Ungoverned Production:
An American View of the Novel Universalism of Japanese Production Methods and Their Awkward Fit with Current Forms of Corporate Governance
Charles F. Sabel
Columbia Law School
Paper prepared for presentation at the
Conference on Socio-Economic Systems of the Twenty-first Century,
Sponsored by the
Institute of Fiscal and Monetary Policy
Ministry of Finance
Japan
Tokyo
February 16, 1996
I thank my colleagues in the Sloan Governance Project and the Corporate Law group at
Columbia Law School, and particularly Mark Roe, for background discussions that
contributed greatly to the view presented here. The usual disclaimers apply with unusual
force, however, as this paper was drafted in trans-Pacific flight, at an altitude
inaccessible to even their wise counsel.
1. This conference, pursuing questions prompted by the study from which it grows, seeks to
weigh the merits of two contrasting claims. This first is that Japan is arriving by an
original route at a universal form of economic and social modernity best represented, in
intellectual spirit if not in fact, by the United States, and called the neoclassical
synthesis to underscore its reliance on coordination through market exchange or
shareholder capitalism, and to underscore the related idea that the corporation exists for
the benefit of its equity owners. The alternative proposition is that Japan has developed
a distinctive form of economic and social organization, rooted in its own culture and that
of the East Asian area of which it is a part. This Japanese form of capitalism is
sometimes referred to as stakeholder capitalism to emphasize its view of the corporation
as a community (like the larger society of which it is a part) jointly controlled and in
that sense "owned" by all those -- suppliers of labor as well as of capital or
components -- who collaborate in production. This shareholder capitalism is said to be as
modern in its way as, no more peculiar in its particularity than, and at least as
competitive economically as the variants of Anglo-American and continental European
capitalism that are commonly offered by partisans of the first thesis as so many examples
of the neoclassical synthesis.
This may in the end be the right question, but it is not the one that would occur first to an American familiar with the organization of production (narrowly understood as the procedures by which goods and services are designed and produced) and governance (the way, and to whose benefit, corporations are monitored) in the US. The reason goes to the pattern of change in the US economy in the last decade. The organization of production in the US has become in that time substantially more collaborative or team-like: in a word, more "Japanese," and, for reasons touched on below, less "neoclassical" or "American." But during this same period there has been surprisingly little change in the basic pattern of US corporate governance. Indeed, if it has changed at all, the US governance system has become more "American" in the sense of treating the corporation as an instrument to be used exclusively for the benefit of its equity owners. Assuming, to avoid obvious circularity, that the US governance system will not automatically become "Japanese" just because the production system is, and, conversely, that the US version of Japanese production methods will not fail simply because the governance system remains American, the reasonable US response to the Japanese question of whether our systems are alike or different is that we don't have a system in the way the question intends.
Let me note, by the way, that I am certainly not the first or only American to remark on this divergence. Michael Porter, commenting on Fukao's recent comparison of governance regimes, on which the present report draws, observes that the Japanese system is portrayed as "supportive" of lean production, while the American system is said to be "not supportive." This, Porter continues, is "clearly too simple," because "many US companies have adopted lean production and closer partnerships with suppliers in recent years." [Porter in Fukao, p. 93]. The comment is ungenerous because Porter forgets to add that he, too, has portrayed the US financial and governance systems as hostile to such cooperation; but it does catch the essence of the matter, and it is, perhaps, especially convincing coming from a past partisan of another view.
Informed observers in societies as different as, say, Brazil and Germany could, I think, reasonably respond in this way as well. For in these countries too production is becoming more Japanese, but governance is not. To my knowledge there is no systematic change of governance in Brazil; and in Germany the well-documented central development is the withdrawal of banks from corporate monitoring [Fischer and Edwards]: a move towards "Americanization." Nor is my American perplexity simply the result of an arbitrarily narrow comparison of the production and governance elements of the respective national bundles. While production was becoming more Japanese in the US, for instance, industrial relations were becoming so much more "American" in their disregard for unions and company or internal labor markets (and, conversely, their increasingly strict dependence on the operation of open or external ones) that many US industrial-relations experts have trouble recognizing what country they are now in. Indeed, considering all this, it may even be that my perplexity is not especially American. Is it completely misguided to think, for instance, that some Japanese observers, looking at the confusion of main-bank governance (and its drift, perhaps, towards "Americanization"), but noting the stability of many other, crucial features of the Japanese production system, may wonder too whether it is judicious to class their own country as either Japanese or neoclassical?
Against this backdrop I want to add to the two possibilities we have been invited to consider a third general view, and show how it bears on problems of corporate governance. This third view is that Japan has pioneered a distinct form of decentralized production organization that is universal, in the strict sense that its core features facilitate adoption in the most diverse settings, regardless of cultural preconditions. But, the general argument continues, the new form of organization, derived from Japanese experience and now developing independently of it, is not neoclassical: For this new decentralization depends on the continuous, disciplined exchange of information among all those collaborating in production, not the contractual relations among autonomous agents, each presumed to know by itself what needs to be done to meet its obligations to the others, which is, of course, the neoclassical view. Finally, the very same features of the Japanese-derived system that allow it to diffuse independent of cultural preconditions and that distinguish it from the neoclassical synthesis render ineffective the current governance structures -- all of them, including the Japanese main-bank system -- which permitted the emergence of the novel type of decentralization in the first place.
Thus, where the conference program asks us to consider the meaning of governance patterns (taken to be "ideas of the corporation" or particular institutions) as parts of coherent bundles including production, political, educational, and cultural systems as well, this third view suggests that the bundles are not presently coherent at all, and that the parts, particularly the production-organization piece, may even be in tension with each other. If that is so, we ought to be thinking hard about how to harmonize the pieces -- perhaps in many, nationally or culturally specific ways -- and not trying to determine whether the bundles are the same or different, and if the latter, which differences are best.
A brief presentation is of course not the place to attempt a full examination of the theoretical underpinnings and practical implications of this third view. Instead, I want to focus on the puzzle of how Japanese production methods can be diffusing to the US and elsewhere even though the other institutions of the Japanese bundle -- especially main-bank or keiretsu governance-- are not. Then I want to suggest how an answer to that question throws light on new forms of governance that are emerging in the US, and why these latter may have relevance to unsolved problems elsewhere. I begin by evoking the change in US production organization that is at the center of what makes the puzzle puzzling.
2. The change I have in mind is the diffusion in the United States of Japanese customer-supplier relations, often taken to be the core of the Japanese production system as a whole. In these new supply relations a small number of top-tier contractors assume responsibility for co-developing crucial modules or subsystems with the final producer, coordinating the production of low-tier supplier producing parts for subassemblies for their module, delivering the components just-in-time, and meeting targets for incremental improvement of production according to targets agreed with the customer. Benchmarking, simultaneous and value engineering, as well as target pricing are the familiar disciplines by which this collaboration is coordinated.
The exact extent of this diffusion is hard to measure; but the combination of anecdote and survey results is compelling. In the auto industry, to take the most familiar example, there are well documented stories of spectacularly successful implementation of the collaborative system, of which the development of the Chrysler LH cars is one of the best known. It is an open secret of the automotive trade press that Chrysler, in general, is better at the new methods than General Motors, and Ford in between. Chrysler is the weakest of the big three American auto manufacturers; and its success therefore casts substantial doubt on the thesis still sometimes heard in the US -- and often advanced five years ago to explain Japanese achievements -- that the true "innovation" of the new methods is nothing more than the use of market power by the strong to shift the costs of adjustment to the weak.
Academic surveys of relations in the industry show accordingly that the length of contracts between customers and suppliers is increasing (again with company-specific differences [Helper] -- so much so, in fact, that suppliers report that they would undertake no further investment in equipment for particular projects if the contracts were further lengthened. Similarly, the proportion of parts which the customer regards as "black boxes" because the supplier's know-how was indispensable to its realization has increased rapidly in the US and is reaching Japanese levels [Fujimoto].
There are, to be sure, still differences between the two countries in this regard. Japanese customer firms are apparently better at conceptualizing the phases of the design process so that, for example, the periods in which innovations in component design with repercussions on adjacent components or on the specification of the product as a whole are clearly separated from periods in which suppliers must devote themselves to improving their products within established parameters; and the clarity of the distinction avoids the disruption that results from blurring this distinction in the US. But this plainly a difference of degree, not of kind [Liker]. As of the mid-1980s, moreover, the same difference existed in Japan between Toyota, which was, of course, the more "Japanese" in the precision of its indications, and Nissan, which was the more "American." Once Nissan noticed the difference -- long hidden because its suppliers did not traditionally work for the Toyota group and vice versa -- it switched to the "Japanese" system. US firms, if they come to the same conclusion, can presumably do the same [Fujimoto].
There are many other examples of the spread of aspects of the Japanese production system; but this one, in addition to being a precondition of the rest, is important because it sheds light on our puzzle: the disjuncture or divergence between production and governance systems, and more generally the strange incoherence of the bundles of features that are supposed to form types of capitalism. The new (to us in America) supplier relations, after all, seem to suppose not only long-term relations between customer and supplier -- at least as long as it takes to design and build a whole model generation -- but also and above all assurances, for both sides, that the relations will in fact be long term. Unless you the customer and I the supplier have some sort of commitment to each other, it seems, you will not make yourself vulnerable to me by making my firm the sole source of a crucial subsystem, and I will not make myself vulnerable to you by dedicating resources to a project that will be worthless if you cancel the order. (This is just a variation of the well known argument that it is necessary to offer the workforce life-time employment, or some equivalent, to secure participation in continuous improvement activities that lead eventually, through cumulative reorganization, to the displacement of every worker from his or her current job.) Whatever the precise form of those assurances, moreover, they all have as their precondition the justifiable expectation on both sides that the relation between us can be stable if we want it to be. If I think someone -- a shareholder, perhaps -- can tell you to cancel your order with me simply because of disappointment with your performance, not mine, even if I want to do business with you I won't, and your response to my situation will be the same. So, if the bundle view is correct, whatever else we may imagine has or has not happened in the organization of corporate governance in the US we should be safe in assuming that nothing has happened in governance (or any other institutions) during the period in which the new supplier relations were being introduced, to increase the likelihood of such paralyzing disruptions. If anything, on this view, the governance should have become more hospitable to long-term cooperation, for example, by favoring finance through "patient" capital which is indifferent to short-term turbulence because of its dedication to long-term results.
But as I have already suggested repeatedly, the change, if any, has been in the opposite direction: to render firms more answerable to shareholders with a keen eye for short-term results, and thus less reliable as long-term partners. I say "if any" because it is not clear, on inspection, that much has changed at the level of institutional design, despite familiar talk of the shareholder revolution and the like. But recall that a finding of no change would leave us just as puzzled as before, since the old -- pre revolutionary -- institutions of corporate governance already were supposed to give impatient capital so much control of affairs that Japanese production relations were excluded. Let me say briefly, in any case, what I think has happened as a result of all the churning of US governance institutions, and use the conclusions from this quick review to reformulate the puzzle.
3. Take first what has changed. As a result of shareholder suits and other legal developments, boards of directors certainly take more seriously today their formal obligations as representatives of the corporation's equity owners than they did a decade ago. This is particularly true in the case of mergers and acquisitions, where recent decisions obligate the board to seek the best price for the company they monitor if it has been put up for sale. But it is also generally agreed that boards play a larger role than before in the selection, or at least the review of the selection, of the CEO, in scrutinizing corporate strategy, and in establishing goals and procedures for selecting members of and assessing the performance of the board itself. It is also true that the board of directors is widely perceived as having greatly increased its power, as demonstrated in the spectacular dethronement of the CEO with the help of the board and institutional shareholders at such corporations as General Motors, IBM, Kodak, AMEX, Westinghouse, and Apple.
But beneath this surface, most everything is as before, where before refers to situations where boards of directors could hardly be thought of as loyal representatives of vigilant and demanding shareholders. Recall that from the 1930s through the late 1970s US boards were widely regarded as abject creatures of the managers who appointed and remunerated them. From the circumstance that the boards are not the abject creatures of management does not necessarily follow that they are its equals, and still less that they exercise mastery over it in the name of the shareholders. The most current evidence suggests strongly, in fact, that they are not and do not. Thus a recent representative survey of some 100 directors of major US corporations commissioned by the Institutional Investor Project of Columbia University (which has close ties to the institutional shareholder activists) found that the "vast majority" of those interviewed dismissed the idea of formally separating the office of CEO from the office of chairman of the board of directors, and reserving the latter for an outsider, who would presumably be a better representative of shareholders than the inside CEO [Neiva, p. 51]. The study found further that the directors were reluctant even to (and infrequently did) meet with representatives of institutional investors, and that this reluctance was "tame relative to their aversion to inviting investors to serve on corporate boards." [ibid., p.63]. The ideal director for the directors interviewed -- the same kind of people who presumably say that the corporation should be managed in the interest of the shareholders -- is, correspondingly, a person like themselves: the CEO (or, increasingly, the division president) of a large corporation [ibid., p. 24]. A recent econometric study of the relation between board composition and corporate performance reinforces the conclusion that the inside managers are still very much in charge. This study finds that many boards now have investment, strategic development and finance committees whose purpose is to evaluate long-term investment and finance decisions. But, crucially, membership on these committees is disproportionately left to inside or management directors, presumably the same people who formulated the plans in the first place, and the higher the percentage of insiders in these bodies the (marginally) better the corporate performance [Klein]. So a crucial result of all the academic and institutional agitation for increased participation of outsiders friendly to shareholders on corporate boards is reorganization that puts inside managers in charge of reviewing their own work, and, for good measure, a study that shows, theory aside, that this is not obviously a bad thing.
I leave it to others to sort out the complex relations between public perception and reality, and the even greater gap, perhaps, between the latter and certain academic reform projects. For present purposes the relevant conclusions are the ones anticipated above: At a minimum, changes in governance have not been a cause of the introduction of the new production methods in the sense of providing assurances of stability that were lacking before. It is simply not clear whether anything of relevance to production organization in this sense has changed. If the governance churning has had any effect on reorganization, it should have to been to obstruct "Japanization," for whatever change has taken place is so ambiguous that it has made an uncertain situation more uncertain still, and so less conducive to long-term relations. Because, however, confused and complex as the governance structures were or are, they plainly have not blocked the introduction of the Japanese supplier relations, the plausible upshot of all this is that formal governance institutions in the US (and by extension elsewhere) have neither helped nor hindered the reorganization of production very much, at least not in any way we can grasp with the help of current theories of the corporation.
(You may object, parenthetically, that the introduction of new methods would have been faster had the governance institutions been more hospitable. But recall that by (almost) all the measures of your report, Germany and (by some) France have more patient capital and more hospitable governance structures than the US. Yet they lag behind the Americans, substantially, in the adoption of the new methods. Of course you might respond that "Japanese" type governance will produce Japanese production relations only in the presence of other features of the Japanese bundle. But then we are back where we started, for how do we explain the diffusion of those methods in the US, where not much seems Japanese?)
All this leads to a reformation of our original puzzle: What is it about the Japanese production system that allows its diffusion given some minimum de facto level of stability, but in the absence of the kind of governance assurances of stability that seem to be its precondition? Or, put another, more tentative way, could it that the construction of Japanese production systems does not suppose the existence of long-term relations, because the system somehow in the course of its operation is able to produce them?
This is, I think, just the innovative and potentially universalizing property of the Japanese system, and it is achieved through the disciplines and institutions the collaborators use to assess continuously one another's capacities while they are solving problems jointly. These disciplines link discussion of actual performance by the cooperating parties -- monitoring -- to discussion of how to improve operations given that performance -- learning; hence I refer to the whole process as learning by monitoring. Just-in-time production, where parts are produced one at a time as needed, is an extreme example: A defect introduced at one work station literally stops the flow of production, so discussion of improving the production setup by identifying and eliminating the disruption becomes a precondition for continuing production at all. Strictly analogous disciplines such as value-added engineering and target pricing allow suppliers and customers to set goals and metrics for assessing progress in achieving them in relation first to design, then to production projects. Think of this as an information-symmetricizing machine, which works to ensure that the parties come to share the same understanding of their situation, and to redefine their purposes and interests accordingly, in the very process of exploring the ambiguities they encounter in pursuit of their initial goals. With some additional argument that would be a distraction here I might convince you that this mutual exploration of unknown states of the world has little to do with the neoclassical view of coordination of independent actors through formal contracts whose very purpose is to fix ex-ante the conditions of exchange, with at most a few exceptions, in the case of relational contracts, for foreseeable "unforeseeable" contingencies such as fluctuations in the price of raw materials or labor.
Notice that in this setting partners do not need equivalent levels of sophistication, nor do they need to depend on a history of prior -- social? economic? -- relations as an assurance of mutual liability. A subcontractor with rudimentary production skills, for instance, can start as a low-tier supplier, and move up the hierarchy of suppliers as it is able to demonstrate greater capacity and reliability. Or both parties can increase in sophistication together, the customer delegating more responsibility to the supplier, to conserve resources for more and more complex design and coordination tasks, and the supplier becoming a more and more equal partner, providing not just black boxes but whole new technologies. If either partner is unable to achieve the stability and discipline needed to maintain the flow of information required to work with the other, this will be evident in disruptions in the information exchange early enough for the more capable collaborator to help correct the problem or seek an alternate partner.
Another, more general way to put the same point is to say that skills (of individuals) and capacities (of firms) in the Japanese system are general purpose, not firm-specific or dedicated in the sense of suited to transactions only with particular, known partners. Indeed, diversification among many clients or employers should actually result in efficient gains: If problem-solving or co-development is the master skill or capacity, then the more different problems solved, the easier it becomes to solve the incremental problem at the margin of experience. Diversification should, in other words, yield economies of scope. The risk of this open, diversified strategy is clearly the possibility of encountering an unreliable partner, and investing so much in the partnership that failure is ruinous when it is finally detected. Learning by monitoring reduces this risk. Given the general-purpose character of the activities and the associated economies of scope, learning by monitoring explains how the world over, in giant, state-owned Chinese steel mills, in Brazilian automobile firms and in the US, parties with no tradition of collaboration, or rather, long histories of mistrust or negligent indifference to one another, are successfully adopting "Japanese" methods.
None of this will strike you as very Japanese. Where is trust? Where is the sense of mutual obligation said to be characteristic of Japan? Surely this is just another effort, fundamentally like the Western economics from which it dissents, to re-describe Japanese experience in a way that simulates the outcome while stripping the system itself of its inherent individuality?
By way of response consider three observations. The first is that the more American observers learn about the Japanese system, the more struck they are, too, by how very un-Japanese it appears. The idea that even first-tier subcontractors compete with each other in the design stage of major projects, each sending its own engineers to work with the potential customer; the notion that rates of improvement are carefully negotiated and recorded in agreements -- this and much more of the measurement and constant comparison of performance with goals does not comport with the picture of the Japanese as relying on mutual dedication to common tasks to carry the burden of intimate relations [Liker]. Second, the more Japanese scholars focus on the Japanese system, the more they, too, come to doubt cultural explanations. Wada, for instance, in an excellent study of the history of supplier relations at Toyota, criticizes two Americans, Monteverde and Teece, for arguing that there was less vertical integration in Japan than in the US because "close cooperative relationships between assemblers and suppliers" reduced the danger of opportunism in Japan. From his study of Toyota, "reputed to be the assembler with the closest cooperative relationships with its suppliers," Wada draws the contrary conclusion that "these close cooperative relationships were realized under a system of evaluations of suppliers by Toyota, which stimulated a competitive spirit among suppliers. It is not that Toyota was not liable to opportunistic exploitation, but that close cooperative relationships in themselves contain the means for preventing the occurrence of opportunism. The evaluation system brought into the close cooperative relationships is the important factor that raised the percentage of Toyota's reliance on external production and that brought about the tiered inter-firm relationships." [Wada, p. 47]
Koike, moreover, makes a second and related argument against the culturalist view in stressing the role in promotion decisions of evaluations of the ability to respond to unusual circumstances and encourage problem-solving by groups [Koike, Southeast Asia, Industrial Relations]. His position is, however significantly different from the one urged here, for instance, in stressing the role of tacit knowledge in production where I am emphasizing the importance of explicating tacit knowledge when necessary through learning by monitoring. (But see on this point [Nonaka]) I introduce his view, and the others from the Japanese debate as well, not to suggest, improbably, that my position turns out to be the "Japanese," one, but rather to indicate only that there are authoritative voices in the Japanese debate -- at least the part accessible to a non-specialist -- which can be construed in favor of the notion that Japanese innovations are of potentially general significance.
4. Where does all this leave the problem of governance? More precisely, what is the relation between the kind of monitoring that goes on in Japanese production systems in connection with project selection, design, and the like, and the kind of higher level monitoring -- of the viability of whole lines of business or divisions, or how to choose among very different but plausible long-term development goals, of how to respond to threats or opportunities facing the corporation as a whole -- that is rightly the province of corporate governance. From the point of view of the report under discussion, the answer is that the relation already is, or will soon be, harmonious: Either "Japanese" style main-bank monitoring will, after a period of turbulence, return to its "natural" role of monitoring "Japanese," corporations, or, after (substantially?) more turbulence a neoclassical synthesis, more American in its outlook, will solve the problem. But the empirical remarks and theoretical thrust of the argument so far suggest a different, much less harmonious conclusion.
Empirically, neither the main-bank nor the US shareholder system has given evidence of doing a good job at monitoring the new kinds of decentralized production when they need to be monitored: in times of trouble. As contingent corporate monitors the Japanese main banks are supposed to take control of corporations when sitting managers demonstrate incapacity; in practice, during the current recession, the banks have not themselves demonstrated much capacity to act on such contingencies. Firms under their supervision have wasted free cash flow in American style. Moreover, current developments have probably accelerated the trend away from bank lending to corporations that began when large Japanese firms were allowed to issue securities directly. For its part, the US shareholder system of monitoring has never failed at the supervision of Japanese style corporations for the simple reason that it has not yet had a chance to: Allowing decentralization to proceed produces improved performance insofar as there are gains from decentralization, but such improvement is no guarantee that the permissive governance conditions are also suited to early detection of errors in the emergent system. On the contrary: If the foregoing is not too far off the mark, the US system as currently constituted might fail too, and for reasons related to the deep-seated difficulties of main-bank monitoring.
The problem is that the information generated by the Japanese or learning by monitoring system for its day-to-day and medium-term needs is presumably also necessary to understand how to correct large errors when they occur, yet is not available to the "contingent" monitors -- main banks or shareholders -- who intervene when such emergencies occur. The paradox is that only agents monitoring the (new) corporation day-to-day -- which is to say participating in its routine project selection and evaluation procedures -- could know enough of its highly decentralized operations to correct large errors in an effective way; but just such agents are discredited when the errors come to light; and outsiders, whatever bundle of interests they are trying to maximize, simply can not learn enough fast enough to be useful. From this perspective, therefore, the differences between main-banks with their view of the corporation as a community and shareholders with their vengeful selfishness are less important than the similarities in their limitations. Doing badly in the name of good cause is presumably better than doing badly in the name of the bad, but this not much of a consolation for the victims of incompetence.
But of course there is a way out. There are in both systems providers of capital who also supply efficiency-increasing information -- ideas about co-investment possibilities, about alternate managerial strategies and the like. In this their relation to the "Japanese" corporation is like the relation of its suppliers of components or skill. In the US these are the venture capitalists, who operate not only in high-tech industries but, increasingly, in the re-structuring of mature sectors through "leveraged build ups" and other novel devices that allow for profound restructuring of the firm's relation to its customers, use of information, and organization of production. Their Japanese counterparts are the managers in the large corporations who decide on the internal diversification strategy. Both know how to combine managerial advice and guidance, rooted in deep knowledge of the decentralized corporation, with expertise in finance. Neither is a shareholder or a stakeholder in the standard sense, for they are too much like managers (or employees or suppliers) to count as the former, and too much like owners -- but of what kind? -- to count as the latter. Perhaps a future conference, dedicated more to the joint exploration of the unknown than to evaluation of what exists, will explore the lessons these groups have to teach for constructing a general governance system that makes the best of the "Japanese" production system as it spreads through the world.
