Law and the boundaries of technology-intensive firms.

• Author: Bar-Gill, Oren

The boundaries of technology-intensive firms are determined not only by economic considerations, but also by legal constraints. The law plays a dual role: First, by granting property right protection to certain types of information and withholding such protection from others, the law determines which innovations will be organized under a property-rights-based model and which will be organized by means of access control and restrictions on employee mobility. When information is protected by property rights, the optimal organization of innovation answers the question, "Who should own the innovation?" When information is not protected by property rights, this basic question becomes meaningless, and other sources of control--like access to the innovation and contractual restrictions on employee mobility--come to the fore. This brings us to the second role that law plays in drawing the boundaries of technology-intensive firms. In the absence of property rights in the innovation, covenants not to compete (CNCs) become critical in determining incentives and overall efficiency. The law imposes substantial restrictions on the scope and substance of CNCs. In some cases it is legal doctrine, rather than economic considerations, that determines the organization of innovation.

INTRODUCTION I. ORGANIZING INNOVATION WITHOUT PROPERTY RIGHTS A. Framework of Analysis B. With Property Rights: The Aghion and Tirole Model C. Contractible Allocation of Property Rights in a Noncontractible Innovation D. No Property Rights 1. Access to the Innovation 2. Restraints on Employee Mobility II. THE LAW A. Property or No Property 1. Patent Law 2. Trade Secret Law 3. Copyright Law 4. The Law of Ideas 5. Summary B. Indirect Protection of Innovation: CNCs CONCLUSION INTRODUCTION

The theory of the firm asks when an intermediate good should be produced in-house and when it should be purchased on the market from an upstream supplier. In technology-intensive industries, one of the central intermediate goods is information or innovation. The question thus becomes which stages of the inventive process should be integrated in a single firm and which should be divided among different firms and traded on the market. The theoretical investigation of the optimal boundary between firm and market cannot be carried out in a legal vacuum. Ideally, only economic considerations should affect the "make or buy" decision. In practice, however, law imposes an important constraint on the economic balance between firms and markets.

We focus on the property rights theory of the firm pioneered by Sanford Grossman, Oliver Hart, and John Moore (1) and applied to the innovation context by Philippe Aghion and Jean Tirole. (2) Aghion and Tirole show both when integration is efficient and when nonintegration is efficient. Their analysis realistically assumes that "the exact nature of the innovation is ill-defined ex ante and the two parties cannot contract for delivery of a specific innovation." (3) But Aghion and Tirole also assume that the contract can specify "the allocation of the property right on any forthcoming innovation." (4)

Implicit in Aghion and Tirole's framework is the notion that property rights in the innovation are legally recognized. Is it necessarily the case that innovation is protected by property rights? The assumption that property rights in the intermediate good are legally recognized is, in many contexts, completely innocuous. Obviously, the law recognizes property rights in the proverbial widget. But this key assumption is not innocuous in the innovation context. Legal doctrine is continuously struggling to define what classes of information are worthy of property right protection. (5)

This is not to say that most types of innovation are not legally protected. Probably most types of economically valuable innovation are protected by property rights. Accordingly, Aghion and Tirole's model clearly covers a broad class of cases. But, we argue, there is another class of cases that the Aghion and Tirole model does not cover--cases where the innovation is not protected by property rights. Aghion and Tirole derived the optimal organization of innovation, assuming legally recognized property rights in the innovation. How would innovation be organized absent such property rights? We answer this question, emphasizing the differences between the organization of innovation with and without property rights. These differences underscore the important effect of legal policy--which determines the scope of property right protection--on the organization of innovation. (6)

To study the organization of innovation in the absence of property rights, we revert back to the notion of control that underlies the property rights theory of the firm. We argue that control can exist, and be allocated, even absent property rights. We focus on two alternative sources of control: access to the innovation and contractual restrictions on employee mobility.

The first, nonproperty source of control is access to the innovation. (7) We define "nonintegration" as the case where the research unit can withhold knowledge of the innovation from the customer, and "integration" as the case where the existence and substance of the innovation are known by both the research unit and the customer. Under our definitions of access-based organization, the research unit may enjoy control via exclusive access in the nonintegration case. The real power that access provides is determined by the choices available to the research unit. Specifically, can the research unit (RU) extract surplus from the customer (C) without relinquishing control--that is, without disclosing the innovation to the customer? (8)

In many cases the answer is no. If RU cannot develop the innovation itself and must sell it to a customer, then the two parties engage in bargaining under asymmetric information--bargaining that may well result in impasse. As Kenneth Arrow famously observed, information that is not afforded legal protection cannot be bought or sold on the market. Absent legal protection, the information holder is in a bind: in order to sell the information, she must disclose it to the potential buyer, but once she does, she has nothing left to sell. (9)

Faced with the disclosure paradox, often the best that RU can do is disclose the innovation to C, free of charge, and then bargain under conditions of symmetric information. As Anton and Yao show, RU may still be able to extract a significant portion of the surplus from C even after disclosing the innovation to C by threatening to disclose the innovation to C's competitors. (10) Since the best strategy for RU is to disclose the innovation to C, the fact that RU initially enjoyed exclusive access to the innovation is meaningless, and the distinction between integration and nonintegration collapses.

There are circumstances, however, where access does imply actual control and can thus provide a basis for a theory of organizational design. First, there are cases where RU will choose to negotiate with C before disclosing the innovation despite the disclosure paradox. (11) Second, when RU can develop the innovation itself, (12) access improves RU's bargaining position and thus implies control. Third, in some cases, RU can disclose only part of the innovation to C and then bargain with C for a share of the surplus larger than what it could obtain after full disclosure. (13) Finally, in some cases RU can leverage access and force C to sign a nondisclosure agreement (NDA).

We characterize the optimal access-based organization of innovation and compare social welfare under access-based organization with welfare when property rights in the innovation are recognized and the organization of innovation can be based on the allocation of these property rights. We show that the organization of innovation, and its welfare consequences, depends on the legal policy that sets the scope of intellectual property rights.

The second, nonproperty source of control derives from contractual restrictions on employee mobility or covenants not to compete (CNCs). CNCs are easy for courts to enforce even in an environment fraught with noncontractibility (like the innovation environment). Oliver Hart observed that in technology-intensive industries, the firm's "source of value may consist of as little as ... a contract that prohibits [the firm's] workers from working for competitors." (14)

Nonhuman assets are the source of control in the property rights theory. The emphasis on nonhuman assets is motivated by "the absence or slavery," (15) which implies that control over human assets is inalienable and thus cannot be allocated. The important role of CNCs, as recognized by Hart, suggests that, at least in technology-intensive sectors, nonhuman assets are not the only, and perhaps not the main, source of control. Control over human assets can be allocated. Of course, the reach of CNCs is limited. And it is the law that sets the limit, taking into consideration, among other things, the inalienability concerns raised by Hart.

CNCs differ from the standard assets that serve as a source of control in yet another way. The standard assets are discrete. A machine is either owned by A or by B. (16) CNCs, on the other hand, allocate control in a continuous manner. The strength of a CNC depends on its geographical and temporal reach--both continuous dimensions. Stronger CNCs increase the customer's control and thus enhance her incentives to invest. At the extreme, a very powerful CNC can mimic the incentives generated when the innovation is protected by property rights and these rights are allocated to C (the integration case). (17) A CNC, as a continuous contracting variable, allows the parties to optimally calibrate incentives, at least within certain bounds. (18)

As suggested above, one such bound is defined by the law. The law imposes an upper bound on the strength of enforceable CNCs. When the...