Bringing the Market Inside the Firm?

Abstract

Academics, consultants, and practitioners have long advocated bringing the market inside the firm. For example, in the 1950's and 1960's economists proposed that the transferpricing problem should be solved by charging market prices for internal transactions. Similarly, in the 1980's, financial economists suggested that the capital-allocation problem should be solved by charging the external cost of capital for internal investments. And wave after wave of organizational restructuring has advocated radical decentralization, empowerment, intrapreneurship, and the like-in short, making employees feel like owners. Proponents of making transactions within firms more market-like often seem to ignore the factors that brought these transactions inside firms in the first place. But Bengt Holmstrom and Paul Milgrom (1991, 1994), Holmstrom and Jean Tirole (1991), and Holmstrom (1999) [hereafter referred to collectively as HMT] remind us that in some cases integration is efficient precisely because it eliminates market incentives. In such cases, bringing the market inside the firm would clearly be undesirable. In this paper we show that bringing the market inside the firm is often not feasible, even if it would be desirable. More precisely, if some aspects of the market transaction are noncontractible (as we define below) then it is impossible to replicate spot-market payoffs inside a firm. This result would be trivial if the firm's only instruments were court-enforceable contracts: it is impossible (by definition) for such contracts to replicate payoffs that were noncontractible in a spot market. Our contribution is to show that this result also applies to relational contracts (i.e., self-enforcing agreements in repeated games). To establish this result, we follow Sanford Grossman and Oliver Hart (1986), Hart and John Moore (1990), and Hart (1995) [hereafter referred to collectively as GHM] in using asset ownership to determine whether a particular transaction is inside a firm, and HMT in adding court-enforceable contracts to asset-ownership models. We also draw on our own work (Baker et al., 2001) to model relational contracts.' The resulting model is the first we know of that simultaneously analyzes asset ownership, courtenforceable contracts, and relational contracts.