"Through the interplay of these often conflicting policies the law has become what it is today: a confusing, seemingly incomprehensible puzzle whose underlying pattern is difficult, if not impossible to discern." (Valiulis, p. 156)
Modern common law courts are often reluctant to enforce contracts in restraint of trade. One partial exception is an agreement by an employee restricting his post-employment freedom to compete with his employer, or by the seller of a business agreeing not to go into competition with the buyer. Such non-competition agreements pose two interrelated puzzles. The first is why they exist-why employer and employee, or seller and buyer, find it in their interest to sign such agreements. The second is why courts are willing to enforce certain non-competition agreements and not others. To put the same puzzles in different words, we would like to know what the parties to the agreement think they are doing, what the courts think the parties are doing, and what the courts think they are doing.
Three different sorts of explanations for such agreements seem to make some sense in terms of both the economics of the situations where such agreements arise and the language of the courts; they are explained and discussed in Parts III-V of this article. In each case, I will start with the question of whether the explanation makes sense from the standpoint of the parties-whether it explains their willingness to sign such agreements. I will next consider whether, assuming that explanation is correct, enforcing such contracts is likely to increase or decrease economic efficiency. Finally, I will look at the words and decisions of the courts in order to see both whether they fit the explanation-whether the courts act as if that is what they think is going on-and whether they promote economic efficiency.
Before going through the alternative explanations, it is worth first looking briefly at the legal and economic context of the behavior being explained. I will therefore start, in Part I, with a brief account of the history and current status of the relevant law. In Part II I will sketch the economic reasons for believing that the terms of contracts take account of the interests of both parties, despite some of the legal arguments to the contrary.
The earliest known common law case involving a contract in restraint of trade was decided in 1414; the defendant claimed to have fulfilled his agreement not to practice his trade for a period of six months in the plaintiff's town. Judge Hull observed that the defendant might have demurred because the condition was illegal at common law.
The first known case in which such a contract was enforced is Rogers v. Parry (1611).  The Court distinguished between general restraints, which would always be invalid, and restraints limited in time and place, which might be valid if for adequate consideration. That position was repeated and amplified by Chief Judge Parker in Mitchell v. Reynolds.
Over the next two centuries, both the prohibition of general restraints and the requirements of adequate consideration were gradually relaxed. The former was finally abandoned in Nordenfelt v. Maxim-Nordenfelt Guns and Ammunition Co., Lim. (1894), where it was held that the real distinction was not between general and limited restraints but between unreasonable and reasonable restraints. Since the plaintiff did business over most of the world, a restraint covering the entire world was reasonable. The requirement of adequate consideration was abandoned in Hitchcock v. Coker (1837), where it was held that while some consideration was necessary, it was for the parties rather than the court to decide how much.
Thus, by the end of the nineteenth century, the courts had moved most of the way to complete freedom of contract. While cases occasionally implied that the agreement must be reasonable with regard to the public interest as well as the interest of the parties, that requirement seems to have had little weight in actual decisions. In practice, an agreement was likely to be enforced if it had been made for consideration and was not broader than necessary to protect the legitimate interest of the promisee.
The main change in the law since the end of the nineteenth century has been a gradual retreat from freedom of contract. The language of the courts suggests two different grounds for this change. One is the unwillingness of the courts to assume that the existence of a contract implies that its terms are in the interest of both parties in situations where one party appears to be in a significantly stronger position. The second, and perhaps less important, is a renewed concern for the possibility that non-competition agreements may be against the interest of the general public.
"In this context (post-employment covenants) a restrictive covenant will only be subject to specific enforcement to the extent that it is reasonable in time and area, necessary to protect the employer's legitimate interests, not harmful to the general public and not unreasonably burdensome to the employee. ... Indeed, our economy is premised on the competition engendered by the uninhibited flow of services, talent and ideas ... Of course, the courts must also recognize the legitimate interest an employer has in safeguarding that which has made his business successful and to protect himself against deliberate surreptitious commercial piracy. Thus restrictive covenants will be enforceable to the extent necessary to prevent the disclosure or use of trade secrets or confidential customer information. In addition injunctive relief may be available where an employee's services are unique or extraordinary and the covenant is reasonable." (Judge Wachtler in Reed, Roberts Associate, Inc. v. Strauman, 40 N.Y.2d 303,386 N.Y.S.2d 677, 353 N.E.2d 590)
As this quotation suggests, modern courts are inclined to disapprove of non-competition agreements for two reasons. The first is that they regard them as unreasonably harmful to the restricted employee. The second is that they regard them as restrictions on competition harmful to the general public. The main argument given in favor of enforcing such agreements in the modern cases is that they protect a firm's legitimate interest in information, whether trade secrets or other valuable information, that an ex-employee might use to his ex-employer's detriment.
The pattern of enforcement, in the case of post-employment restrictions, is largely summed up in the principle that reasonable restrictions will be enforced and unreasonable ones will not be. In practice, this means that the narrower the restriction, in both time and space, the more likely it is to be enforced, on the theory that the narrower the restriction, the less the cost it imposes on the employee and the more likely it is to be limited to what is required for the employer's protection. Further, restrictions are more likely to be enforced against employees who have access to confidential information that was expensive to produce and against "key employees." Finally, restrictions must be given in exchange for some valid consideration and they must protect some valid interest of the employer.
Restrictions associated with the sale of a business are generally subject to a much looser scrutiny, and are much more likely to be enforced.
"(this argument) assumes that all employment contracts are contracts of adhesion, the burdens of which fall entirely on the employee by virtue of his lesser bargaining power." (Blake p. 683) "Personnel offices report that hard-to-get, qualified men are refusing to agree to the impairment of mobility that such covenants entail, or are demanding other concessions because of them." (Blake, p. 627)
From the standpoint of economic theory, the argument described in the first sentence quoted makes very little sense. An employer offers a potential employee some package of salary and other employment conditions which he either accepts or rejects. If the employer really had enough bargaining power to force the potential employee to do whatever he wished-or, put differently, if the potential employee really had no other alternatives-then the package would include a salary of zero (or the legal minimum, in cases where there is one). That is not what we observe.
Assuming that the employer is rational, he will offer the employee the cheapest package he believes the employee would accept-say twenty thousand dollars a year and no non-competition agreement. The reason the employer does not lower his offer to nineteen thousand is that it would then be less attractive than the employee's next-best alternative. But adding a non-competition agreement will have the same effect-since the agreement imposes a cost (in decreased future mobility) on the employee, an offer that was just worth accepting without the agreement is no longer worth accepting with it.
It follows from this argument that, if the cost of the agreement to the employee as perceived by the employee is (say) a thousand dollars a year, then an employee who will just accept an offer of twenty thousand without the agreement will also just accept an offer of twenty-one thousand with the agreement. So the employer who insists on an employee signing a non- competition agreement will find that he must pay, in additional wages or other terms of employment, the cost that the agreement imposes upon the employee, as measured by the employee and revealed in his actions. It follows that the employer will insist on such an agreement only if he believes that its value to him is greater than its cost to the employee. What seems odd about the second sentence quoted from Blake is the assumption that the situation described is in some way new or unusual.
Nothing in this argument depends on the two parties having "equal bargaining power." If the employer has many attractive alternatives and the employee does not, then the two are likely to agree to terms more favorable to the employer than if the situation is reversed. That is true whether or not non- competition agreements are enforceable. But the employer will find it in his interest to use his advantage to insist on a non-competition agreement, instead of using it to get his way on other issues, such as salary, only if the non-competition agreement benefits him by more than it costs the employee.
Nor is the argument affected by who writes the contract. Modern cases have sometimes suggested that the general presumption in favor of freedom of contract ought not to apply to "contracts of adhesion"-contracts written by one party and offered to many others on a take it or leave it basis. The underlying argument is that the party writing the contract can afford to ignore the interests of the party it is being offered to.
But that makes no economic sense. The contract is designed, after all, with the objective of getting the other party to sign it. If I am designing the contract and offering it to many other parties, that may put me in a position to commit myself to insisting on terms that give me a large fraction of the benefit that the contract produces. But it is still in my interest to maximize the size of that net benefit-which I do by only insisting on terms that are worth at least as much to me as they cost the other party. Arguing that contracts of adhesion ignore costs to the other party is rather like claiming that, since automobiles are designed by the manufacturer rather than the purchaser, they will not be equipped with wheels, padded seats, or brakes-and forgetting that what the manufacturer wants to design is a car that the consumer will want to buy.
It appears, from this line of argument, that at least one of the changes in the law during this century has been in the wrong direction, reflecting a decrease rather than in increase in the economic sophistication of the courts. If one accepts the economist's analysis of the situation, a further conclusion follows-as long as the only interests affected are those of the parties to the contract, any contract the parties are willing to sign produces net benefits and should, from the standpoint of economic efficiency, be enforced. If courts refuse to enforce such contracts -as they often do -it follows either that the court is not seeking economic efficiency, that it believes third parties are affected, or that it for some reason rejects the economic argument given above.
One reason a court might reject the argument is paternalism-the belief that employees are not competent to judge their interests for themselves and that the court should therefore step in to cancel a contract that the employee did sign but should not have. While this provides a possible explanation for some behavior of the courts, it is not an entirely convincing one. The employees who sign non-competition agreements are typically executives or professionals, likely to know at least as much about the likely effect on their future career as their employer and a great deal more than a random judge. And courts are frequently willing, in other contexts, to accept individuals' judgements of their own interest on equally difficult issues.
By signing an enforceable non-competition agreement, an employee reduces the alternatives available to him if he leaves his present employment. If the range of the restriction is narrow and the employee's skills are not highly specialized, the effect may be insignificant; a waiter who agrees that if he quits his present employer he will not work for another Middle-Eastern Restaurant in Hyde Park restricts his future options very little. The effect is more serious for a lawyer specializing in bankruptcy who agrees that if he quits his present firm he will not work for any of its competitors.
So one may view a non-competition agreement as a way in which an employer increases the chance that an employee will continue to work for him by eliminating some of the employee's other options. Much of the legal discussion of such agreements seems to assume that doing so is almost always in the interest of the employer and rarely in that of the employee. The employer gets something-an employee less willing to quit-for nothing.
In the previous section, I argued that that view made no sense. In order to induce an employee to sign a non-competition agreement the employer must compensate him for its cost. The employer will insist on such an agreement only if he believes that its value to him is greater than its cost to the employee.
How can that be the case? Consider a situation where the employer provides his employee with valuable and expensive training. The trained employee must later be paid less than his actual productivity in order that there be something left over to pay back the cost of his training. But this is impractical if the employee is free to leave and the training equally useful to a competitor.
The standard solution in the literature is for the employee to pay for such general human capital and for the employer to pay only for firm-specific human capital-training that is useful only to that particular firm. But if the training is expensive, its cost may be more than the employee is in a position to pay. If, for instance, the training costs more than the trainee produces, paying for it himself means accepting a negative wage during the training period-which for many workers is not a viable option.
One solution might be for the employee to borrow, from the employer or from someone else, the money necessary to pay for the human capital. The problem with this is that human capital cannot be used as security. The employee, after receiving the training, can declare bankruptcy and walk away from the debt, taking his human capital with him.
In such situations, either the employer pays for the human capital or it does not get produced. But, once the employer has produced general human capital embedded in a particular employee, what is to keep the employee from walking away with it? A non-competition agreement helps solve that problem. The employee can walk away, but the agreement, if properly designed, prevents him from taking his human capital to any other employer who can use it.
So we have at least one explanation for why employers might propose, and employees accept, such agreements-in order to make it in the interest of the employer to provide the employee with general human capital, either by training him or by giving him access to proprietary information necessary for him to do his job.
A slightly different version of the same general explanation may be important in explaining one of the legal doctrines associated with non-competition agreements in employment-the idea that they are more legitimate when applied to a key employee or one with unique skills. Consider the case of a theater that spends money advertising and promoting a particular performer or a firm that adapts its organizational structure to the talents, tastes, and requirements of a particular high ranking executive. While one could describe either case in terms of firm specific human capital, it seems more natural to describe it as employee specific capital of the firm.
In such a situation, the firm faces two problems. The first is that the employee may leave, wiping out the firm's investment. The second is that the employee may use the threat of leaving to persuade the firm to alter the employment contract in favor of the employee. Both possibilities give the firm an incentive to avoid making such investments, even if they would otherwise be profitable. A long term employment contract enforceable against the employee would solve the problem, but if that option is not available a non-competition agreement provides an alternative solution.
So one interpretation of non-competition agreements is that they are a way in which an employee commits himself to continue working for his present employer. If this is the correct explanation, it seems clear that enforcing such agreements promotes economic efficiency. The contract is in the interest of the parties-that is why they sign it. It harms nobody else. Hence it is efficient.
It is equally clear that courts do not act that way. One of the standard reasons for refusing to enforce a non-competition agreement is that it imposes unreasonably large costs on the employee. But, if the purpose of the agreement is to commit the employee by reducing his future options, that is precisely what it should do. The most natural interpretation of court decisions and dicta from this standpoint is that courts believe that non-competition agreements are used as bonding devices-and shouldn't be.
One explanation for this is that the courts are making a mistake-the particular mistake embodied in terms such as "bargaining power" and "contracts of adhesion." If the court believes that employers can coerce employees into signing such contracts at no cost to the employer, then the refusal to enforce them makes a good deal of sense-the existence of the contract is evidence not that it is efficient but only that it provides some benefit to the employer at the cost of the employee.
A second explanation is that the courts are behaving paternalistically, substituting their judgement of the employee's interest for his. As suggested earlier, this is not an entirely convincing explanation, given the sorts of employees who typically sign (and litigate) non-competition agreements.
A third explanation is that the courts correctly interpret the economics of what is going on, but are still opposed to it. An employee who guarantees that he will continue to work for his employer by signing a binding contract restricting his other alternatives is achieving, in a weaker form, the same result as one who sells himself as a slave or indentured servant-contracts that modern courts refuse to enforce. Consider, as an extreme case, the situation of an employee contractually bound neither to work for any other employer nor to set up in business for himself without the permission of his present employer.
Various reasons have been suggested for the refusal of courts to enforce labor contracts by specific performance. One-that it would be impractical for the court to enforce the contract because of the difficulty of monitoring the performance-does not apply here, since what is being enforced is not an agreement to work for one employer but an agreement not to work for another. But, whatever the reason, courts are in fact unwilling to enforce such contracts, and that may explain why they are reluctant to permit non-competition agreements to serve as substitutes for them.
Another explanation for non-competition agreements is that they are a way of protecting trade secrets. Under the existing common law, trade secrets already enjoy a certain amount of protection. An employer who informs his employees that certain information is confidential and takes reasonable precautions to keep it from becoming public may well have causes of action against both an ex-employee who misappropriates the information and a competitor who knowingly uses it.
Protecting trade secrets in this way involves, however, serious difficulties. For one thing, trade secret law is useful mostly for the sort of information that it is practical to conceal; it provides no protection for information, such as the design of a product, that is revealed when it is used. The same sort of information that can be easily concealed by its creator can also be easily concealed by someone who has stolen it, so the victim of such misappropriation may never know that it has happened. Another problem is that, by enforcing trade secret rights against a competitor, an injured party gives notice to the rest of the industry of the existence of the secret.
To what extent can non-competition agreements provide better protection? At first glance, it might seem that an employee willing to sell trade secrets could easily do so without taking a job with the firm he was selling them to. For some trade secrets this is undoubtedly true; for others it is not. A complicated industrial process that takes a year to learn will also take a year to teach.
There are several other advantages to hiring someone who is selling you a trade secret. For one thing, it makes it much easier to claim that you did not know he was doing anything wrong. Handing someone an envelope with a hundred thousand dollars in unmarked bills in exchange for a set of drawings is much harder to explain away than hiring an experienced professional at a generous salary.
A further advantage arises from the nature of property in information. Once the ex-employee has sold the information to one competitor, what is to stop him from selling it to another? Yet the more competitors have the information, the less valuable it will be to each-so the first purchaser may be reluctant to pay very much for what will soon become common knowledge. One solution is for the ex-employee to start his own firm, on the theory that he can trust himself to keep his mouth shut.
For all of these reasons, non-competition agreements provide a way of protecting trade secrets. They also provide a way of protecting proprietary information that, for one reason or another, does not qualify for protection under trade secret law-a point I shall return to in the next section.
In a previous section, I argued that such agreements would be signed only if they were in the joint interest of both parties. In this case they are. The value to a defaulting employee of being able to misappropriate a trade secret is what the competitor will pay for it. That is less than the value of the secret to the firm that created it; if it were not, the firm would have sold the secret to the competitor itself. In addition, the security provided by such an agreement will make the firm more willing to invest resources in producing information, which is a net gain for the firm that costs the employee nothing. So this is a plausible explanation of why such agreements are signed.
Are they efficient? To answer that question, it is worth considering why a trade secret is likely to be worth less to a competitor than its loss costs its owner. At first glance, one might expect the opposite. The owner can still use his "secret"-and so can the competitor. The owner may lose some business to the competitor-but is that not simply a transfer from one firm to the other?
The answer is no. Consider a trade secret that lowers production cost by $10/unit. If one firm has it, the secret is worth $10 times that firm's volume of production. If ten firms have it, they compete the price of the product down by $10 and each is no better off than it would have been if the secret had never been discovered.
As this example suggests, the transfer of a trade secret typically benefits not only the recipient but also the customers of the two (or more) firms. So the fact that keeping the secret produces a net gain from the standpoint of firm and employee, or firm and competitor, does not tell us whether it produces a net social gain. Once a trade secret exists, increasing the number of firms that can use it produces net benefits-the gains to the customers outweigh the loss to the firms.
The stealing of trade secrets affects not only how they are used but also whether they exist. The more secure trade secrets are, the more it will pay firms to invest in producing them. So protection of trade secrets produces a benefit (more spent in research whose social payoff is larger than its cost) and a cost (less use of existing secrets). The logic is the same as in the case of patent law, where the restriction of output due to the patent monopoly is justified as a way of encouraging valuable research.
The existence of trade secret law suggests that the courts believe the benefits are sometimes worth the costs. The limitations in the law suggest that the courts believe they sometimes are not.
The pattern of enforcement of non-competition agreements seems consistent with this explanation of why they exist. Courts are more willing to enforce agreements against employees who had access to valuable proprietary information. They are also more willing to enforce agreements restricted, in time and space, in a way that seems designed to protect such information.
The one part of the pattern that seems not to fit this explanation is the court's concern with balancing the benefit to the employer against the burden to the employee. The employee must gain by signing the contract, else he would not do so. If the contract generates net costs, it is because of costs imposed on third parties who are not signatories.
One way of resolving this puzzle is to combine the discussion of this section with that of the previous section, assuming that courts believe that firms use non-competition agreements for both of the purposes we have discussed. Thus one might interpret the pattern of enforcement as an attempt to limit firms to non-competition agreements that protect trade secrets while preventing agreements intended to keep employees from leaving.
Courts, in refusing to enforce non-competition agreements, sometimes suggest that such agreements may reduce market competition. It is worth considering under what circumstances we would or would not expect to observe such contracts used for that purpose.
Consider a perfectly competitive market. Each existing firm produces a very small fraction of total output. In this context, it is hard to see why a firm would be willing to pay anyone, whether its previous owner or an employee, any substantial sum to agree not to become the millionth competitor.
Next, consider an oligopoly-an industry with a small number of firms selling essentially identical goods. Each firm would like to prevent new firms from coming into existence, so they have some incentive to try to arrange non-competition contracts. But the injury done to existing firms by an employee of firm X who starts a new firm is divided among all the existing firms, so it is unlikely that firm X by itself will be willing to bear the cost of preventing him from doing so. While non-competition contracts are conceivable in such a situation, they seem unlikely unless there is some way in which firms can fund them jointly.
In the case of a natural monopoly, competition is not an issue. The same is true for a monopoly that maintains its position by legal privilege. We would not expect to see non-competition contracts in either case.
We are left with the case of monopolistic competition. A firm spends resources acquiring knowledge and skills specialized to a particular part of the marketplace-a particular variety of the product in a particular geographical location. A long term employee who leaves the firm may possess much of that knowledge and so be able to start a new firm at a much lower cost. Since the knowledge he has is specialized to the particular market and geographical niche occupied by his previous employer, his new firm will compete with that employer more directly than with other firms in the industry. It therefore seems plausible that the employer will be willing to offer him a payment in exchange for his agreement not to do so.
The logic of the situation is analyzed below in terms of a simplified example, and in more detail in the appendix. Under at least some circumstances, as I will demonstrate, the reduction in the profit of an existing firm due to a new firm started by an ex-employee is greater than the profit of the new firm; the employer will therefore be willing to offer the employee more to stay than he would make by leaving. The same analysis applies to the case of the owner of a firm who sells his firm, along with the knowledge and skills embodied in it, but preserves the ability to reproduce them.
Monopolistic competition thus provides a plausible environment in which non-competition agreements might be expected to arise and one that fits the observed nature of such contracts-agreements not to compete in a particular geographical area for a fixed length of time. After enough time has passed, the specialized knowledge of the employee or previous owner will have depreciated enough so that it is no longer worth his while to start a new firm, or, if he does, it will no longer be so similar to the old firm as to compete directly for the same market niche.
We start by assuming an industry characterized by monopolistic competition. A firm comes into existence by spending $10,000,000 producing the specialized skills and knowledge necessary to sell to a particular market segment; we may think of this as purchasing a particular location. We will assume that this is the only fixed cost and that marginal cost is constant on the margin of how much is produced. An individual sufficiently familiar with an existing firm-either a high ranking executive or the firm's founder-can take advantage of the knowledge produced by the firm to purchase a nearby location at a cost of only $1,000,000.
We will assume that, when there are two (or more) firms at almost the same location, they divide the market evenly. A more realistic model of firm behavior is analyzed in the appendix.
Using these assumptions, we will first look at the case of non- competition agreements between buyers and sellers of firms, then at agreements between employers and employees. In each case, we are interested both in whether the agreement is in the interest of the parties, and is therefore likely to be made, an in whether it is in the net interest of everyone affected-is economically efficient.
We imagine that the founder of a firm can expect to operate it effectively for ten years. After that time, his ability declines as he gets older, so marginal production cost increases and profits fall. He can, however, sell the firm to a new and younger manager who will be able to restore the original level of costs and profit.
Suppose that non-competition agreements associated with the sale of a firm are enforceable. Further assume a market interest rate of 10%. In equilibrium, firms enter the industry until there are enough so that the individual firm receives $1,000,000/year after paying its marginal cost, giving it a normal market return on its $10,000,000 investment. After ten years a firm's profits begin to decline, due to the increasing age of its founder, so he gets his money back by selling out to a new owner at a price of $10,000,000, agreeing not to start a new firm.
Now suppose that non-competition agreements are not enforceable. If the aging owner sells his firm, he can start a new one at a cost of $1,000,000 and earn slightly less than $500,000/year (slightly less because of his decreased managerial ability); clearly it is worth doing so. But his action reduces the revenue of the firm he just sold to $500,000 a year, so the purchaser, anticipating the situation, is only willing to pay $5,000,000 for it (I assume, again for simplicity, that the founder can only pull the trick once-after that he is too old, or too rich, for it to be worth his while to found a third firm).
One conclusion is that if founders can sign enforceable non-competition agreements, they will. The agreement increases the selling price by $5,000,000. Its cost is that the founder cannot found a second firm; since the second firm would cost $1,000,000 to found and have a present value of under $5,000,000, the founder's return from founding it is under $4,000,000.
A second conclusion is that, if such agreements are not enforceable, the founder will continue to run his firm well past the time when it would be more efficient for someone else to do so-because he is the only one who can enforce a non-competition agreement against himself. As he becomes less and less capable of running it, both the price at which he is better off selling and the profit he will make by starting a second firm after selling decline. When either his sales price drops below $5,000,000 or his profit from starting a second firm drops below zero, he sells. One consequence of the unenforceability of non-competition agreements in this context is a sizable deadweight cost, in the form of inefficient management of the firm.
A third conclusion is that the unenforceability of such contracts will change the initial equilibrium. If it did not, if we still had a density of firms such that a well managed firm could make only $10,000,000/year, the founder of a new firm would get less than the market return on his investment. Instead of making $1,000,000/year for ten years and then selling for $10,000,000, he would make $1,000,000/year for ten years, then a return declining gradually from $1,000,000/year to $500,000 (possibly beyond), then sell out for $10,000,000. He would get back less than a present value of $10,000,000 in exchange for his $10,000,000 investment.
It follows that, if non-competition agreements are not- enforceable, fewer firms will enter the industry, giving an equilibrium where a well managed firm makes (say) $11,000,000 a year, providing an above market return in the early years to balance the later below market return. If founders of such firms end up selling out for $10,000,000 and not starting new firms (they are too old), the number of firms when non-competition agreements are not enforced is permanently lower than the number when they are enforced.
In this case, at least, enforcing non-competition agreements leads to more competition, not less. From the standpoint of the people signing the agreement its purpose is to prevent competition-specifically to prevent the founder of the firm from competing with the purchaser. But the ability to use such contracts to increase the price at which firms can be sold increases the profitability of starting firms, and thus the number of firms in the industry.
So far I have implicitly assumed that there are no good substitutes for non-competition agreements, no alternative ways in which the seller can commit himself not to compete with his old firm. If such substitutes exist, then making non-competition agreements unenforceable may merely mean that the parties substitute other, and presumably more costly, arrangements. One obvious example, and one that is sometimes observed, is for the payment to the seller to be spread out over time, with the amount depending in part on the performance of the sold firm. Alternatively, the firm may hire its previous owner as a consultant-with the understanding that they will stop paying him if he starts his own firm or goes to work for a competitor.
So far we have been considering agreements between buyers and sellers of firms. Next we will consider the case of non-competition agreements between firms and their employees, using the same model of monopolistic competition but dropping the assumption that costs rise after ten years.
We start with the simple situation where there is only one employee with sufficient expertise to start a competing firm. If non-competition agreements are legal, the firm demands such an agreement as a condition of employing him. He agrees, since the cost to him of signing the agreement is $4,000,000 foregone (the value to him of being able to leave and start his own firm once he has acquired the necessary information and skills) and the value to the firm of his signing it is $5,000,000. Since the firm has access to alternative employees at their alternative wage, he does not actually get paid anything extra-if he refuses to sign, the firm offers him his alternative wage minus $5,000,000, and he declines.
Next suppose that non-competition agreements are not enforceable and that there is no way of running the firm without eventually giving the employee the information necessary to run his own firm. In this case, once the employee acquires the information, the firm gives him a raise of somewhat over $400,000/year, thus making it in his interest to stay with the firm rather than founding his own. While this is an expensive solution from the standpoint of the firm, it is better than letting the employee start his own firm and reduce profit by $500,000.
Just as in the case of selling firms, refusing to enforce non-competition agreements alters the initial equilibrium. The firm's costs are higher by about $400,000/year. So the present value of the firm's profits would be less than $10,000,000 if we started with the same density of firms that we had with non-competition agreements. This means that in equilibrium, where every firm just covers its costs, there will be fewer firms than if non-competition agreements were enforceable. Here again, non-competition agreements are used to prevent competition, but their actual effect is to increase it.
Finally, consider the case where there are several employees, any one of whom could found a new firm. If non-competition agreements are legal, the employees will sign them. If they are illegal, the firm will find that paying a bonus of $400,000/year to each such employee reduces its profit below zero, yet if it does not do so some employee will start his own firm. The problem here is that unless the firm pays each of (say) five employees as much as that employee would make if he were the only one starting a new firm, at least one of the five will start a new firm. The existing firm is not willing to do that, so ex-employees start firms until the profit from doing so (net of cost of starting) is zero, which in this case means until the return (net of marginal cost, gross of starting cost) is down to $100,000/year.
In this case, the initial number of firms in the industry is smaller than if non-competition agreements were enforced, since firms know that after their employees acquire the necessary information, profits will drop abruptly with the entry of many new firms. But the ultimate number of firms in the industry will probably be much higher than if the agreements were enforced. In this case at least, non- competition agreements may actually reduce competition-eventually.
So far, I have been assuming that firms have no substitute for non- competition agreements, other than overpaying their workers enough so that it is not worth quitting to start a new firm. In fact, there are at least three important substitutes worth considering. One is modifying the firm's activities in such a way as to prevent any single employee from acquiring the knowledge necessary to start his own firm. A second is to hire for the relevant positions family members, close friends, people who for one reason or another can be trusted to abide by a non-competition agreement even if it is not enforceable in the courts. A third is to use some combination of entrance fee and salary schedule to charge the employee in advance for the high salary that will later keep him from quitting to start his own firm.
Consider the situation with one key employee, and assume he has unlimited access to capital. When he gets the job, he pays an initiation fee of several million dollars-just enough to compensate his employer for the extra $400,000 a year he is going to be paid once he acquires the information necessary to start his own firm. Alternatively, the employee could post a $4,000,000 bond with the firm, to be returned as a pension dependent on the firm's profits after he leaves it, on a schedule that eliminates the pension if profits fall to or below $500,000.
There would obviously be some practical problems in drawing up and enforcing such contracts-in particular, there must be terms preventing the employer from firing the worker and keeping the bond. There would also be legal problems with at least some versions. And, if the sums are as large as in my example, there are also practical problems in finding potential employees with access to that much money. For smaller sums, however, some version of bonding or buying future high salaries, perhaps by accepting a low salary in the early years of employment, might be a workable substitute for non-competition agreements.
One conclusion suggested by this example is that if the alternatives to non-competition agreements, other than overpaying key employees to keep them from starting their own firms, are either impractical or illegal, then enforcing non-competition agreements in the case where there is only one employee capable of starting his own business results in more firms, not fewer. In the many-employee case, enforcing such agreements results in more firms in the short term, but probably fewer in the long term.
In the Appendix, I work out this analysis more precisely in two versions of a somewhat more realistic model of monopolistic competition. I calculate the net effect in terms of the tradeoff between short term (more firms are started if non-competition agreements are enforceable, since firms can expect to maintain their market position longer) and long term (if the agreements are not enforceable, ex-employees eventually start additional firms). The conclusion is ambiguous: If there are no good substitutes for non-competition agreements, making such agreements unenforceable may or may not increase social welfare.
In the first version of the model, the result of making the contracts unenforceable is not only superior to the result of making them enforceable, it results in the economically efficient pattern of behavior-no alternative rule for controlling the number of firms started by ex-employees could produce a superior result. In the second version, making the contracts unenforceable means that no firm in the industry will be able to cover its fixed cost, so no firm is ever started-a result economically inferior to the result if the contracts are enforceable. It seems fairly clear that further variations in the details of the model could produce additional results between these two extremes. It follows that, so far as economic theory is concerned, we cannot predict whether forbidding non-competition agreements will increase or decrease competition-or, more precisely, whether the effect of doing so on the number of firms will or will not be desirable from the standpoint of economic efficiency.
If substitutes for non-competition agreements are available, the case against enforcing the agreements becomes weaker. Refusal to enforce them may simply mean that firms switch to producing the same result in more costly ways. If so, we will have fewer firms initially (with higher costs, fewer firms enter the market), fewer firms later, and an economically less efficient outcome than if the agreements were enforceable.
What Courts Do
The general standard of "reasonableness" adopted by courts seems to suggest that, if non-competition agreements are intended to protect monopolistically competitive firms from the competition of their ex-employees, the courts approve. The limitation of such agreements in both area and duration seems consistent with the idea that firms are entitled to keep ex-employees out of their market niche, at least until the relevant specialized information becomes obsolete. While the courts do not regard the restriction of competition as itself a legitimate interest of the employer deserving of protection, the categories of interest that are considered legitimate seem sufficient to cover most cases of monopolistic competition; if the argument of this section is correct, that means that they cover most of the situations in which a firm would find it in its interest to use such contracts to restrict competition.
On the other hand, the tendency, especially in modern cases, to refuse to enforce such agreements against employees who do not have access to proprietary information narrowly defined might be taken as evidence that it is the protection of trade secrets rather than the prevention of competition that they consider a legitimate objective of the firm, to be balanced against the interest of the employee and perhaps the public.
We have seen that in two of the three situations considered (selling firms and firms with one employee capable of starting a new firm) the enforcement of non-competition agreements actually increases the number of firms and thus, in some sense, "increases competition." In the third case, the long run results are more complicated. There are fewer firms in the first generation but may be more in the second, and the effect on overall economic efficiency is ambiguous.[55 ]This seems to fit, at least roughly, the behavior of modern American courts, which have tended to enforce non-competition agreements in the first (sale of firm) and in the second (key employee) cases but not always in the third.
If so, we have a context in which the position of the courts makes sense. In the two cases where refusal to enforce non-competition agreements has only bad consequences, they tend to enforce them. In the one case where the refusal may have desirable consequences, they frequently refuse-although they often seem to do so for the wrong reason.
Section VI: Conclusion
In this essay I have tried to suggest some possible explanations for the selective enforcement of non-competition agreements by modern common law courts. One reason for not enforcing such contracts, and the only one that makes sense out of the courts' emphasis on avoiding contracts that impose unreasonable costs on the employee, is that the court rejects the economic argument for freedom of contract. One possible reason is that judges do not understand economics. Another is that they understand it but reject the underlying assumption that the contracting parties are rational-the judges are substituting their opinion of what is in the employee's interest for his. A third is that the judges reject the objective of economic efficiency, at least in the case of contracts that verge on indentured servitude.
A second reason for rejecting such contracts is suggested by occasional dicta on the subject of protecting the public or maintaining competition. Under some circumstances, non-competition agreements in a monopolistically competitive industry may produce a sort of ex ante cartel. They are used to enforce an agreement not to compete between a firm and the individuals who will be most qualified to start competing firms. The economic analysis of such situations is complicated, but it seems that in such circumstances the courts' reluctance to enforce such agreements may sometimes be justified on efficiency grounds.
Do the changes that have occurred in this part of the common law over the past century represent progress or regress? The answer is both. They represent a slight improvement in economic sophistication insofar as modern courts seem somewhat more willing than 19th century courts to take seriously the possibility that non-competition agreements might actually prevent competition, and thus impose costs on the customers of the contracting parties. In another respect they represent a slight worsening-the retreat from freedom of contract has been justified by doctrines that, as I argued in Part II, make very little economic sense. In this respect, the much maligned judges of the period of laissez-faire were better economists than their successors. Yet even they were unwilling to carry the principle of freedom of contract as far as might seem appropriate from the standpoint of economic theory. By requiring that post-employment restraints must be "reasonable" in order to be enforceable, they, like modern judges, implicitly refused to enforce agreements designed as commitment devices.
The purpose of this appendix is to work out the consequences of refusing to enforce non-competition agreements, using two models of a monopolistically competitive market. The objective is to learn whether forbidding such agreements affects economic welfare in a predictable way.
Certain assumptions will be common to both models and all cases considered. They are:
1. There are no effective substitutes, through bonding or the like, for non-competition agreements; if such agreements are not enforceable (or not used), employees are free to start their own firms.
2. The cost of starting a firm at time t=0 is C1. By working for such a firm for a period of time T1 some fixed number of employees acquire the information that makes it possible for them to start his own firms at a cost of C2<C1. After a further period of time T2 the world, or at least the industry, ends.
3. The only cost that firms face other than the startup cost is a marginal cost of MC per unit of output.
4. Employees have enough resources so that they can, if they wish, accept a reduction in salary corresponding to the value of the opportunity to later start their own firm.
Assumption 1 is important, since without it a legal rule forbidding non-competition agreements may merely force the parties to achieve the same result in a more expensive way. Assumptions 2 and 3 are made for purposes of analytic simplicity. Assumption 4 is convenient but not essential; I will show that the principal result does not depend upon it.
Under monopolistic competition, firms enter the industry until profit is driven to zero. In the first generation, firms will enter until total expected revenue during T1+T2 is driven down to total expected costs, including C1. In the second generation, new firms will be started by ex-employees until either all ex-employees who can do so have started firms or the expected revenue during T2 has been driven down to the cost of entry, equal to C2 plus the cost (if any) of refusing to sign a non-competition agreement. I define:
P1: The equilibrium price during T1
P2: The equilibrium price during T2
Q1: Number of units sold per month per firm during T1
Q2: Number of units sold per month per firm during T2
[[pi]] 1: Profit per month per firm during T1
[[pi]] 2: Profit per month per firm during T2
N: The number of ex-employees per firm who start their own firm at T1
For simplicity, I assume an interest rate of zero. Introducing a positive interest rate would have the same effect as increasing T1 relative to T2.
Figure 1 shows a very long street with barber shops evenly spaced along it. Customers (not shown) are evenly distributed along the street, with a density of k customers per block; they are all identical except for where they live. Every customer has his hair cut once a month. Customers wish to minimize the combined cost, in money and walking, of getting a haircut. A customer regards an extra block of distance from the barber shop as equivalent to an extra dollar in price, so the combined cost is W+P, where W is the distance from the customer to the barbershop and P the price of a haircut in dollars. Each customer goes to the barbershop for which W+P is smallest.
Figure 2 shows the situation from the standpoint of barber L on figure 1. The lower the price charged, the more customers come, as shown by the demand curve D. Each time he lowers his price by a dollar he gets an additional block of customers-half a block of customers to the west, who now find that L is more attractive than K, and half a block to his east, who now find that L is more attractive than M. Since a block has k customers living on it, D has a slope of 1/k-each one dollar drop in price increases the number of customers (per month) by k. The equation of D is Q=k(Pmax-P)
For a given d, we solve for P* as follows:
Solving the equation of D for P we have:
P=Pmax - Q/k
The equation of MR is:
MR = Pmax - 2Q/k
for Q = kd we have:
MR = MC = Pmax - 2d
Pmax = MC+2d
P* = Pmax - d = MC+d
If the fixed cost of being a barber is C dollars/month, we have:
Profit = [[pi]] = TR - VC - C =(P*-MC)kd - C=kd2 - C
In equilibrium profit must be zero, so we have:
C=kd2 Equation 1
So far, we have been considering the situation for a single month. Generalizing to a situation in which d may be changing over time, Equation 1 becomes:
where the integral is over the entire time period during which the firm operates and C is the fixed cost necessary to operate over that time period.
We now apply this analysis to our simple two period model. At time t=0, anyone who wishes can start a firm at a cost of C1. At time t=T1, N ex-employees of each firm start their own firms, each paying a cost C2.
I assume that, as discussed earlier, ex-employees who start firms start them "close to" their old firm-what they have learned from their employment lowers the cost of starting a firm, but only a firm similar to the one they worked for.  During the second period, the old firm and its daughter firms share the market (represented by a section of the street d1 blocks long) that was controlled by the old firm in the first period.
Without that assumption or something similar, non-competition agreements are unlikely. The ex-employees of a firm that did not require them to sign non-competition agreements could spread out over the whole market. So unless the number of employees of one firm was very large or the market very small they would not drive profit in the second period down to zero. They would therefore be willing to pay their original firm a significant price for permission to start their own firm. But a firm's ex-employees would be competing mostly with other firms, so giving them permission would cost their employer very little. This is the point I made earlier about why non-competition agreements are more likely under monopolistic competition than under oligopoly.
A potential employee who knows that at time T1 he will be able to start his own firm will take that fact into account in his initial decision to accept employment at time t=0. Suppose there is a pool of employees willing to accept employment from t=0 to t=T1 for a total reimbursement of w. If an employee does not expect to start his own firm, either because he has signed a non-competition agreement or because he knows that starting a firm at T1 is not going to be profitable, his reimbursement consists entirely of wages. On the other hand, if the employee expects to start his own firm and make profit [[pi]] , he will require only w-[[pi]] in wages to make him willing to accept the initial job offer. In effect, he is paying his employer a price for the training that makes it possible for him to later start his own firm. This price equals [[pi]] and is a function of N, since the more employees start firms at t=T1 the less each will make. So we may write [[pi]] =[[pi]] (N).
We then have two zero profit conditions. For the first generation firm, we have:
C1=(P1-MC)Q1T1 + (P2-MC)Q2T2 + N[[pi]] (N)
= k(d12T1+d22T2) + N[[pi]] (N) (1)
Here the term N[[pi]] (N) represents the "revenue" (in the form of a lower wage bill) that a first generation firm receives for training N employees.
For the second generation firm, we have:
C2=(P2-MC)Q2T2 - [[pi]] (N) = kd22T2 - [[pi]] (N) (2)
Here the final term represents the payment by the ex-employee to his previous employer (in the form of accepting lower wages) for the training that permitted him to start his firm. We could rearrange equation 2 to read:
[[pi]] (N)=(P2-MC)Q2T2 - C2 = kd22T2 - C2 (2')
It then shows gross profit (not including payment for training) as equal to revenue minus cost. Net profit is, as we would expect, zero, since the owner of the new firm has paid his gross profit to his previous employer.
Solving for d1 and d2 we get:
In addition to these equations, we also have two constraints. Employees cannot start a negative number of firms, so N>= 0. Employers will not pay employees to start firms, since an increase in the number of firms in the second generation lowers the profits of the existing firms, so we must have a value of N such that [[pi]] (N)>= 0. With a little manipulation, these yield the constraints:
The next question is how N affects social welfare. So far as the producers are concerned, it does not matter; free entry to the industry implies zero profit; whatever the value of N, the producers make just enough to cover their costs. What about the consumers?
Under all circumstances, consumers get the same number of haircuts-one per consumer per month. The difference is in what they pay for them. The average consumer is a distance d/4 from the nearest barbershop, so his cost per haircut, in walking and money, is P*+d/4=MC+d+d/4 = MC+5d/4. Integrating this over the entire time period, we have:
Cost of haircuts=(MC+5d1/4)T1+(MC+5d2/4)T2
=MC(T1+T2)+(5/4)(d1T1+d2T2) = MC(T1+T2)+
= MC(T1+T2) +
We would like to know how the cost of haircuts depends on N. We have:
(Cost of haircuts)=
By inequality 6 above, the final bracket, and thus the whole expression, is <= 0.
So cost falls as N, the number of second generation firms per first generation firm, rises. This is true as long as N is sufficiently small so that a second generation firm generates non-negative gross profit, that being the requirement from which inequality 6 was derived. Beyond that point the derivative becomes positive; a further increase in the number of firms raises cost. So cost to customers is minimized and social welfare maximized at the level of N for which [[pi]] (N), gross profit to a second generation firm, is zero. This is precisely the outcome that occurs if non-competition agreements are forbidden. Employees leave first generation firms to start second generation firms until the profit from doing so is driven to zero. 
What happens if non-competition agreements are permitted? A first generation firm chooses N, the number of its employees not required to sign non-competition agreements, to maximize its profit. Since each first generation firm must take the behavior of all other first generation firms as given, d1 is fixed; the firm chooses N (and hence d2) to maximize its profit.
Equation 7 shows that the first generation firm, by charging its employees for permission to start second generation firms, is receiving both its own profits and the profits of N second generation firms. It receives the income of one firm (itself) during the first period and N+1 firms (itself plus N daughter firms) during the second period, while paying the corresponding costs.
The firm chooses N to maximize its profits, d1 fixed, so we have:
Profit(N) = <0
The firm's profits decrease as N increases, so the firm chooses N=0; all employees are required to sign non-competition agreements.
We have now shown that, in this particular case of monopolistic competition, a legal rule making non-competition agreements enforceable results in N=0 (no daughter firms, all employees agree not to compete), and a legal rule making them unenforceable results in the optimal level of N. It follows that, for this model, the efficient rule is for such agreements to be unenforceable.
So far, we have been assuming that employees have sufficient resources to pay [[pi]] (N) in reduced wages in exchange for not signing a non-competition agreement. As I suggested earlier, this may not always be a realistic assumption. Fortunately, the result of the previous paragraph does not depend on it. If the assumption does not hold, non-competition agreements become even more attractive from the standpoint of the employer, since the employees are less able to pay him for not signing them. So the proof that, if non-competition agreements are enforceable, the employer will choose to require all employees to sign them (set N=0) holds a fortiori. So comparing the results of permitting or not permitting non-competition agreements again means comparing the efficiency of two outcomes, of which one (permitted) has N=0 and the other has [[pi]] (N)=0. Since in the first case nobody is paying the price and in the second the price is zero, employees' ability to pay the price is irrelevant to the efficiency of the outcome.
So it appears that, in this model, refusing to enforce non-competition agreements is economically efficient. Does this result hold in general? The answer is no. To prove that, it is sufficient to give a second model of monopolistic competition-one in which refusing to enforce noncompetition agreements lowers social welfare.
Consider our street of barbers, with one additional assumption. Individuals can cut their own hair, and will do so if the cost of a haircut, in money plus travel, is greater than some Pmax > MC. How are the results changed?
So long as d, the distance between barbers, is small, the solution is the same as before-going to a barber is always cheaper than cutting your own hair. As d increases, however, it eventually reaches a point dcritical at which the customer halfway between two barber shops finds it in his interest to cut his own hair. Thereafter, further increases in the distance between barber shops have no effect on the price they charge or the profits they make, since what determines the demand curve faced by one barber is not the customer's alternative of going to another barber but his alternative of cutting his own hair. So there is some maximum profit per month [[pi]]max = [[pi]](dcritical) that a firm can earn.
Now suppose that (T1+T2)[[pi]]max > C1 >T1[[pi]]max. Further suppose that C2=0. If noncompetition agreements are enforceable, there will be some d<dcritical at which firms can just cover their costs. If they are not enforceable, new firms will be able to enter costlessly at t=T1, so profit will be zero in the second period. Firms cannot earn more than T1[[pi]]max in the first period, so it is impossible for a firm to cover its fixed cost, so no firms are ever started. Those customers who, if the agreements had been enforceable, would have gone to barbers now have to cut their own hair, so are worse off. Under this particular model, with these values of C1 and C2, society is unambiguously worse off if noncompetition agreements are not enforced.
The first model demonstrates that there are circumstances in which a legal rule making non-competition agreements between employers and employees unenforceable is economically efficient; the second model demonstrates that there are circumstances in which it is not. The conclusion is that, so far as we can tell from the theoretical analysis, either result might apply to an actual case in the real world.
It is worth noting some important limitations in the analysis. I have assumed away all objectives for non-competition agreements other than preventing competition. Furthermore, I have assumed that non-competition agreements are the only way in which firms can prevent competition by ex-employees. Model 1 might give a very different result if firms could, at a cost, prevent employees from acquiring the information necessary to start their own firms. In that situation, the unenforceability of non-competition agreements might merely mean substituting costly precautions for inexpensive contracts.
Gary S. Becker, Human Capital, Columbia University Press (1964).
G.C. Cheshire and C. H. S. Fifoot, The Law of Contract, seventh edition, London, Butterworths (1969).
Harlan M. Blake, "Employee Agreements Not to Compete," 73 Harv. L. Rev. 625, 681-4 (1960)
E. H. Chamberlin, The Theory of Monopolistic Competition, Cambridge, MA: Harvard University Press, 1933.
David Friedman, Price Theory: An Intermediate Text 2d Edn., South-Western Publishing Co., Cincinnati, 1990.
David Friedman, William Landes and Richard Posner,"Some Economics of Trade Secret Law," Journal of Economic Perspectives, Volume 5, Number 1 (Winter 1991), pp. 61-72.
Edmund W. Kitch, "The Law and Economics of Rights in Valuable Information," JLS IX (1980), p. 683
Joseph Bridges Matthews and Herbert M. Adler, The Law relating to Covenants in Restraint of Trade, 2nd edition London Sweet & Maxwell, 1907.
Milton Handler and Daniel E. Lazaroff, "Restraint of Trade and the Restatement (Second) of Contracts," NYU Law Review 57:669, (1982).
Margaret N. Kniffin, "Employee Noncompetition Covenants: The Perils of Performing Unique Services," Rutgers Camden Law Journal, Volume 10 Number 1 (Fall 1978).
Sir Frederick Polock, Bart., Principles of Contract, Stevens and Sons, Ltd., London 1889.
Todd D. Rakoff, "Contracts of Adhesion: An Essay in Reconstruction," 96 Harv. L. Rev. 1174, 1242-43 (1983).
Joan Robinson, The Economics of Imperfect Competition," London: Macmillan, 1933.
Paul H. Rubin and Peter Shedd, "Human Capital and Covenants Not to Compete," JLS 1981 p. 93
William Alexander Sanderson, Restraint of Trade in English Law, London: Sweet &Maxwell 1926
George J. Stigler, "Restraints on Trade in the Common Law," in The Organization of Industry 255 (1968).
Michael J. Trebilcock, The Common Law of Restraint of Trade. Carswell, Toronto, 1986.
Anthony C. Valiulis, Covenants Not to Compete, Wiley New York 1985.
Oliver E. Williamson, "Credible Commitments: Using Hostages to Support Exchange" (1983), 73 Am. Ec. Rev. 519.
Back to my home page.