Longer ago than either of us cares to remember, one of us attended junior high in Tokyo. On Saturdays, he worked at a printed circuit factory. Or maybe "factory" makes it all sound too grand. A small building in back of a gas station, it had three or four punch presses. The "president" supervised matters (though he actually spent more time hanging out at the gas station), together with a sidekick who did assorted odd jobs besides. Several middle-aged women with no apparent technical education or skill ran the presses.
The junior high kid spent his time trimming the sheets to which others would eventually attach the transistors. The women then punched the holes and margins onto the boards, and the president's sidekick loaded the finished boards onto a truck. Periodically, he returned them to the firm that had ordered the work and brought more sheets to punch along with any press dies the firm needed. The punch presses were standard generic affairs, and the buyer seems to have kept title to the dies.
Thirty years later, the other one of us knows the president of a factory near Nagoya. For many years, the firm has done machining work for a first-tier Toyota subcontractor. Unfortunately for the firm, Toyota has increasingly substituted integrated plastic units for the steel shock absorber parts the firm machines. Worried that the Toyota-bound work might disappear, the president has begun to move the firm toward machining materials for computer hard disks on the side.
A machining firm can make a wide variety of products, the president seemed to explain. His firm could make products for the automobile industry or otherwise, Toyota-bound or otherwise. If the demand for shock-absorber parts fell, well then it would simply make computer disks instead.
What neither of us saw in either firm was any evidence of investments that were specific to the firm's trading partners. Yet whether such relationship-specific investments ("RSIs") structure the arrangements firms make matters. Indeed, for at least two independent reasons, whether they structure the Japanese automobile industry matters crucially.
Within law and economics, the prevalence of RSIs matters because of the way the issue goes to the heart of market contracting. At root, RSI theory challenges our routine assumption that straightforward market contracting produces something close to socially optimal arrangements. Although the theory is clear, the empirics are less so. Scholars have looked hard for evidence of governance arrangements driven by large relationship-specific investments. To date, they have reached only mixed conclusions. They find substantial evidence of the relation between RSIs and governance within idiosyncratic industries like public utilities, aerospace, and defense. Although they find some evidence of the relation within "ordinary" industries, they find considerably less. In that empirical vacuum, the Japanese automobile industry has stood as a prominent exception -- an important example of RSI-driven extra-contractual governance arrangements in an "ordinary" industry.
Within Japanese studies, RSIs provide a convenient theoretical rationale for taking the conventional tales of "socially embedded" contracts and relational stability at face value. To date, all too many scholars have been all too happy to "explain" these tales by citing strong cultural norms of integrity or obligation. The theoretically more astute justifiably find the "explanations" hollow. For them, RSIs have offered an analytically coherent incentive-compatible rationale for exactly the same tales.
In this Article, we argue that the usual accounts of the industry are myth. Notwithstanding those accounts, the industry does not contain widespread, substantial physical-asset or human-capital RSIs. To the extent that we are right, theorists might do well to rethink the empirical role RSI theory has played over the past two decades. We do not argue that firms never make RSIs or that contracts will always solve incentive problems. Far be it from us to make such a claim, especially since this Article is only about one industry in one country. Neither do we claim that RSI theory is wrong as theory. Neither of us is a theorist, this is not a theoretical paper, and the intuition behind RSI theory has generally made sense to us anyway. Instead, we make a more modest point: that modern production may require lower levels of idiosyncratic investment than we have usually supposed; that market contracting may work better than usually asserted; and that, as a result, RSI theory may explain less of the contracting and governance patterns in place than scholars have often asserted.
In this Article, we argue that RSIs in the Japanese automobile industry are usually quite small and usually play a minor role. Toward that end, we begin by summarizing the implications RSI theory poses for contract theory (Section I.A) and surveying the empirical evidence (Section I.B-.C). We then turn to the Japanese automobile industry. First, we anecdotally canvass the practices at Honda (Section II.B), and provide a background to the industry as a whole (Section II.C). Second, we examine the evidence of RSIs among second- and third-tier suppliers (Section III.A). Finally, we examine the evidence among first-tier suppliers (Section III.B).
A. The Idea:
Relationship-specific investments matter -- and matter deeply -- argue Oliver Williamson, Benjamin Klein, Robert Crawford, and Armen Alchian.(1) Dozens of scholars have since repeated the logic they pioneered, and today it graces such mainstream sources as the industrial organization text of Dennis Carlton and Jeffrey Perloff and the management text of Paul Milgrom and John Roberts.(2) According to this intuition, the scope and size of RSIs can directly affect the governance arrangements firms choose. Whether business partners negotiate long-term contracts, spot contracts, equity investments, franchise arrangements, or even mergers can depend vitally on the RSIs at stake.
Crucially, investments specific to a relationship generate appropriable quasi rents. In a world of incomplete contracting, as Scott Masten, James Meehan, Jr., and Edward Snyder put it, that appropriability may increase the "resources expended attempting to negotiate a favorable distribution of the gains from trade."(3) In the words of Klein, Crawford, and Alchian themselves, "[a]fter a specific investment is made and such quasi rents are created, the possibility of opportunistic behavior is very real."(4) To avoid such rent-seeking and rent-avoidance costs, firms may sometimes introduce governance arrangements that are otherwise unnecessary (and probably problematic, given the way most of them weaken market incentives). RSIs can potentially transform a competitive market exchange into a bilateral monopoly, in other words. When appropriate contractual arrangements are infeasible, that transformation may call forth arrangements that otherwise would be superfluous at best.(5) Or as Klein, Crawford, and Alchian write:
The crucial assumption underlying the analysis of this Article is that, as assets become more specific and more appropriable quasi rents are created (and therefore the possible gains from opportunistic behavior increases [sic]), the costs of contracting will generally increase more than the costs of vertical integration. Hence, ceteris paribus, we are more likely to observe vertical integration.(6) B. The Evidence
Consider a short summary of the anecdote Klein, Crawford, and Alchian used to popularize this analysis: the 1926 merger between General Motors and Fisher Body. Before 1919, claim Klein, Crawford, and Alchian, car companies used wooden or wood-and-metal coaches. Making these early coaches involved standard tools and standard knowledge. Making a good one took skill, but it was a skill a coachmaker could use as easily to fit a coach onto a frame by assembler A as onto one by assembler B. Conversely, assembler A could as easily use a coach from coachmaker X as from coachmaker Y.
In this pre-1919 world, continue Klein, Crawford, and Alchian, assemblers and coachmakers traded on what was virtually a spot market. In doing so, they took little risk. If a coachmaker stopped selling, the assembler could buy its coaches elsewhere. If an assembler stopped buying, the coachmaker could sell its coaches elsewhere. As neither had invested much in either assets or skills that were specific to the relationship, neither had much to lose from switching contract partners.
By the next decade, Klein, Crawford, and Alchian write, car makers started to make standardized coaches out of steel. Fashioning these steel coaches required dies. In turn, these dies cost large sums, and could be used only for specific models. Now, the assembler and coachmaker faced the prospect of investing in an asset that paid off only within the relationship. As such, the asset generated appropriable quasi rents: if the coachmaker bought the die, the assembler could threaten to end the relationship in order to shift the terms of the deal in its favor.(8)
Rather than risk this opportunism, reason Klein, Crawford, and Alchian, assemblers and coachmakers integrated vertically. In 1919, Fisher Body and General Motors entered into a long-term contract. Alas, given the problems inherent in long-term contracts in the real world, opportunism-related problems persisted. By 1926, GM simply acquired Fisher Body outright. Given the large RSIs involved, the two firms found it paid to eliminate the risk through vertical integration.
To Klein, Crawford, and Alchian, the risk of opportunism in the GM-Fisher Body relationship lay in the investment in large stamp dies: "The manufacture of dies for stamping parts in accordance with the above specifications gives a value to these dies specialized to [the assembler], which implies an appropriable quasi rent in those...