Coase and the Myth of Fisher Body
| Peter Klein |
I vividly recall, at the inaugural meeting of the International Society for New Institutional Economics in 1997, a discussion about the best empirical strategy for that emerging discipline. Harold Demsetz stood up and said “Please, no more papers about Fisher Body and GM!” The Fisher-GM case had become the canonical example of holdup in transaction cost economics and was considered stale and even trite. Ronald Coase, who was at the podium, replied (I’m paraphrasing from memory) “Sorry, Harold, that is exactly the subject of my next paper!”
The GM-Fisher case was introduced into the transaction cost literature by Klein, Crawford, and Alchian in their 1978 paper “Vertical Integration, Appropriable Rents and the Competitive Contracting Process.” They cited the case as a classic example of vertical integration designed to mitigate holdup in the presence of asset specificity. As the story is told, Fisher refused to locate its plants near G.M. assembly plants and to change its production technology in the face of an unanticipated increase in the demand for car bodies. This led G.M. to terminate its existing ten-year supply contract with Fisher and to acquire full ownership of Fisher.
The basic facts of the account, and the interpretation of these facts, were challenged in five independently written papers, all appearing in 2000. Three of the papers, by Coase, Casadesus-Masanell and Spulber, and Freeland, are in the April 2000 Journal of Law and Economics. A fourth paper by Helper, MacDuffie, and Sabel appears in Industiral and Corporate Change and one by Miwa and Ramseyer is in the Michigan Law Review. These papers showed that nearly every detail of the canonical account is wrong.
First, as Coase reveals, the original ten-year supply contract included provisions that GM would acquire 60 percent of Fisher’s stock and that three of the five members of Fisher’s finance committee would be appointed by G.M. Moreover, in 1921 one of the Fisher brothers became a director of GM, with two other brothers joining him in 1924, one of whom became president of GM’s Cadillac division. A fourth brother was added to the board in 1926 when GM acquired the remainder of Fisher’s stock. As Coase points out, the interests of the two companies were sufficiently aligned during the period covered by the original contract that it is unlikely that Fisher would have used the contract to extract rents from GM. Also, contrary to the standard story, Fisher did in fact build eight new body plants between 1922 and 1925 that were close to GM facilities and had incentives to use the most efficient technology available. Helper, MacDuffie and Sabel suggest that vertical integration promoted collaborative learning, while Casadesus-Masanell and Spulber argue it improved the coordination of production and inventories. In short, GM did not acquire the remaining 40 percent of Fisher’s stock in response to an inappropriate alignment between transactional attributes and an existing governance structure. Rather, the long-term contract signed in 1919 was adequate for mitigating holdup in the face of asset specificity and uncertainty, and was replaced by vertical integration for secondary reasons.
Coase’s latest article — yes, at age 96 he is still publishing in top-tier journals — appears in the Summer 2006 Journal of Economics and Management Strategy. Titled “The Conduct of Economics: The Example of Fisher Body and General Motors,” the article traces the development and evolution of the canonical account, its challenge by Coase and others, and Benjamin Klein’s response to the challenge (which Coase finds shamefully inadequate). The most interesting section is the last one, in which Coase tries to explain how economists came to accept the fable. (Coase has performed this same exercise before, of course, most famously in his 1974 article on lighthouses.) For me, Occam’s Razor suggests a simple explanation: economists like illustrative examples (more so, in reality, than comprehensive econometric evidence, despite their protests to the contrary), and Klein, Crawford, and Alchian offered one that seemed intuitively plausible, was easy to remember, and was useful for pedagogical purposes. No one bothered to check the details because there seemed to be no need to do so.
Coase’s explanation is more complex. First, because he thinks the basic asset specificity-holdup-opportunism story of transaction cost economics is wrong, he thinks researchers have exaggerated the importance of this case because it supports what would otherwise be a suspect theory. I don’t agree with Coase here, because the holdup story is supported by more than mere armchair speculation, but by dozens — perhaps hundreds — of case studies, econometric investigations, and other forms of empirical analysis. (My summary of this literature is here.) The truth of the basic holdup story is not dependent on the facts of one particular — if highly celebrated — case.
Coase has a broader purpose in mind, however. He uses this episode to criticize economists for overreliance on deductive methods, for failing to investigate the “facts on the ground,” and for general sloppiness in empirical work.
If it is believed that their theory tells us how people would behave in different circumstances, it will appear unnecessary to many to make a detailed study of how they did in fact act. This leads to a very casual attitude toward checking the facts. If it is believed that certain contractual arrangements will lead to opportunistic behavior, it is not surprising that economists misinterpret the evidence and find what they expect to find.
This is not quite fair, because Coase has not subjected his own, preferred, inductive approach to equally critical scrutiny. In his own hands, Coase’s method has led to several brilliant insights. In the average economist’s hands, however, I shudder to think of the consequences. After all, it was Coase himself who said of the “old” institutional economists: “Without a theory, they had nothing to pass on except a mass of descriptive material waiting for a theory, or a fire.” Let’s not throw out sound transaction cost and incomplete-contracting theory with the GM-Fisher bathwater.