[* ]Olin Fellow, the University of Chicago Law School. I would like to thank my colleagues, especially Walter Blum, Frank Easterbrook, Abner Green, Geoffrey Miller and Richard Posner for useful comments and suggestions on earlier drafts. I am grateful to the Olin Foundation for financial support.


1Non-competition agreements also appear in the context of franchising; a franchisee may be required to agree that, if he gives up his franchise, he will not start a competing business. "A franchisee is entitled to the same protection as an employee, and a covenant not to compete in a franchise agreement is subject to the same strict rules of construction as a covenant not to compete in an employment contract. They are thus enforceable if reasonable and consistent with the public interest." Valiulis pp. 8-9. Other contexts in which non-competition agreements appear are among shareholders in a corporation (agreeing not to compete with the corporation), in patent licensing agreements, and in connection with the sale or leasing of real estate. (Valiulis pp. 10-12). These contexts will not be discussed in this article.


Many of the states of the U.S. have restricted the use of non-competition agreements by statute; see Valiulis p. 7 . In some cases, the Sherman Act may make restrictive covenants illegal and unenforceable (Valiulis pp. 33-42), although in practice it has rarely been used for that purpose. "Judge Taft held that any restrictive covenant affecting interstate commerce that was unenforceable at common law also violated the Sherman Act. Subsequent cases have firmly established that inquiry under the Sherman Act is coextensive with that at common law." Handler and Lazaroff (1982) pp. 750-751, commenting on United States v. Addyston Pipe & Steel Co. In this article I will be concerned with the common law treatment of such agreements, not with statutory modifications thereof.

2See Blake, pp. 640-642 and Sanderson, pp. 18-19.

3The case (Dyer's Case, Y.B. Mich. 2. Hen. 5, f.5, pl. 26 (C.P. 1414) ) involved a writ of debt upon an obligation which, according to the defendant, was "to to lose its force if he did not practice his trade for a period of six months in the plaintiff's town" (Blake p. 636).

4(2. Bulst. 136, reprinted in 80 English Reports 1012, also 11.Jac. 1. Roll 223, reprinted in 79 Eng Repts 278-279). It is sometimes asserted that prior to Rogers v. Perry all such contracts were invalid. Since many of the early cases survive only in the form of pleas and verdicts in unpublished manuscripts, it seems unlikely that any such categorical negative has been or can be established.


Various explanations for the refusal of early common law courts to enforce contracts in restraint of trade have been suggested. One is that contemporary labor markets were so rigid, due in part to guild restrictions, that a craftsman bound by contract not to practice his trade was likely to become a burden on the public. Another is that the courts were protecting apprentices and journeymen from " `unethical' masters attempting to prolong the traditional period of subservience" (Blake p. 632). There seems to be very little evidence for any of these guesses. It is worth noting that Dyer's Case predates by a century and a half the establishment of a national welfare system (the Elizabethan Poor Laws) upon which the unemployed might be a burden-although there were, of course, earlier local institutions for dealing with the poor.


Blake's conjecture that the court's were protecting apprentices and journeymen from their masters finds little support in the four cases he cites. In only one is the contract stated to be between a master and an apprentice, and it is unclear in that cases whether the agreement was an attempt to prolong the period of apprenticeship or to protect the master against the competition of an apprentice setting up on his own before the period of apprenticeship is over. None of the other three cases contains any evidence that the promisee was a master and the promisor his journeyman or appentice.

5 1 P. Wms. 181, 24 Eng. Rep. 347 (Q.B. 1711). This is probably the most cited case on common-law restraints of trade. It involved the transfer of a bake shop with restriction (seller not to bake in the parish for the term of the lease). The contract was upheld. It is the first reported case to discuss in some detail what sort of non-competition agreements should or should not be upheld, and why.


Lord Macclesfield rejected the view that bonds (liquidated damages provisions) were invalid and promises valid, arguing that: "the true distinction of this case is, not between promises and bonds, but between contracts with and without consideration;... ."

6 "When once it is admitted that whether the covenant be general or particular the question of its validity is alike determined by the consideration whether it exceeds what is necessary for the protection of the covenantee, the distinction between general and particular restraints ceases to be a distinction in point of law." (Lord Herschell, L.C. in Nordenfelt v. Maxim-Nordenfelt Guns and Ammunition CO., Lim., (1894) A.C. 535. This was confirmed by the House of Lords in Mason v. Provident Clothing and Supply Company, Ltd. (1913) A.C. 724. A similar position was taken earlier by Justice Fry in Rousilon v. Rousillon, 14 Ch. D. 351 (1880).

7Hitchcock v. Coker (1837), 6 A. & E. 438.

8 Paternalistic considerations appear to become less important over this period. In Mitchel v. Reynolds (1711), Chief Justice Parket offered, as a possible explanation for earlier cases invalidating contracts in restraint of trade enforced by a bond, the following hypothetical story:


"A weaver forced by the necessity of his circumstances to sell to his loss, another takes him, thus chagrined, to the tavern, where, for a trifle, he extorts such a bond from him, and when the cries and tears of his half-starved wife and children call him again to the loom, this bond is put in suit against him." (Modern Reports vol. 10 p. 137.)


Contrast this with:


"This case (Mumford v. Gething (1859) 7 C.B.N.S. 305) is remarkable in this respect: that it contains one of the few dicta to be found in any case subsequent to Hitchcock v. Coker to the effect that covenants in restraint of trade are held void if unreasonable, `because it is expedient for the law to interpose for the protection of the ignorant or of those who would otherwise be subjected to undue influence or pressure.' Byles, J., is responsible for this dictum .. which is, with submission, inconsistent with the principle established by Hitchcock v. Coker. ... In all of (the more recent cases) it will be found that the reasonableness of a covenant in restraint of trade is regarded solely with reference to the protection which it affords to the covenantee. The notion of protecting the covenantor against his own folly in entering into an improvident bargain has long since been laid aside. " Matthews and Adler pp. 76-77.


On the other hand, in American law, "in the application of the reasonableness test, almost from the beginning more emphasis was placed on protecting the employee from overly heavy burdens and less on the conclusiveness of contractual terms." Blake pp. 643-644.

9In the words of an early twentieth century treatise: "The doctrine that within the limits which the protection of the covenantee may reasonably require a covenant in restraint of trade is valid appears to be subject to the somewhat vague limitation that the covenant ... must not be of a nature injurious to the interests of the public-a saving clause the precise limit of which no one has ever yet attempted to define." Matthews and Adler, pp. 131-132. And in Att.-Gen. of the Commonwealth of Australia v. Adelaide Steamship Co., (1913) A.C. 781 , at p. 795, the Judicial Committee stated that "their Lordships are not aware of any case in which a restraint though reasonable in the interests of the parties has been held unenforceable because it involved some injury to the public."

10"It is a sufficient justification, and, indeed, it is the only justification, if the restriction is reasonable-reasonable, that is, in reference to the interests of the parties concerned and reasonable in reference to the interests of the public, so framed and so guarded as to afford adequate protection to the party in whose favour it is imposed, while at the same time it is in no way injurious to the public. " Lord Macnaghten in Nordenfelt v. Maxim-Nordenfelt Guns and Ammunition Co.


For more detailed analysis of the relevant law, see Blake. An important case is Horner v. Graves, 7 Bing. 735, 131 Eng. Rep. 284 (C.P. 1831), which held that a restraint preventing a dentist's assistant from practicing dentistry within 100 miles of his employer's town while the latter was practising was invalid because unreasonably broad.

[11 ]But, according to two modern authorities, "Today ... the recognized method of decision is that of balancing the employer's claims to protection against the burden on the employee. Once the judgement is made, almost never does a court proceed to consider possible injury to society as a separate matter." (Blake, pp. 686-7) and "In general, the courts have continued to assert, as in the laissez-faire era, that a covenant that is reasonable as between the parties is likely also to be reasonable in the public interest ... . However, there have been several cases of restrictive covenants ancillary to medical and legal partnership or employment agreements in Canada and the United States in which the courts have been willing to ... consider the protection of the rights or interest of the clients or patients of the parties to these restrictive covenants in continued access to the services of parties restrained thereby." (Trebilcock, p. 51) And according to Lord Atkinson in Herbert Morris Ltd. v. Saxelby, the onus of proving that a restraint is reasonable rests on the promisee but the onus of proving that it is against the interest of the public is on the promisor, which seems to imply a presumption that it is not. (Discussed in Trebilcock, p. 71).


On the other hand, in Esso Petroleum Co., Ltd. v. Harper's Garage Ltd., (1968) A.C. 269 "three of the Law Lords deprecated this dismemberment of the principle of public policy. The doctrine of restraint of trade is based on public policy, and in every case "there is one broad question: Is it in the interests of the community that this restraint, should, as between the parties, be held to be reasonable and enforceable?" (Cheshire and Fifoot p. 345).

12Restatement, Contracts, para 515 (1932). A restraint is reasonable only if (1) no greater than required for protection of employer and (2) no undue hardship on employee and (3) not injurious to the public (p. 649 Blake). Injury is "tendency or purpose to create a monopoly, to control prices, or to limit production." See Annot., 43 A.L.R.2d 94, 144 (1955).

13For a discussion of the special case of key employees or employees providing unique services, see Kniffin (1978), pp. 25-56. The author concludes that "... since 1963 a significant number of New York courts and of federal courts applying New York law have employed a uniqueness test in determining whether to enforce a noncompetition covenant. ... Courts in Maryland, Ohio, and Texas have also adopted the uniqueness criterion." Purchasing Associates, Inc. v. Weitz 13 N.Y.2d 267, 196 N.E. 2d 245, 246 N.Y.S. 2d 600 (1963) contains dictum which has frequently been cited as the foundation of the uniqueness test. Possible justifications for such a rule are suggested in Parts III and V below.

14 The modern doctrine is summed up in Valiulis, pp. 2-4.

15A non-competition agreement cannot exist solely in order to restrict competition; in order to be valid it must be "ancillary to the main purpose of a lawful contract and necessary to protect the covenantee in the enjoyment of the legitimate fruits of the contract, or to protect him from the dangers of an unjust use of those fruits by another party." Justice Taft in United States v. Addyston Pipe & Steel Co., 85 F. 271 (6th Cir. 1898l), modified, 175 U.S. 211 (1899). For a discussion of the equivalent doctrine in modern English law see Cheshire and Fifoot pp. 347-351. For a good modern discussion of the difference between naked and ancillary restraints on trade, see Judge Easterbrook in Polk Bros., Inc ., v. Forest City Enterprises, Inc., U.S. Court of Appeals for the Seventh Circuit, 776 F.2d 185 (1985). The issue of what interests the covenantee might have proprietary rights in is discussed at some length in Trebilcock pp. 52 , 66-68. fn 53 below contains a brief list of "valid interests."

16This distinction was made by the House of Lords in the important English case of Mason v. Provident Clothing and Supply Company, Ltd. (1913) A.C. 724. That case is also important for confirming Lord Macnaghten's view in the Nordenfelt Case that all covenants in restraint of trade, whether partial or general, are prima facie void, with enforceability depending on whether they are reasonable. An earlier case making the distinction between restraints in the context of employment contracts and in the context of the sale of a business is Leather Cloth Co. v. Lorsont (1869), L.R. 9 Eq. 345.


The most common justification given for the distinction between the two contexts is the difference in relative bargaining power of the parties. The buyer and seller of a business are, it is argued, in roughly equal positions, while employer and employee are not. I shall argue that, from an economic standpoint, this is not a useful distinction.

17I am ignoring the question of whether the employee correctly perceives the cost of the agreement to him. If not, there is a paternalistic argument for substituting the judgement of the court for that of the employee, a point that will be discussed later.

18 Blake might be describing a situation where something has recently increased the cost to employees of such restrictions, such as an increase in the degree of specialization of human capital, and employers have not yet recognized the change and adjusted salary offers accordingly. Employees are being offered too little compensation for the restriction, so they either refuse to sign or bargain for greater compensation. What is special is not that employees must be compensated for the restriction, but that the amount of the compensation is a subject of explicit bargaining.

19 The term was apparently invented by Salleilles in France in 1901. The earliest use of the term in an American court that I have found is by Judge Clark in 1947; references become more common from about 1955 on.


Rakoff (1983) contains a discussion of the history of the doctrine and citations to earlier articles, along with a sophisticated analysis of the reasons why such contracts exists and an interesting argument for why the terms of a contract of adhesion ought to be viewed as presumptively invalid. The essential error in his argument, in my view, is the same as in Trebilcock's defense of the unwillingness of courts to enforce non-competition agreements as written (see fn 24 below). In any case, his analysis has limited relevance to non-competition agreements in employment, since he is considering the sort of form contract that is routinely signed without reading.

20 See Shute v. Carnival Cruise Lines, 111 S.Ct. 39, 112 L.Ed.2d 16 (1990), a case recently decided by the Supreme Court, for evidence that courts are becoming more willing to accept the idea that compensation for unfavorable contract terms is included in the associated price. The point is made explicitly by Judge Posner in Northwestern National Ins. Co. v. Donovan, 916 F.2d 372 (7th Cir. 1990), p. 378 : "We may assume, since the market in surety bonds is a competitive one, that the cost savings that accrue to Northwestern from contractual terms that facilitate the enforcement of one of its bonds will be passed on, in part anyway, to the purchaser of those bonds-the enterprise in which the defendants invested-in the form of a lower premium. If so, the defendants were compensated in advance for bearing the burden of which they now complain,..."

21 For a more extensive discussion see Kitch (1980). See fn 8 above for examples of both support for and opposition to paternalistic arguments in early cases.

22 Becker (1964).

23 For others it is; we call the resulting institution a school.

24This is the explanation offered by Rubin and Shedd (1981). In their view, problems with such contracts arise due to opportunistic behavior both by employees, trying to leave with human capital paid for by their employers, and by employers, trying to use non-competition agreements to expropriate human capital they have not paid for.


This explanation fails to explain the courts' unwillingness to accept the principle of freedom of contract. Under Rubin and Shedd's analysis, the same outcomes that are inefficient are also, ex ante, unprofitable. Employer and employee can both be better off by signing a non-competition agreement in the situations where it is efficient and not signing it in the situations where it is not. So if the court wants efficient outcomes it should enforce all contracts, analysing the desirability of an outcome only when the terms of a contract are ambiguous. That is not what courts do.


The same problem arises with Trebilcock's much more extensive discussion (Trebilcock, pp. 120, 139-154). He argues that in a world of full information and low transactions costs contracts would always be efficient so far as the parties are concerned, and should therefore always be enforced unless they impose costs on third parties. But in the real world various problems may prevent the parties from reaching a contract that is fully efficient under all circumstances. Courts may then step in to rewrite the contract into something closer to what the parties would have chosen if they had anticipated the situation that arose.


This argument ignores the fact that the parties know they are living in a world of imperfect information, positive transaction costs, etc. If they believe that the court can do a better job than they can of dealing with certain contingencies, they can choose to leave those parts of the contract ambiguous-providing, for example, that the employee "will not directly compete with his previous employer for a reasonable period of time." If the parties choose instead to agree that the employee "will not engage in the selling of shoes anywhere in the U.S. for five years after the termination of his employment," that is evidence that they believe the costs to them from potential problems associated with a rigid and overly broad restriction are less than the costs of leaving the court, lacking much of their specialized information and their incentive to produce an efficient contract, free to rewrite the terms. Unless one has some reason to believe that contracting parties systematically underestimate the ability of the court system, the standard argument for freedom of contract implies that that the parties should be free to decide which issues are left open for the court's judgement.


Blake, writing some twenty years before Rubin and Shedd, made the "hostage for human capital" argument but concluded that "These arguments have not proven sufficiently persuasive." (Blake, p. 652)


For a general discussion of the use of hostages to guarantee future behavior, see Williamson. For a discussion in the context of non-competition agreements, see Trebilcock pp. 129-133.

25Lumley v. Wagner 1 De G.M. & G. 604, 42 Eng. Rep. 687 (1852).

26Bradford v. New York Times Co., 501 F.2d 51 (2d Cir. 1974). The defendant had been General Manager, Vice President and a director of the New York Times Company.

27Under modern American law, the employee could always evade the obligation via bankruptcy. It is not clear whether going through bankruptcy would relieve an employee of the obligations of a non-competition agreement.

28"she has no cause of complaint if I compel her to abstain from the commission of an act which she has bound herself not to do [singing for a rival theater], and thus possibly cause her to fulfill her engagement [singing for the theater with which she had an employment contract]... though ... I disclaim doing indirectly what I cannot do directly." Lumley v. Wagner 1 De G.M. & G. 604, 42 Eng. Rep. 687 (1852). The case was decided by an Equity court, which could grant an injunction (against singing for the rival theater) but not damages for breach of contract. Enforcing the contract directly was not an option, since that would have meant requiring specific performance of a contract for personal services. A more recent discussion of the distinction between compelling specific performance and inducing performance by preventing alternatives is in Mission Independent School District v. Diserens, 188 S.W.2d 568 (Tex. 1945).

29One can imagine reasons for doing so other than human capital investments. For instance, an employee whose productivity increased sharply over his lifetime might want to even out his consumption, being paid more than he is worth in the early years in exchange for accepting less than he is worth later. Here again, absent some bonding mechanism the employee has an incentive to behave opportunistically, quitting as soon as his wage falls below his output.


A similar incentive to sign such an agreement might arise from asymmetric information. Suppose the employee knows that he much prefers working for his employer to any alternative, and therefore has no incentive to leave even if he can take with him human capital provided by the employer. In this situation, there is no need for a hostage to prevent him from from behaving opportunistically, since doing so is not in his interest. Unfortunately, the employer does not know that. By agreeing to sign a non-competition agreement, the employee demonstrates that he has no intention of leaving.

30 "And the second reason (for careful scrutiny of covenants by an employee not to compete with is employer) is that such covenants in an extreme form may practically compel a servant to continue his service with an employer on terms which may no longer be advantageous by reason of the fact that the covenant, if enforced, would, when he leaves, prevent his obtaining beneficial employment elsewhere" Lord Blanesburgh in Dewes v. Fitch, [1920] 2 Ch. 159, at pp. 185, 186.


"A Covenant that serves primarily to bar an employee from working for others or for himself in the same competitive field so as to discourage him from terminating his employment is a form of industrial peonage without redeeming virtue in the American economic order."Josten's, Inc. v. Cuquet, 383 F. Supp. 295 (E.D. Mo. 1974) at 299.


"... the courts in the modern era have fairly consistently adopted the view that employer investments in employee training are not a protectable interest." (Trebilcock, pp. 87-88)


This position contrasts with the dictum of the 19th century English court in Lumley v. Wagner, cited in footnote 28 above. On the other hand, English courts in the nineteenth century were unwilling to enforce restrictions that they considered "unreasonably broad," which suggests that they, like modern American courts, were generally hostile to the use of non-competition agreements as bonding devices.

31A long term employment contract could in principle be enforced by money damages against the breaching employee. One problem would be the difficulty of measuring the quality of service provided by the employee. The defendant in Lumley v. Wagner could get out of the employment part of her contract by remaining with her original employer-and singing badly. Alternatively, she could breach and then declare bankruptcy.


It might also be difficult for a court to measure damages from loss of services. The parties could agree in advance to a liquidated damages provision. But the breaching employee might successfully evade the contract by claiming that the provision was really a penalty clause, designed to force him to continue to work for the employer-"a form of industrial peonage without redeeming virtue in the American economic order."

32The proprietary information itself is supposed to be kept confidential in the course of a trial. Even if none of the parties to the trial reveals the secret itself, the resulting public information-that a certain firm has a trade secret of a certain sort-may well reduce the future value of the secret by making it easier for others either to steal it or independently invent it. These issues are discussed at greater length in Kitch (1980).

33I ignore , for simplicity, changes in quantity produced due to changes in cost due to possessing the secret.

34One important difference is that a patent holder can more easily license his invention, since his property right in it is more secure; this reduces but does not eliminate the cost. In both the patent and trade secret case, a further benefit may be a reduction in expenditures to protect secrets. For a general discussion of this and related issues, see Friedman, Landes, and Posner (1991).

35As suggested by the passages quoted in fn 8, this is particularly true of the American cases. British cases tend to ignore the cost to the employee. They do, however, require the minimum restriction adequate to the employer's purpose, which suggests that any further restriction imposes an unreasonable cost either on the employee or the public.

36But note that the final sentence in the passage quoted from Judge Wachtler in Part I above suggests that this use of non-competition agreements may be legitimate in special cases.

37 "Another reason is, the great abuses these voluntary restraints are liable to; as for instance, from corporations, which are perpetually labouring for exclusive advantages in trade, and to reduce it into as few hands as possible; as likewise from the masters, who are apt to give their apprentices much vexation on this account, and to use many indirect practices to procure such bonds from them, lest they should prejudice them in their custom when they come to set up for themselves." Lord Macclesfield in Mitchel v. Reynolds. His case of the master and apprentices corresponds fairly closely to the case of the employer and employees under monopolistic competition, analyzed here and in the appendix.

38At this point I am assuming away both of the alternative explanations previous discussed, in order to focus on the issue of using such contracts to reduce competition.

39For a brief explanation of monopolistic competition, see Friedman 1990, pp. 303-310. The original discussions are Chamberlin (1933) and Robinson (1933).

40This is very similar to the situation analyzed in the previous section under the rubric of "Trade Secrets." The difference is, first, that the "secrets" are of a sort unlikely to qualify for common law protection under trade secret law, and second that they have a very narrow range of application. What we are dealing with in this section is the attempt by a firm to protect information it has generated about how best to do business in a particular market niche.


As we will see in the Appendix, the tradeoff between the benefit of creating information and the benefit of disseminating it shows up here as the tradeoff between number of initial firms (fewer if non-competition agreements are not enforced, reducing the returns to the information generated by the first generation of firms) and number of final firms.

41I ignore, for purposes of the example, the fact that such an agreement would be in violation of the anti-trust laws. Readers may wish to imagine the case of firms, such as barber shops or dry cleaners, small enough so that enforcement of anti-trust laws against an informal agreement between two firms is very unlikely. This problem does not arise with the somewhat more realistic model in the Appendix.

42 The firm may sell for somewhat more than $5,000,000, since as the founder gets older he may eventually find that it is not worth continuing to run his second firm. On the other hand, if he is able to sell the second firm as well it can continue to compete with the first firm even if he does not.

43Alternatively, he may choose to run his firm past the point where he is making only $500,000 a year, so that when he sells he will be too old to start a second firm and so will be able to get $10,000,000 instead of $5,000,000 for his firm.

44"But to refuse to enforce reasonable restraints accompanying the transfer of a business would result in unnecessary hardship or loss to a craftsman ready to retire but forced to continue in trade or to sell out at a lower price because no one would risk the purchase of his business without the protection of an enforceable covenant not to compete." (Blake, p. 629, summarizing the court in Mitchell v. Reynolds, 1 P. Wms. 181, 24 Eng. Rep. 347 (Q.B. 1711). "Suppose a man engaged in trade is desirous when old age approaches of selling the goodwill of his business-why may he not bind himself to enter into the service of another and to trade no more on his own account?" Lord Abinger, C.B. in Wallis v. Day (1837), 2 M.&W. 273.

45Alternatively, he might sell out for $5,000,000 and then start a new firm. Either way, he gets less than if he were able to sign an enforceable non-competition agreement and sell out for $10,000,000.

46Both numbers should be discounted to allow for the time between his initial employment contract and his acquiring the information necessary to start his own firm.

47The second case considered in the appendix is a counterexample.

48Using a forfeitable pension as a bond, at least in the ways in which it was most often done ("bad-boy clauses"), is currently outlawed by ERISA. See "Note, ERISA's restrictions on the Use of Postemployment Anticompetitive Covenants," 45 Alb. L. Rev. 410 (1981). A leading pre-ERISA case is Bradford v. New York Times Co., 501 F. 2d 51 (2d Cir. 1974), where a non-competition clause in an employee benefit plan was enforced. Even post-ERISA, "Vested benefits under employee benefit plans can be forfeited in two situations: (1) either where the vesting provisions of the plan are in excess of the minimums required by ERISA or (2) when the benefits under the plan are payable to normal retirement age." (Valiulis, p. 180). The distinction between non-competition agreements enforced by injunction and those enforced by forfeiture of benefits is discussed by Judge Posner in Norton Sarnoff and Carl Fletcher v. American Home Products Corporation, 798 F. 2d 1075.


In the English case of Bull v. Pitney-Boweds, Ltd. (1966) 3 All E.R. 384; (1967) 1 W.L.R. 273., making a pension conditional on the ex-employee not competing with the ex-employer was ruled equivalent to a covenant in restraint of trade.

49This issue is discussed briefly by Trebilcock, who puts it in terms of the conflict between "static efficiency" (once the information is generated, it is more efficient to have it more widely used) and "dynamic efficiency" (without restrictions on use, less such information is generated). Referring to "any post-termination activities of the employee that reduce or are likely to reduce the market value of the employer's business, to an extent that is appreciably greater than the activities of other rivals in the market (where the market is workably competitive)" he asserts that "This form of competition is inefficient and welfare-reducing because while its prohibition may generate short-term allocative inefficiencies, these are likely to be outweighed by the long-term or dynamic productive efficiencies realizable by protecting returns to firm-specific investments of the employer." (Trebilcock p. 147) No evidence is offered for the assertion. The first model analyzed in the appendix shows a case in which the "static" gains of prohibiting non-competition agreements unambiguously outweigh the "dynamic" losses.

50 It is worth noting that, despite early dicta implying that contracts in restraint of trade were prima facie illegal, English courts early in this century were willing to enforce contracts in restraint of trade whose purpose was clearly to reduce competition. "Two rival dealers in imitation jewelry were carrying on business near one another in London. To avoid competition the one sold his business to the other and covenanted that he would not for the period of two years `...carry on or be engaged concerned or interested in ... the business of a vendor of or dealer in real or imitation jewellery in the county of London or .... .' ... The Court of Appeal held that the covenant was too wide in area ... but the covenant was held to be severable in both these matters ... and the decision of Neville, J., in granting an injunction was affirmed ... .Goldsall v. Goldman, [1914] 2 Ch. 603; [1915] 1 Ch. 292." Sanderson pp. 67-68. And, in the words of Scrutton, L.J. in English Hop Growers v. Dering (1928) 2 K.B. 174 and (1928) All E.R. Rep. 396, "There was nothing unreasonable in hop growers combining to secure a steady and profitable price, by eliminating competition amongst themselves, and putting the marketing in the hands of one agent, with full power to fix prices and hold up supplies ... ." "judicial benevolence in England to outright cartels continued up to the second world war, as exemplified in the "stop" (or "black") list cases, before being overtaken by legislation shortly after the war" [WW II]. (Trebilcock p. 41).


On the other hand, it is generally agreed that a valid non-competition agreement cannot exist solely in order to restrict competition; see fn 15 above.

51In Horner v. Graves reasonableness was described a matter of "whether the restraint is such only as to afford a fair protection to the interests of the party in favour of whom it is given, and not so large as to interfere with the interests of the public." The restraint was found unreasonable and thus unenforceable, in a context that appeared to endorse preventing competition as a reasonable objective while ruling the area covered by the restriction too large.

52 "Those interests that the court have protected include the following: 1. Trade secrets; 2. Other confidential information; 3. Near permanent customer relationships; 4. Goodwill; and 5. Unique or extraordinary skills." (Valiulis, p. 2).

53 "Similarly, most courts would not allow an employer to restrict its former employees from competing in an industry that is highly competitive, with numerous salespeople from many companies selling to the same customers, where no legitimate trade secrets are involved, where customer lists and prices are readily available, and where unusual training is not needed to do the job." Valiulis p. 16, citing in support Diamond Match Div. of Diamond Int'l Corp. v. Vernstein, 196 Neb. 452, 243 N.W. 2d 764 (1976). This is precisely the sort of industry where using non-competition agreements to reduce competition makes little sense from the standpoint of the firm.

54 In the first version of the model analyzed in the appendix this is not true; even if only one employee acquires the skills necessary to start his own firm, making non-competition agreements unenforceable still results in more firms in the second period. But in that model I assume that there are no substitutes for non-competition agreements, no other ways in which the employer can pay the employee not to start his own firm. In the single employee case, there is a simple substitute-once the employee has acquired the information, the employer pays him at least as much as he would make by starting his own firm, thus giving him an incentive not to do so. This does not work in the many-employee case.

55In most of the cases considered, efficiency is equivalent to making the customers better off, since firms exactly cover costs whatever the legal rules. One possible exception is the case where the key employee makes a windfall through being paid not to start his own firm. One can probably prove the result in this case as well, since his windfall is at the expense of his firm and its customers, but I have not done so.

56It appears from the discussion in Cheshire and Fifoot (pp. 347-355) that modern British courts do not make the "key employee" distinction, although they also are more willing to enforce non-competition agreements associated with the sale of a firm than those associated with employment contracts. The degree to which modern American courts accept the distinction varies considerably; see Kniffin(1978). To the extent that courts refuse to enforce the agreement against key employees, their behavior is evidence against that they are not seeking economic efficiency.

57For the key employee case, this is true only in some U.S. states.

58 Careful readers will realize that I am being somewhat imprecise about the effect on one firm of what is happening in the market areas of adjacent firms. In effect, I am assuming that a firm without non-competition agreements is in about the same situation whether or not other firms have such agreements, thus ignoring the fact that an increase in the density of the adjacent firms will cost the firm some customers at the ends of its market area. Dropping this simplifying assumption would seriously complicate the analysis.

59 This assumes that signing the agreement imposes no cost on the employee other than preventing him from profiting from his experience by starting his own competing firm-does not, for instance, keep him from deciding, for some unrelated reason, to move out of town and switch to another employer in the same industry. That assumption is appropriate here, where we are analyzing noncompetition agreements as ways of preventing competition, but not appropriate in a more general analysis, in which they may have a variety of other purposes and effects.

60 This assumes that the number of employees who acquire the information necessary to start their own firm is large enough so that if all of them start firms profit will be driven to or below zero. If that is not the case, then the efficient outcome is for all of the employees who can start such firms to do so-which is what will happen if non-competition agreements are forbidden.

61 I have not proved that this result also holds if the number of employees in a position to acquire the necessary skills is sufficiently low so that, without non-competition agreements, [[pi]] (N)>0, although I suspect that it does.

62 As in the previous model, the result does not depend on Assumption 4.


Readers familiar with the economics of price discrimination may note that these examples suggest a familiar pattern. Requiring a firm that sells in two markets to charge the same price in each results in a more efficient allocation of a given quantity of output, and is thus an economic improvement-provided that the requirement does not change the total output the firm chooses to produce. One situation in which the requirement results in a worse outcome is if the firm requires price discrimination in order to cover its costs, and will therefore not exist if price discrimination is forbidden.