and prohibited from using other platforms. Critics argue that firms evade the spirit of the employment laws by giving workers minimal freedoms or responsibilities necessary to qualify them for contractor status rather than allowing those conditions to be determined by the nature of the business.

We can use the controversy over Uber’s classification of drivers as an illustration. Drivers use “effort” to drive expeditiously and maintain their cars. By allocating ownership of the cars to the drivers, Uber gives them high-powered incentives to maintain their cars. But it is easy to imagine the alternative approach in which Uber owns the cars. While the drivers’ care incentives would be diminished, Uber would gain greater control over other aspects of the drivers’ work—for example, the level of courtesy they offer passengers, which may be important for Uber’s brand. If Uber owns the cars, then its threat to punish discourteous drivers is strengthened: it can not only kick the drivers off the platform (where drivers can use other platforms); it can take away their cars as well. The analysis is similar to the analysis of trucking where (in my Uber example) the courtesy problem takes the place of the backhaul problem in trucking.

From a legal perspective, the question is whether Uber classifies its drivers as contractors because the care incentives are more important and more responsive to the assignment of property rights than courtesy incentives. If the answer is yes, then Uber’s classification of drivers reflects the underlying economic reality of ride-sharing rather than an attempt by Uber to evade the law. The drivers do not make relationship-specific investments in Uber’s platform, and so the drivers are protected by competition—they can find work with other platforms or independently find clients—and so do not need the protection of the law.

Unfortunately, it is difficult to determine what Uber (or any other ride-sharing or taxi or limousine company) would do in the absence of a legal regime that will push it to classify workers as contractors in order to avoid the costs of complying with employment and labor laws. A court or regulator can at best do a rough analysis. Applying the relational work test, the most striking aspect of the Uber case is that Uber (like the trucking companies after the introduction of onboard monitoring) imposes a considerable level of control over drivers through contract specificity (thanks to the platform technology). This allows Uber to control the drivers’ level of courtesy with the star rating system, for example. And this allows Uber to assign the property right over the car to the worker so as to enhance the worker’s motivation to take care. On the other hand, the monitoring technology that enables Uber to control the routes, courtesy, and other elements of the service would also enable it to monitor the driver’s care level (as the trucking example shows). And the empirical evidence suggests that drivers cannot easily abandon the Uber platform once they begin using it,^[50] suggesting a nontrivial relationship-specific investment in using and mastering the technology.

While we will not try to resolve the Uber question here, the example illustrates how the relational work test should be applied.

B. Antitrust Law

The gig economy has spawned antitrust suits that allege that firms like Uber have cartelized labor markets. However, while many commentators allege that Uber has exploited drivers, at least one group of plaintiffs argues that Uber coordinated price-fixing by drivers.^[51] Which is it? Are drivers passive victims or complicit wrongdoers?

The problem comes back to the question of whether drivers are employees or contractors. Labor law assumes that employees lack bargaining power—that is, they face an employer monopsonist—and gives them the right to organize so that they can counter the employer’s bargaining power with their own. Antitrust law, for this reason, recognizes a labor exemption: employees do not violate antitrust law by organizing even though a union is a type of price- (or actually wage-) fixing arrangement.^[52] Labor union organization when employees are subject to monopsony should increase production and (in many cases) lower prices, while raising wages as well. However, because contractors are not employees, they do not benefit from the labor exemption, and they are not permitted to organize.^[53] Organization by contractors in a competitive market would result in a cartel, and hence less production and higher prices. That is why drivers who organize violate the law if they are contractors but not if they are employees.

Sanjukta Paul argues that Uber should not be permitted to set the prices for workers if antitrust law prohibits workers from doing it themselves.^[54] She argues that it would be anomalous if Uber were allowed to organize a sellers’ cartel by inviting drivers to its platform, fixing prices, and collecting the rents paid by consumers—when drivers are prohibited by antitrust law from fixing prices.^[55] However, the seeming paradox dissolves upon inspection. Uber is not allowed to organize drivers and set their prices if the result would be a monopoly. That would violate Sections 1 and 2 of the Sherman Act.^[56] Lawsuits against Uber for cartelizing the market have failed because Uber faces competition from taxis, other rideshare companies like Lyft, public transportation, and other services. And while it is true that if some drivers fixed prices then they would violate the antitrust laws, that is only so because of the crude per se ban on price-fixing. The drivers could easily avoid liability by merging into a single firm rather than fixing prices as long as the merger does not encompass an excessive share of the market.

The discussion should make clear how employment law protections fit in with antitrust and labor law to create a general legal structure that governs the relationships between workers and labor buyers. Antitrust law prohibits workers in competitive labor markets from forming cartels, while labor and employment law protects workers in monopsonized labor markets—labor law by allowing them to organize, and employment law (including the minimum wage law) by regulating prices and conditions. On this view, workers are divided into “employees,” who are subject to monopsony, and contractors, who are not.

Thus, it is important in the antitrust context to get straight the reasons why workers should be classified as employees or contractors. When workers operate in a competitive labor market, they should be classified as contractors, and thus forbidden to organize, because the right to organize would result in cartelization and above-market wages. When workers operate in a monopsonized labor market, they should be classified as employees, and thus allowed to organize, because the right to organize should allow them to counter employer market power and thus raise wages toward the competitive level.

We see analogies on the product-market side. When firms have natural monopolies, they are generally immune to antitrust challenge, but they may be subject to price regulation and related regulations designed to prevent them from abusing their market power. Given the parallel nature of product and labor markets, it should not be surprising that there are analogous rules on the labor side even if they have rarely been recognized as such because of the different legal terminology.

It is possible to argue that if employment and labor law counters labor monopsony power, the antitrust laws are not needed, and workers should not be permitted to bring antitrust actions against employers. But the two approaches to monopsony are complements. On the product-market side, antitrust law coexists with various forms of price regulation, like usury laws, anti-price-gouging laws, and insurance premium limits. Antitrust law focuses on tactics employed by firms to increase their market power or extend monopolies into new markets, while price regulation limits the negative effects of firms that have achieved market power lawfully. Similarly, while antitrust law can be used to prevent firms from increasing their power over labor markets through collusion, mergers, and the like, other areas of law are needed to counter the negative effects of labor market power that is achieved lawfully.^[57]

A final point is that while I have assumed that discrete-work markets are competitive, they may not be. Discrete-work markets should be more competitive than relational-work markets because the discrete worker does not make a relationship-specific investment in a primary-labor buyer and hence does not have high exit costs derived from the relationship. But a discrete worker may still face high exit costs if there are few labor buyers. Concentration arises simply because there are few buyers, and concentration could occur either legally (because there are economies of scale, for example) or illegally (because mergers result in a single dominant labor buyer). In this case, the logical source of legal protection for the discrete worker is antitrust law, which distinguishes between concentration that is regarded as socially harmful and concentration that is regarded as tolerable.^[58]

C. Respondeat Superior

Tort law imposes liability on employers for torts committed by employees under certain conditions. The law distinguishes employers and independent contractors using the control test.^[59] If a labor buyer exerts control over a worker, then it will normally be classified as an employer and be held liable. If a labor buyer does not exert control, then it will be classified as a “customer” (with the worker classified as a contractor) and not be held liable. However, there are numerous exceptions to this principle. An employer is not liable for torts committed by employees outside the scope of employment, and a firm may be liable for torts committed by independent contractors, if the tort is partly the result of the labor buyer’s negligence or is the result of inherently dangerous activity.^[60]

Alan Sykes influentially argued that respondeat superior (and vicarious liability for labor buyers more generally) is justified when the worker is judgment-proof and the labor buyer can sufficiently control the worker so as to deter the worker from committing torts—for example, through training, inspections, monitoring, and sanctions.^[61] If the worker is not judgment-proof, the law can impose liability directly on the worker in order to deter torts. Respondeat superior is necessary because a firm and a worker might otherwise agree that the worker’s judgment-proof status will be used to shield the firm from liability in return for the firm compensating the worker. But if the firm has no practical means of controlling the worker’s behavior, then respondeat superior does not improve incentives to take care, but is simply a tax on the enterprise.

One puzzle in Sykes’ analysis is that it does not explain why the law uses the categories of employee and contractor. His analysis suggests instead that the law should always impose liability on person A if person B performs a task at A’s request and for A’s benefit, person B is judgment proof, and person A can control person B. The categories of employee and contractor do not work in this analysis. Indeed, when courts evaluate respondeat superior claims, they do not rely on the firm’s and worker’s own classification of the worker. Courts simply look for evidence of control, and then classify the worker based on how much control the firm exerts over her. Thus, the association of respondeat superior with the employment relation seems all the more puzzling.^[62]

The solution to this puzzle brings us back to relationship specificity. A worker who makes relationship-specific investments in a firm (its assets and other workers) is subject to the firm’s control because, as a result of these investments, the worker’s cost of exit is high. Because the cost of exit is high, the worker will be concerned that if she commits torts for which the employer is liable, she will be denied bonuses, raises, and promotions—or that she will be fired. By contrast, a discrete worker does not fear such sanctions—because she can work for anyone else. Sykes implicitly acknowledges this distinction, noting that “agents often earn returns in excess of what they can earn in their next-best employment opportunity, and expect those returns to continue into the future. Such agents have an important stake in retaining their current positions.”^[63] Labor monopsony, which ensures that the exit cost is high, thus is important for giving labor buyers leverage for disciplining workers who commit torts. This distinction explains why respondeat superior is associated with employment relationships rather than the broader class of buyer–worker relationships encompassing discrete workers.

D. The Employee/Contractor Distinction in Other Areas of the Law

Congress and state legislatures have built an enormous administrative structure on the distinction between employee and contractor. Employees are entitled to unemployment insurance; contractors are not. Employees benefit from tax-subsidized retirement and health insurance plans operated by employers; contractors do not. And taxes on the wages of employees are withheld by employers; taxes on payments made to contractors are not withheld by the buyers of their services. A question arises why these programs and practices should apply to employees only, and not to workers more broadly.

The existing system may not be optimal but it follows a certain logic. Relationship specificity creates monopsony, which is another way of saying that workers develop thick relationships with labor buyers and cannot easily exit those relationships. Employees are therefore more vulnerable if laid off than contractors who lose business. A locksmith who loses one customer can find another; but a factory worker whose value to a firm is relationship-specific will have trouble finding a comparable job at a similar wage if laid off. Thus, unemployment insurance may be more appropriate for employees than for contractors, at least in normal times, as opposed to a recession that dries up demand for the contractor’s services.

For health insurance, the stickiness of the relationship between employee and employer may help address adverse selection problems in the insurance market and justify group-based insurance. By contrast, contractors who try to create an insurance pool take the risk that the healthiest will opt out. And as for retirement subsidies, there does not seem to be a good reason for offering them to employees but not contractors. But the law does offer various self-employed retirement subsidies that contractors can take advantage of. The law seems to draw a distinction between employee-based retirement subsidies and contractor retirement subsidies so as to enlist employers to manage retirement plans—taking advantage of the administrative capacity of large labor buyers who have continuous relationships with workers.

Finally, tax withholding is a form of monitoring and enforcement, and these tasks should be assigned to the party (labor buyer or worker) who is most likely to perform them efficiently and conscientiously. As a rough approximation, labor buyers who employ people will have the infrastructure in place for withholding, while households, and even firms that rely on contractors, will not.

Conclusion

Let us sum up the analysis. A worker is an employee when she makes a relationship-specific investment in, and applies her labor to, assets owned by, or other workers employed by, another party. A worker is a contractor when the value of her output is independent of such relationships. An employee’s output is thus relationship specific or “relational”: the output is valued more by the labor buyer than by outsiders. A contractor’s output is “commodified”: its value does not depend on the identity of the labor buyer. Because of relationship specificity, the employer is a monopsonist with respect to the worker’s output. The buyer from a contractor is not (normally) a monopsonist. If it is, it is not because of the relationship with the worker but because of market concentration.

Because an employer is a monopsonist, it can (and will) pay the worker less than marginal revenue product (at least after the initial hire), resulting in inefficiency (lower output) and often inequity. Thus, there is a labor market failure. When the worker is a contractor, there is no such labor market failure. The labor buyer is not a monopsonist (normally). Exit is costly for the employee, cheap for the contractor.

In light of these distinctions, employment and labor law should be understood as legal mechanisms for addressing the labor market failure caused by labor monopsony. Employment law is a form of price (wage) and quality (conditions) regulation, understood to raise wages/conditions toward the competitive rate. Labor law is a form of power aggregation, understood to convert monopsony into bilateral monopoly, with better outcomes closer to the competitive rate. This is not to say that all employment law protections work as intended; unions can also produce perverse consequences. But these areas of law are best understood as the legal response to the problem of labor monopsony. Antitrust law provides a residual set of protections that are invoked when anticompetitive efforts to cartelize or monopsonize markets result in higher market power.

Firms misclassify workers as contractors in order to avoid employment and labor law protections so that they can exploit their monopsony power and suppress wages. To prevent misclassification, courts and regulators must understand the true economic relationship between a labor buyer and a worker—essentially whether the labor buyer exerts monopsony power because of relationship-specificity. The factors used in the various misclassification tests should be interpreted in this light.

This account should be understood as an attempt to rationalize labor and employment law, and the law of classification—in the sense of offering the best justification for these laws. I do not try to prove that this body of law does what it is supposed to do. There is some evidence that it does, but the evidence is mixed, and varies for different types of rights and protection.

This analysis leaves some questions. One could imagine an alternative legal regime in which no formal distinction is made between employees and contractors and instead employment protections are extended to all workers on the basis of market structure. For example, a minimum wage law could be designed to apply to all labor markets in which the level of monopsony (typically measured as residual labor supply elasticity, which measures the extent to which a worker will tolerate a below-market wage before quitting), whatever its source (including concentration), exceeds a threshold.^[64] What would be wrong with such a system? One possible answer is that measurement is too complex, and courts and regulators require a simpler, even if blunter, test. It is also possible that our current system reflects historical contingency. Policymakers have found the contractor/employee distinction useful in a range of contexts, where the consequences of the distinction are different. That may explain both why the distinction is so entrenched and why the modern test has gotten so muddled.

1. .Some commentators claimed that these alternative work arrangements (outside the employment relationship) have transformed labor markets throughout the United States, a claim that received a boost from a 2016 study, which found that alternative work arrangements increased from 10.7% in 2005 to 15.8% in 2015. See Lawrence F. Katz & Alan B. Krueger, The Rise and Nature of Alternative Work Arrangements in the United States, 1995–2015, at 1 (Nat’l Bureau of Econ. Rsch., Working Paper No. 22667, 2016) (discussing the findings of a survey on alternative work arrangements). However, the authors later revised their estimate down to a 1% increase. See Lawrence F. Katz & Alan B. Krueger, Understanding Trends in Alternative Work Arrangements in the United States 2 (Nat’l Bureau of Econ. Rsch., Working Paper No. 25425, 2019) (explaining analysis that “indicates a modest upward trend in the incidence of alternative work arrangements of about 1 percent of the workforce”). ↑
2. .See, e.g., Keith Cunningham-Parmeter, From Amazon to Uber: Defining Employment in the Modern Economy, 96 B.U. L. Rev. 1673, 1677 (2016) (“[C]urrent judicial pronouncements . . . embrace a cabined vision of employment that shields firms from liability.”); Brishen Rogers, Employment Rights in the Platform Economy: Getting Back to Basics, 10 Harv. L. & Pol’y Rev. 479, 500, 505–07 (2016) (arguing that underinclusive employment classification tests leave workers at risk of excessive employer domination in violation of fundamental democratic and egalitarian values, suggesting that companies like Uber should be required to treat drivers as employees); Jennifer Pinsof, Note, A New Take on an Old Problem: Employee Misclassification in the Modern Gig-Economy, 22 Mich. Telecomm. & Tech. L. Rev. 341, 344 (2016) (suggesting that Uber drivers and similar workers should be classified as employees, which would provide necessary worker protections and benefits); Seth D. Harris & Alan B. Krueger, A Proposal for Modernizing Labor Laws for Twenty-First-Century Work: The “Independent Worker,” Brookings Inst. (2015) (proposing protections for non-employees). ↑
3. .See John A. Pearce II & Jonathan P. Silva, The Future of Independent Contractors and Their Status as Non-Employees: Moving On from a Common Law Standard, 14 Hastings Bus. L.J. 1, 3 (2018) (“The claim that a business’s economic motivations led to classifying workers as independent contractors has been the basis for numerous legal challenges by gig-economy workers claiming that they were improperly classified as independent contractors and that the nature of their work makes them employees.”); Pinsof, supra note 2, at 349–54 (describing “rampant” misclassification, particularly in the gig economy); Alexia Fernández Campbell, Companies Often Mislabel Employees as “Freelancers” to Cut Costs. Workers Are Fighting Back., Vox (Mar. 20, 2019, 4:30 PM), https://‌www.vox.com‌/policy-and-politics‌/2019‌/3‌/20‌/18272918‌/conde-nast-epicurious-employee-freelancer-contractor [https://‌perma.cc‌/EA7N-YQ56] (suggesting that financial incentives to misclassify are powerful and pervasive, potentially affecting millions of workers); Harris & Krueger, supra note 2, at 7 (noting that legal uncertainty has created both “intentional and unintentional” misclassification). Allegations of misclassification have led to a flurry of class action lawsuits. See Liya Palagashvili, Disrupting the Employee and Contractor Laws, 2017 U. Chi. Legal F. 379, 382–83, 405–08 (describing such cases). ↑
4. .The literature on the question of worker classification has grown massive. For research from the last few years, see, e.g., Rogers, supra note 2, at 482–83 (suggesting that courts should not shy away from reliance on normative values to determine employment status as imposing duties on employers when doing so advances the underlying policy goals of employment law such as preventing employee domination); Naomi B. Sunshine, Employees as Price-Takers, 22 Lewis & Clark L. Rev. 105, 110 (2018) (suggesting a rebuttable presumption that workers lacking power to set their rates should be classified as employees); Cunningham-Parmeter, supra note 2, at 1677–78 (noting that judicial focus on daily, direct control over working conditions is underinclusive, and instead arguing that courts should use a broader definition of control that is less formalist and considers power imbalances); Brishen Rogers, Redefining Employment for the Modern Economy, Am. Const. Soc’y for L. & Pol’y, Oct. 2016, at 7 (available at https://‌www.acslaw.org‌/wp-content‌/uploads‌/2018‌/04‌/Redefining‌_Employment‌_for‌_the‌_Modern‌_Economy.pdf [https://‌perma
   .cc‌/YGC8-7R6C]) (proposing expanding the definition of employee to include all workers who are “economically dependent” on the employer); Richard R. Carlson, Employment by Design: Employees, Independent Contractors and the Theory of the Firm, 71 Ark. L. Rev. 127, 130 (2018) (proposing a test based on Ronald Coase’s “make or buy” theory of the firm that would incorporate analysis of the firm’s motives for hiring rather than buying labor); Matthew T. Bodie, Participation as a Theory of Employment, 89 Notre Dame L. Rev. 661, 665–66 (2013) (proposing a definition of employee based on the motive of the employer); Benjamin Means & Joseph A. Seiner, Navigating the Uber Economy, 49 U.C. Davis L. Rev. 1511, 1539 (2016) (suggesting a flexible, case-by-case approach that avoids “sweeping all workers . . . into one category or the other” due to the varied circumstances of workers even within a single company). Another group of commentators has suggested the creation of a third, middle category between employee and contractor, to be granted some, but not all, of the protections afforded to employees. See, e.g., Harris & Krueger, supra note 2, at 5 (suggesting creation of an “independent worker” category); Pearce & Silva, supra note 3, at 2 (suggesting creation of a “dependent contractor” to capture workers that are functionally independent contractors but are nevertheless economically vulnerable); Michael L. Nadler, Independent Employees: A New Category of Workers for the Gig Economy, 19 N.C.J.L. & Tech. 443, 480–81 (2018) (suggesting an “independent employee” intermediate category of worker); Miriam A. Cherry & Antonio Aloisi, Dependent Contractors in the Gig Economy: A Comparative Approach, 66 Am. U. L. Rev. 635, 637 (2017) (observing that an intermediate, “hybrid” category is found in different legal systems around the world). ↑
5. .Preetika Rana, Uber Tests Feature Allowing Some California Drivers to Set Fares, Wall St. J. (Jan. 21, 2020, 7:29 PM), https://‌www.wsj.com‌/articles‌/uber-is-testing-a-feature-that-lets-some-california-drivers-set-fares-11579600801 [https://‌perma.cc‌/AFH4-YHU5]. ↑
6. .Aarian Marshall, Why Uber Thinks It Can Still Call Its Drivers Contractors, Wired (Sept. 12, 2019, 7:00 AM), https://‌www.wired.com‌/story‌/why-uber-still-call-drivers-contractors‌/ [https://‌perma.cc‌/Z6QV-FLTA]. ↑
7. .David Ingram & Diana Dasrath, More Drivers Sue Uber, Saying They’re Employees, Not Contractors, NBC News (Dec. 18, 2019, 12:08 PM), https://‌www.nbcnews.com‌/tech‌/tech-news‌/
   more-drivers-sue-uber-saying-they-re-employees-not-contractors-n1103936 [https://‌perma.cc‌/
   JN5V-LTEE]. ↑
8. .See supra note 4 and accompanying text. ↑
9. .See, e.g., Pearce & Silva, supra note 3, at 2–3 (suggesting a third employment classification that would provide so-called “dependent contractors” with certain employment rights); Nadler, supra note 4, at 480–81 (suggesting “independent employees” as a new classification since many gig-economy workers have multiple sources of income and are not necessarily dependent on their employers); Harris & Krueger, supra note 3, at 5 (suggesting “independent workers” as a third classification for those falling somewhere between employees and independent contractors). See also supra notes 2–4 and accompanying text. ↑
10. .Two earlier law review papers draw on the theory of the firm, as I do (see infra), but offer different tests and do not derive the distinction between contractor and employee from the problem of labor market monopsony. See Bodie, supra note 4, at 665–66 (arguing that an employee is a worker who “participates” in a firm); Carlson, supra note 4, at 127, 130 (proposing a test that draws on employers’ economic motives for hiring rather than buying labor). ↑
11. .As I will discuss later, it is not clear that these laws, or all of them, are actually effective; but that is an empirical question that I put to one side. ↑
12. .See, e.g., Harris & Krueger, supra note 2, at 2 (explaining that “independent workers” should qualify for certain benefits since they generally do not qualify for hours-based unemployment insurance and other benefits that employees receive). ↑
13. .See Marina Lao, Workers in the “Gig” Economy: The Case for Extending the Antitrust Labor Exemption, 51 U.C. Davis L. Rev. 1543, 1543, 1547 (2018) (arguing that the gig economy’s hybrid work relationships and the underlying purpose of the labor exemption suggests that extension to non-employees would be appropriate). ↑
14. .Ronald H. Coase, The Nature of the Firm, 4 Economica 386, 403–04 (1937). ↑
15. .See Scott E. Masten, Legal Basis for the Firm, 4 J.L. Econ. & Org. 181, 186 (1988) (discussing the historic emphasis courts have placed on the ability of the employer to direct the means by which an employee performs their work). ↑
16. .Armen A. Alchian & Harold Demsetz, Production, Information Costs, and Economic Organization, 62 Amer. Econ. Rev. 777, 778 (1972). ↑
17. .The starting point for this literature is a series of articles by Sanford Grossman, Oliver Hart, and John Moore (GHM), who pointed out that because contracts are necessarily incomplete (in the sense of being unable to specify all the optimal actions for both parties), it is important for the parties to specify which party will have discretion within the scope of incompleteness. The parties can assign discretion by allocating property rights: the party with a relevant property right has discretion over the use of that property—“residual” discretion or control in the sense that the discretion is limited by any specified contractual terms. The parties should then assign that discretion (via the assignment of property rights) to whichever party is more likely to use discretion in the joint interest of both parties—and that is roughly the party whose incentive to invest in the joint project is more sensitive to the return on that investment. The difference between an employee and a contractor is that the employee does not own the asset in which she applies her labor—that asset is owned by the employer—while a contractor does own the asset. The contractor has greater bargaining power vis-à-vis the firm and the stronger incentive to maintain the asset but a weaker incentive to use the asset to benefit the firm. See, e.g., Sanford J. Grossman & Oliver D. Hart, The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration, 94 J. Pol. Econ. 691 (1986) (theorizing that contractual rights can be categorized as either specific rights or residual rights, and that one party may purchase all residual rights when listing all specific rights over assets in the contract would be costly); Oliver Hart & John Moore, Property Rights and the Nature of the Firm, 98 J. Pol. Econ. 1119, 1119 (1990). In demonstrating the value of these rights, Hart and Moore argue:[T]he crucial difference for party 1 between owning a firm (integration) and contracting for a service from another party 2 who owns this firm (nonintegration) is that, under integration, party 1 can selectively fire the workers of the firm (including party 2), whereas under nonintegration he can “fire” (i.e., stop dealing with) only the entire firm[.]

    Id. Philippe Aghion and Richard Holden provide a lucid discussion. See Philippe Aghion & Richard Holden, Incomplete Contracts and the Theory of the Firm: What Have We Learned over the Past 25 Years?, 25 J. Econ. Persps. 181 (2011) (discussing incomplete contracts and vertical integration and showing how the incomplete contracts approach can be extended beyond the firms’ boundaries issue). Other authors further developed the special role of the employee. See, e.g., Bengt Holmström & John Roberts, The Boundaries of the Firm Revisited, 12 J. Econ. Persps. 73, 79 (1998) (noting that employees typically own no assets, yet they often work effectively); George P. Baker & Thomas N. Hubbard, Make Versus Buy in Trucking: Asset Ownership, Job Design, and Information, 93 Am. Econ. Rev. 551, 551 (2003) (developing a model of asset ownership in trucking); Eric Van den Steen, On the Origin of Shared Beliefs (and Corporate Culture), 41 RAND J. Econ. 617, 617 (2010) (discussing how corporate culture originates and persists, and why culture and employee performance may be inversely related); Wouter Dessein, Incomplete Contracts and Firm Boundaries, 30 J.L. Econ. & Org. 13, 13 (2014) (proposing two approaches that introduce managerial direction and authoritative decision-making in a formal theory of the firm). For an early piece that anticipated some of these arguments, see Alchian & Demsetz, supra note 16, at 777–78, which argues that firms need control over employees because of the difficulty of controlling team production by contract. ↑

18. .George P. Baker & Thomas N. Hubbard, Contractibility and Asset Ownership: On-Board Computers and Governance in U.S. Trucking, 119 Q.J. Econ. 1443, 1444 (2004). ↑
19. .Id. at 1446. ↑
20. .Id. at 1447. ↑
21. .Id. at 1446–48. ↑
22. .Id. at 1476. ↑
23. .The distinction between discrete and relational work is similar to the distinction between routine and nonroutine work. Nonroutine work, like relational work, is more vulnerable to labor monopsony than routine (and discrete) work. For discussion and evidence, see Ronald Bachmann, Gökay Demir & Hanna Frings, Labour Market Polarisation and Monopsony Power, J. Hum. Res. (forthcoming 2021). ↑
24. .See Gary S. Becker, Human Capital 33–34 (3d. ed. 1993) (discussing the economic incentive to provide general job training to employees). ↑
25. .See Carlson, supra note 4, at 146–47 (“[A] master craftsman’s servants were often minors and student workers, residing with the craftsman and dependent on the craftsman for instruction, direction, access to assets, lodging, and daily needs.”). ↑
26. .For a detailed illustration based on the Winchester Company, see John Buttrick, The Inside Contract System, 12 J. Econ. Hist. 205 (1952), and the very helpful discussion in Carlson, supra note 4, at 151–58. Buttrick observes that the “employees,” more than the contractors, were subject to monopsony (because the employees ended up working exclusively for Winchester while the contractors sold their services to other companies as well), Buttrick, supra note 26, at 210, as does Carlson, supra note 4, at 154—my point here. But neither of them recognizes its centrality to the distinction between employee and contractor. ↑
27. .See Suresh Naidu, Eric A. Posner & Glen Weyl, Antitrust Remedies for Labor Market Power, 132 Harv. L. Rev. 536, 538–39 (2018) (describing the creation of monopsony conditions where employers are the buyers operating in a labor market). ↑
28. .Alan Manning, Monopsony in Motion: Imperfect Competition in Labor Markets 199–200, 241 (2003). ↑
29. .See Hart & Moore, supra note 17, at 1121 (explaining residual cost theory in relation to how owners exclude others from their assets). ↑
30. .If contractors require legal protection or other policy responses, the reasons lie elsewhere. For example, one might favor the extension of antidiscrimination norms in employment law to contractors because one believes that discrimination against contractors is morally reprehensible, or that such an extension will advance public values. These arguments lie outside of the narrow economic perspective that I take in this Essay. ↑
31. .The relationship between worker and labor buyer may produce other sources of monopsony power, for example, from the buyer’s private information about the worker’s ability. Daron Acemoglu & Jörn-Steffen Pischke, Why Do Firms Train? Theory and Evidence, 113 Q.J. Econ. 79, 81–82 (1998). In such cases, the relational work test will be underinclusive. It is possible that it should be expanded to address this problem. ↑
32. .For a discussion, see Suresh Naidu & Eric A. Posner, Labor Monopsony and the Limits of the Law 8, 12, 16, 26 (Jan. 31, 2019) (unpublished manuscript), http://‌papers.ssrn.com‌/sol3‌/
    papers.cfm?abstract‌_id=3365374 [https://‌perma.cc‌/F7C3-LB97]. In the legal literature, Richard Epstein has used this insight to launch a wholesale assault on labor and employment law. See, e.g., Richard A. Epstein, Forbidden Grounds: The Case Against Employment Discrimination Laws 9 (1992) (arguing that anti-discrimination laws should be repealed); Richard A. Epstein, Contractual Solutions for Employment Law Problems, 38 Harv. J.L. & Pub. Pol’y 789, 790 (2015) (positing that anti-discrimination is partially responsible for high unemployment rates); Richard A. Epstein, Labor Unions: Saviors or Scourges?, 41 Cap. U. L. Rev. 1, 30 (2013) (arguing the same for labor union laws). Epstein’s argument is, of course, based on the assumption that labor markets are competitive, or nearly so. ↑
33. .For a recent overview of the empirical literature on labor market concentration, see Naidu, supra note 27, at 560–69. ↑
34. .See Arindrajit Dube, T. William Lester & Michael Reich, Minimum Wage Shocks, Employment Flows, and Labor Market Frictions, 34 J. Lab. Econ. 663, 664 (2016) (explaining empirical findings that minimum wage increases do not always decrease employment levels); José Azar, Emiliano Huet-Vaughn, Ioana Marinescu, Bledi Taska & Till von Wachter, Minimum Wage Employment Effects and Labor Market Concentration 3–4 (Nat’l Bureau of Econ. Research, Working Paper No. 26101, 2019) (same). ↑
35. .However, not all elements of employment law can be justified from a labor monopsony standpoint. When workers are paid above the minimum wage because of market competition, employment mandates and related protections do not benefit them and may harm them. See Naidu & Posner, supra note 32, at 17 (arguing that minimum wage laws help only a small class of relatively poor workers who would otherwise be paid slightly less than the minimum wage, and not workers higher on the wage scale). ↑
36. .This holds true unless labor buyers violate the antitrust laws. See infra subpart II(B). ↑
37. .Indeed, even minimum wage laws may not help the poor. The reason is that they are very crude rules that can end up raising wages above the competitive rate even when labor monopsony exists; when they do, the higher costs may be passed on to low-income buyers. For empirical analysis, see Thomas MaCurdy, How Effective Is the Minimum Wage at Supporting the Poor?, 123 J. Pol. Econ. 497, 503–05 (2015) and Doruk Cengiz, Arindrajit Dube, Attila Lindner & Ben Zipperer, The Effect of Minimum Wages on Low-Wage Jobs, 134 Q.J. Econ. 1405, 1408 (2019). ↑
38. .There remains a significant question how much employment and labor law helps workers even when labor markets are monopsonized. I have abstracted from this question for the purpose of this Essay, but a few comments are in order. First, standard employment law protections—including the minimum wage—can probably help only low-skill workers who would otherwise be paid below the minimum wage. The minimum wage does not help higher-skill workers, and it is unlikely that other protections, for example, mandatory family medical leave, can benefit them since employers can reduce wages to offset the cost of mandates. See Lawrence H. Summers, Some Simple Economics of Mandated Benefits, 79 Amer. Econ. Rev. 177, 180 (1989) (explaining that government-mandated leave policy will have downward effect on wages of skilled workers). Second, even the minimum wage may not benefit low-income workers directly if employers offset the cost by raising prices in product markets. The extent of this effect remains empirically ambiguous, and probably varies by location. See MaCurdy, supra note 37, at 505 (asserting that higher minimum wages often translate into higher prices for goods and at least some of the burden of the increased wage is passed onto consumers through higher prices); Peter Harasztosi & Attila Lindner, Who Pays for the Minimum Wage?, 109 Amer. Econ. Rev. 2693, 2708 (2019) (explaining that scholars still disagree about the precise correlation between minimum wage increases and price increases across the board, but where prices increase, they occur most often in the manufacturing sector). Third, labor laws that encourage union organization probably do benefit workers; union organization is probably the most effective remedy for labor monopsony. Several recent empirical papers have found as much. See Efraim Benmelech, Nittai Bergman & Hyunseob Kim, Strong Employers and Weak Employees: How Does Employer Concentration Affect Wages? 4, 15 (Nat’l Bureau of Econ. Rsch., Working Paper No. 24307, 2018) (arguing that unionization strengthens labor’s bargaining position and enables employees to diminish employers’ monopsony power); Elena Prager & Matthew Schmitt, Employer Consolidation and Wages: Evidence from Hospitals, 111 Amer. Econ. Rev. 397, 399–400, 419–20 (2021) (finding that high levels of unionization and a pro-union environment appear to mitigate the estimated negative wage effects). ↑
39. .See Stewart J. Schwab, Law-and-Economics Approaches to Labour and Employment Law, 33 Int’l J. Comp. L. & Lab. Rel. 115, 140–42 (2017) (arguing that even in the absence of a government mandate, the monopsonist employer has a profit-maximizing incentive to provide an efficient wage‌/benefit combination to workers). ↑
40. .For a discussion with citations to the literature, see Ryan Bubb, Patrick Corrigan & Patrick L. Warren, A Behavioral Contract Theory Perspective on Retirement Savings, 47 Conn. L. Rev. 1317, 1324–25 (2015). ↑
41. .See generally Omri Ben-Shahar & Carl E. Schneider, More Than You Wanted to Know: The Failure of Mandated Disclosure (2014) (demonstrating why disclosure mandates repeatedly fail). ↑
42. .For a statement of the common law test, see Cmty. for Creative Non-Violence v. Reid, 490 U.S. 730, 751–52 (1989). ↑
43. .Carlson, supra note 4, at 159–60. I discuss respondeat superior below. ↑
44. .See, e.g., Bartels v. Birmingham, 332 U.S. 126, 130 (1947) (noting that application of social legislation such as the Social Security Act does not hinge on employer “control,” but rather the extent to which the employees “as a matter of economic reality are dependent upon the business to which they render service”). ↑
45. .See Carlson, supra note 4, at 130 (providing that, over time, state courts have added “economic reality” factors to account for the modern workplace). ↑
46. .The factors are quoted from Scantland v. Jeffry Knight, Inc., 721 F.3d 1308, 1312 (11th Cir. 2013). ↑
47. .The Supreme Court recognized the difficulty of applying multi-factor employment tests over 70 years ago. See NLRB v. Hearst Publ’n, Inc., 322 U.S. 111, 121 (1944) (“Few problems in the law have given greater variety of application and conflict in results than the cases arising in the borderland between what is clearly an employer–employee relationship and what is clearly one of independent entrepreneurial dealing.”). See also Carlson, supra note 4, at 171–74 (describing difficulties with multi-factor tests for employee status). ↑
48. .In re Vega, 149 N.E.3d 401, 416 (2020) (emphasizing the fact that Postmates benefits from the labor of unskilled workers and persons of low income). ↑
49. .Cal. Assem. Bill 5, 2019–2020 Reg. Sess. (Cal. 2019), http://‌leginfo.legislature.ca.gov‌/
    faces‌/billNavClient.xhtml?bill‌_id=201920200AB5 [https://‌perma.cc‌/29H2-XF9T]. ↑
50. .See Joshua D. Angrist, Sydnee Caldwell & Jonathan V. Hall, Uber vs. Taxi: A Driver’s Eye View 1–2 (Nat’l Bureau of Econ. Rsch., Working Paper No. 23891, 2017) (finding that many Uber drivers who were offered a lease failed to accept, even when doing so would have been to their advantage). ↑
51. .See Meyer v. Kalanick, 174 F. Supp. 3d 817, 820 (S.D.N.Y. 2016) (“Plaintiff claimed . . . that Mr. Kalanick . . . had conspired with Uber drivers to use Uber’s pricing algorithm to set the prices charged to Uber riders, thereby restricting price competition among drivers . . . .”). ↑
52. .See United States v. Hutcheson, 312 U.S. 219, 232, 235–37 (1941) (holding that the Clayton Act and Norris-LaGuardia Act together provide unions a statutory exemption from antitrust liability). ↑
53. .See Columbia River Packers Ass’n v. Hinton, 315 U.S. 143, 144–45 (1942) (excluding “independent entrepreneurs” from the labor exception); L.A. Meat and Provision Drivers Union v. United States, 371 U.S. 94, 96, 99–102 (1962) (same). ↑
54. .Sanjukta M. Paul, Uber as For-Profit Hiring Hall: A Price-Fixing Paradox and Its Implications, 38 Berkeley J. Emp. & Lab. L. 233, 236–39 (2017). ↑
55. .Id. ↑
56. .15 U.S.C. §§ 1, 2. ↑
57. .Naidu & Posner, supra note 32, at 2, 8. ↑
58. .Naidu, supra note 27, at 587. ↑
59. .Carlson, supra note 4, at 127, 130. ↑
60. .Restatement (Third) of Agency § 7.05, § 7.07 (2006); Alan O. Sykes, The Economics of Vicarious Liability, 93 Yale L.J. 1231, 1259–60 (1984). ↑
61. .Sykes, supra note 60, at 1232, 1235–39, 1244. ↑
62. .As noted by Keith Cunningham-Parmeter, Gig-Dependence: Finding the Real Independent Contractors of Platform Work, 39 N. Ill. U. L. Rev. 379, 406–08 (2019). ↑
63. .Sykes, supra note 60, at 1254 (citations omitted). ↑
64. .Recent papers have measured the residual labor supply elasticity in various settings, e.g., David W. Berger, Kyle F. Herkenhoff & Simon Mongey, Labor Market Power (Nat’l Bureau of Econ. Rsch., Working Paper No. 25719, 2019), and related variables, like the workers’ outside option or what I have called exit options. See Sydnee Caldwell & Oren Danieli, Outside Options in the Labor Market (Nov. 7, 2018) (unpublished manuscript), http://‌scholar.harvard.edu‌/
    files‌/danieli‌/files‌/danieli‌_jmp.pdf [https://‌perma.cc‌/5MC9-P4JG] (developing a model to estimate outside employment opportunities available to workers, based on an “outside options index”). ↑