Firm Boundaries are Responsive to More than Investment Incentives


Firm Boundaries are Responsive to More than Investment Incentives

The examples above make clear that there are many alternatives to integration when one tries to solve hold-up problems. The examples also suggest that ownership may be responsive to problems other than underinvestment in specific assets. Speaking broadly, the problems relate to contractual externalities of various kinds, of which hold-ups are just one.

Resolving Agency Problems

An example of how agency issues can affect the boundaries of an organization is whether a firm employs its sales force directly, or whether it uses outside sales agents. The best-known example here involves electronic parts companies, some of which hire their own sales agents while others sell through separate supply companies (Andersen 1985, Andersen and Schmittlein 1984). Originally, Andersen (1985) appears to have expected that the observed variation in this choice would relate to the degree of asset specificity—for example, the extent to which investment by sales people with knowledge about products was specific to a particular company. Instead, measurement costs and agency concerns turned out to be central. An employee sales force is used when individual performance is difficult to measure and when non-selling activities (like giving customer support or gathering information about customers' needs) are important to the firm; otherwise, outside companies are used.

Holmström and Milgrom (1991, 1994) rationalize this pattern with a model of multi-task agency, in which sales people carry out three tasks: making current sales, cultivating long-term customer satisfaction, and gathering and relaying information on customer needs. If the latter two activities are important and if the three activities compete for the agent's time, then the marginal rewards to improved performance on each must be comparable in strength; otherwise, the ill-paid activities will be slighted. Because performance in non-selling activities is arguably hard to measure, it may be best to provide balanced, necessarily lower-powered incentives for all three activities.

Offering weak incentives to an outside sales agent can be problematic, however, because the agent may then divert all effort to selling other firms' products that come with stronger rewards for sales. With an employee, this problem can be handled with a salary and a low commission rate, because the employee's outside activities are more easily constrained and promotion and other broader incentives can be used within the firm to influence the agent's behavior.[16] This logic also explains why outside agents commonly receive higher commission rates than does an inside sales force.

A less familiar illustration of how ownership responds to agency concerns comes from multi-unit retail businesses. Some of these businesses are predominantly organized through traditional franchise arrangements, in which a manufacturer contracts with another party to sell its products in a dedicated facility, as in gasoline retailing. Others, including fast-food restaurants, hotels, and pest-control services, are organized in what is called "business concept" franchising. The franchiser provides a brand name and usually other services like advertising, formulae and recipes, managerial training, and quality control inspections, collecting a fee from the franchisee in return, but the physical assets and production are owned and managed by the franchisee. Sometimes franchisers (like McDonald's) own and operate a number of outlets themselves. Finally, other businesses are commonly organized with a single company owning all the multiple outlets and hiring the outlet managers as employees. Examples are grocery supermarkets and department stores. What accounts for such differences?

It is hard to see how the specificity of the assets—real estate, cash registers, kitchens, and inventories—differs between supermarkets and restaurants in such a way that transactions cost arguments would lead to the observed pattern. Indeed, the assets involved are often not very specific at all. Alternatively, applying the HartMoore property rights model here would involve identifying non-contractible investments that are unavailable to the other party if the franchise agreement is terminated or the store manager's employment should end. Noncontractible investments by the center in building the brand might qualify on the one hand, but in many cases it is hard to see what the investments of the operator might be. For example, a fast-food restaurant manager might invest in training the workers and building a clientele, but these investments would presumably still be effective even if the manager were replaced by another. Further, these should also be investments that vary across cases in such a way that it is more important to provide the strong incentives of ownership to the manager of the outlet in one case and to the central party in the other.[17]

An alternative approach based on the need to offer incentives for effort has been proposed by Maness (1996). This approach begins by noting that any elements of the retail outlet's financial costs that are sufficiently difficult to measure must accrue to the owner of the outlet as residual claimant, because they cannot be passed by contract to another party. Suppose then that all costs are non-contractible in this sense; that is, since the level or appropriateness of various costs cannot be well monitored from outside, such costs cannot be part of an agreed-upon contract. Then, the only possibility for payments from the owner to the other party is on the basis of revenues. Indeed, actual franchise fees are almost always based on revenues and not on costs (Maness, p. 102) and incentive pay for employee managers is also often based on sales. In such a structure, the employee-manager has no direct incentive to control costs under central ownership, while the franchiser has no incentives for cost reduction under local ownership. Because the efforts of either party might affect costs, this creates a potential inefficiency. The solution is to lodge ownership with the party to whom it is most important to give incentives for cost control. Maness then argues that cost control in a fast-food operation is more influenced by the local manager's efforts at staffing, training, controlling waste, and the like, while costs in supermarkets are most influenced by the inventory and warehousing system, which can be centrally managed. Thus, an explanation emerges for the observed patterns: Ownership is assigned to give appropriate incentives for cost control.

A more complex example involves gasoline retailing in the United States and Canada, which has been studied by Shepard (1993) and Slade (1996), respectively. They document a variety of contractual arrangements that are used in each country between the gasoline refining company and the station operator and, in the United States, significant variation in ownership of the station. While the physical assets used in gasoline retailing are quite specific to that use, a station can be switched from one brand to another with a little paint and new signs. Consequently, neither study attempted to explain the variation in contractual and ownership arrangements in terms of specific assets and hold-up.

Both studies find that the observed patterns are consistent with the arrangements being chosen to deal with problems of inducing effort and its allocation among tasks. These arrangements differ over the strengths of the incentives given to sell gasoline and other, ancillary services like repairs, car washes, and convenience store items. In turn, these ancillary services differ in the ease and accuracy of performance measurement. The observed patterns were generally consistent with their being selected to provide appropriately balanced incentives. For example, Shepard's (1993) work notes that in repair services, effort is hard to measure and, more importantly, monitoring the realized costs and revenues by the refiner may be tricky. This should make it less likely that the refinery will own the station and employ the operator, and more likely that an arrangement will be adopted where the operator is residual claimant on sales of all sorts, either owning the station outright or leasing it from the refiner on a long-term basis. This is what the data show. In Slade's (1996) data, the presence of repair did not affect the ownership of the station (essentially all the stations were refiner-owned). It did, however, favor leasing arrangements, where the operator is residual claimant on all sales, and diminished the likelihood of commission arrangements, which would offer unbalanced incentives because the operator is residual claimant on non-gasoline business but is paid only a small commission on gasoline sales. The presence of full service rather than just self-serve gasoline sales also favors moving away from the company-owned model, since it matches the returns to relationship-building with the costs, which are borne by the local operator.[18] However, adding a convenience store actually increases the likelihood of using company-owned and-operated stations in the U.S. data, which goes against this logic unless one assumes that monitoring of such sales is relatively easy.

Considering a broad variety of retailing businesses more generally, LaFontaine and Slade (1997) document that the contractual and ownership arrangements that are used are responsive to agency considerations.

Market Monitoring

Ownership also influences agency costs through changes in the incentives for monitoring and the possibilities for performance contracting. A firm that is publicly traded can take advantage of the information contained in the continuous bidding for firm shares. Stock prices may be noisy, but they have a great deal more integrity than accounting-based measures of long-term value. For this reason, stock-related payment schemes tend to be superior incentive instruments. This factor can come to play a decisive role in organizational design as local information becomes more important and firms are forced to delegate more decision authority to sub-units and lower-level employees. Such moves require stronger performance incentives and in many cases the incentives can be offered most effectively by spinning off units and exposing them to market evaluation.

This, at least, is the underlying philosophy of Thermo Electron Corporation and its related companies. Thermo is an "incubator". It finances and supports start-up companies and entrepreneurs within a modern-day variant of the conglomerate. As soon as a unit is thought to be able to stand on its own feet, it is spun off. A minority stake is offered to outside investors, with Thermo and its family of entrepreneurs (particularly the head of the new operation) retaining a substantial fraction. The principal owners of Thermo, the Hatsopolous brothers, make it very clear that getting to the spin-off stage is the final objective and a key element in their strategy to foster a true entrepreneurial spirit within the company. Besides making managerial incentives dependent on market information, spin-offs limit the amount of intervention that Thermo can undertake in the independent units. This, too, will enhance entrepreneurial incentives.[19]

While Thermo has been remarkably successful (at least until recently), few companies have emulated its strategy. One likely reason is that Thermo's approach requires real commitment not to interfere inappropriately in the management of the spun-off units, as this would undercut entrepreneurial incentives and would also destroy the integrity of the independent businesses in the eyes of outside investors. While laws protecting minority shareholders help to achieve this commitment to some extent, Thermo's founders have worked for more than a decade to establish a reputation for neither intervening excessively nor cross-subsidizing their units. Another reason may be that Thermo has enjoyed all the benefits of a booming stock market since the early 1980s; it is not clear how well Thermo's strategy would work in a flat stock market like the 1970s.

Knowledge Transfers and Common Assets

Information and knowledge are at the heart of organizational design, because they result in contractual and incentive problems that challenge both markets and firms. Indeed, information and knowledge have long been understood to be different from goods and assets commonly traded in markets. In light of this, it is surprising that the leading economic theories of firm boundaries have paid almost no attention to the role of organizational knowledge.[20] The subject certainly deserves more scrutiny.

One of the few economic theory papers to discuss knowledge and firm boundaries is Arrow (1975), who argued that information transmission between upstream and downstream firms may be facilitated by vertical integration. As we saw in the examples of Nucor and the case of Japanese subcontracting, however, this type of information transfer may actually work fairly well even without vertical integration. More significant problems are likely to emerge when a firm comes up with a better product or production technology. Sharing this knowledge with actual or potential competitors would be socially efficient and could in principle enrich both parties, but the dilemma is how to pay for the trade. Until the new ideas have been shown to work, the potential buyer is unlikely to want to pay a lot. Establishing the ideas' value, however, may require giving away most of the relevant information for free. Again, repeated interactions can help here; in fact, even competing firms engage in continuous information exchange on a much larger scale than commonly realized. An example is the extensive use of benchmarking, in which the costs of particular processes and operations are compared between firms. But when big leaps in knowledge occur, or when the nature of the knowledge transfer will involve ongoing investments or engagements, the issues become more complex. A natural option in that case is to integrate. Any claims about the value of knowledge are then backed up by the financial responsibility that comes with pairing cash flow and control rights.[21]

We think that knowledge transfers are a very common driver of mergers and acquisitions and of horizontal expansion of firms generally, particularly at times when new technologies are developing or when learning about new markets, technologies, or management systems is taking place. Given the current level of merger and acquisition activity, and the amount of horizontal rather than vertical integration, it seems likely that many industries are experiencing such a period of change. The trend toward globalization of businesses has put a special premium on the acquisition and sharing of knowledge in geographically dispersed firms.

Two organizations that we have studied in which the development and transfer of knowledge are particularly central are ABB Asea Brown Boveri, the largest electrical equipment manufacturer, and British Petroleum, the fourth-largest integrated oil company. Both firms see the opportunity to learn and to share information effectively as key to their competitive advantage, and both operate with extremely lean headquarters that are too small to play a central, direct role in transferring knowledge across units. ABB spends a huge amount of time and effort sharing technical and business information across its more than 1,300 business units around the world through a variety of mechanisms. This would hardly be possible if these businesses were not under the single ABB umbrella. Similarly, BP's 100 business units have been encouraged to share information extensively through "peer assists," which involve business units calling on people from other units to help solve operating problems. BP also has a network of different "federal groups," each of which encourages technologists and managers from units around the world to share knowledge about similar challenges that they face.

The problem with knowledge transfers can be viewed as part of the more general problem of free-riding when independent parties share a common asset. If bargaining is costly, the situation is most easily solved by making a single party responsible for the benefits as well as the costs of using the asset. Brand names are another example of common assets that typically need to be controlled by a single entity.

[16]For a further discussion of the idea that low-powered incentives are a major virtue of firm organization and can help explain firm boundaries, see Holmström (1996).

[17]See Lutz (1995) for a formal model of franchising along these lines.

[18]A hold-up story is consistent with the fact that the presence of repair services favors dealer ownership over leasing arrangements in the U.S. data: A lessee who invests in building a clientele for repair work might worry that the refining company will raise the lease payments to appropriate the returns from this investment. This argument, however, does not do much to explain the pattern in the Canadian data, where the refiners own all the stations. One might also attempt to apply this logic to the choice between company-owned and leased stations by arguing that if the company owns the station it cannot motivate the employee-manager to invest in building a clientele because it will appropriate all the returns. However, this argument is not compelling without explaining how firms in other industries succeed in motivating their employees to undertake similar investments.

[19]See Aghion and Tirole (1997) for a model along these lines. In general, the role of firm boundaries in limiting interventions by more senior managers, thereby improving subordinates' incentives in various ways, has been a basic theme in the influence cost literature (Milgrom and Roberts 1990, Meyer et al. 1992).

[20]In contrast, researchers outside economic theory have made much of the role of knowledge. See, for instance, Teece et al. (1994).

[21]Stuckey (1983), in his extraordinary study of the aluminum industry, reports that knowledge transfer was an important driver of joint ventures.