Aoki Firma Japoneza

American Economic Association
Toward an Economic Model of the Japanese Firm

Author(s): Masahiko Aoki
Source: Journal of Economic Literature, Vol. 28, No. 1 (Mar., 1990), pp. 1-27
Published by: American Economic Association
Stable URL: http://www.jstor.org/stable/2727189 .
Accessed: 21/10/2013 04:00
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to Journal
of Economic Literature.
http://www.jstor.org
This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

All use subject to JSTOR Terms and Conditions
Journal of Economic Literature

Vol. XXVIII (March 1990), pp. 1-27
Toward
an
Economic
Japanese
Model
of
the
Firm
By MasahikoAoki
Stanford University
and
Kyoto University
I owe a great deal to Professor Moses Abramovitz who suggested
that I write this article. I appreciate his constant encouragement
and guidance which aided me throughout its preparation. I am also
indebted to three anonymous referees of the Journal and to Professors
Ryutaro Komiya and Michael Riordan for their critical comments
and helpful suggestions. Needless to say, I am solely responsible for
the views expressed in this paper.
OVER
THE LAST DECADE
a considerable
on shop-floor industrial relations; Taishiro Shirai (1984) on enterprise unionism; Masanori Hashimoto and John Raisian (1985) and Jacob Mincer and Yoshio
Higuchi (1988) on employment duration
and tenure-wage profile; Richard Freeman and Martin Weitzman (1987) and
John Taylor (1989) on the bonus system;
Akiyoshi Horiuchi, Frank Packer, and
Shinichi Fukuda (1988), Paul Sheard
(1989), and Erik Berglof (1989) on bankoriented financing and the main bank system; Albert Ando and Alan Auerbach
(1988) and B. Douglas Bernheim and
John Shoven (1989) on the cost of capital;
Aoki and Nathan Rosenberg (1989), Kenichi Imai, Ikujiro Nonaka, and Hirotaka
Takeuchi (1985), and Eleanor Westney
and Kiyonori Sakakibara (1985) on R & D
organization and rapid product development; Banri Asanuma (1985, 1989)
and Seiichi Kawasaki and John McMillan
(1987) on subcontracting; Abegglen and
economic and management literature has been produced in English dealing with various aspects of Japanese firms
and with comparisons with their Western
counterparts. The wide-ranging subject
matter covered in the literature may be
indicated by reference to the following
representative works: James Abegglen
and George Stalk (1985, ch. 5) and Yasuhiro Monden (1983) on the "kanban"
system just-in-time method) and other
practices of manufacturing operations;
Masahiko Aoki (1986, 1989) and Hideshi
Itoh (1987) on the nonhierarchical mechanism of operational coordination; Harvey Leibenstein (1987), James Lincoln,
Mitsuyo Hanada, and Kerry McBird
(1986), and Toshiaki Tachibanaki (1987)
on the nature of internal hierarchies; Kazuo Koike (1984, 1988, 1989) and Haruo
Shimada and John Macduffie (1986) on
skill formation systems on the job and
1

Journal of Economic Literature, Vol. XXVIII (March 1990)
Stalk (1985, ch. 8), Aoki (1984a, 1988a),

and Ryutaro Komiya (1987) on the nature
of Japanese management.
This list of representative papers is incomplete and possibly biased, but it is
certainly sufficient to indicate the rich
variety of subject matter studied in the
growing literature on Japanese firms.
The aim of this paper, however, is not
to present an exhaustive survey of the
literature, but rather to provide a "unified" treatment of various features of Japanese practices and to suggest a consistent economic model that explains how
Japanese firms operate and what they
achieve. This paper leaves the more detailed assessment of their comparative efficiency to future studies, but the underlying presumption is that one of the
important sources of the industrial
strength (and weakness in certain respects) of the Japanese economy can be
found in the micro-micro (internal) structure of firms. Thus the theme of this paper is parallel to that of a recent study
by the MIT Commission on industrial
productivity (Michael Dertouzos, Richard Lester, and Robert Solow 1989) dealing with the comparative assessment of
the American productive performance
with strong emphasis on the impact of
organizational factors.
This paper asks such questions as: Is
there any logical connection between the
"Japanese" mode of internal operation
and coordination on the one hand and
the so-called Japanese employment system on the other? Is the bank-oriented
business financing related to the internal
structure of Japanese firms in any essential manner? If so, how, and what are
its implications for the nature of management and behavior of firms? The model
that emerges to answer these questions
turns out to be somewhat at odds with
the standard contractual model of the
firm that now prevails in the AngloAmerican theoretical literature. Is this
because the Japanese model is culturally

unique? Or, given the competitive challenge posed by Japanese firms in global
markets, does the difference suggest a
need to broaden the scope and reorder
the focus of the theory of the firm? Toward the end of this essay, I will touch
briefly on these issues as well.
The organization of this paper is as follows. Section I describes a few important
examples of operational practices in Japanese firms and suggests that horizontal
coordination among operating units
based on knowledge sharing, rather than
skill specialization, is an important internal characteristic of Japanese firms. It
also makes an elementary inquiry into
the efficiency implications of horizontal
coordination. The employees of Japanese
firms are, nonetheless, arranged in a hierarchy of rank, and Section II argues
that Japanese firms depend on employee
competition to achieve higher status
within their hierarchies of rank as a primary incentive device. The hierarchical
nature of the incentive scheme, therefore, complements the nonhierarchical
tendency of operational coordination and
so helps maintain organizational effectiveness and integrity. Section III deals
with the nature of financial control over
Japanese firms in a bank-oriented financial system. It contends that the financial
and internal characteristics of Japanese
firms are closely related to, and complement, each other. Section IV proposes
that, as a result of this complementarity,
the managerial decisions of Japanese
firms are subject to the dual influence
of financial and employee interests. It
discusses the nature and behavioral
consequences of mutual commitments
between management, on one hand, and
employees and banks, on the other. Fundamental theoretical propositions developed in Sections I through IV are summarized in the form of three Duality
Principles. Section V contains concluding

Aoki: Toward an Economic Model of the Japanese Firm

remarks. The Appendix describes the use
of supply contracting by Japanese firms
and relates this use to other aspects of
Japanese companies. Because of space
limitations, detailed data supporting
stated stylized facts are not presented in
this paper. They appear in works cited
above and in my recent book (Aoki
1988a). Moreover, I limit references to
works available in English.
I. Horizontal Coordination
A key to an understanding of Japan's
industrial performance can be found in
the ability of firms in certain industries
to coordinate their operating activities
flexibly and quickly in response to changing market conditions and to changes in
other factors in the industrial environment, as well as to emergent technical
and technological exigencies. Representative Japanese firms have cultivated an
ability for rapid response by developing
an internal scheme in which emergent
information is utilized effectively on-site
and in which operating activities are coordinated among related operating units
on the basis of information sharing. In
this section, I first illustrate this claim
with three examples drawn from operating practices in the automobile and steel
industries and from R & D activities in
manufacturing industry generally. I then
characterize generic aspects of these and
other examples as a mechanism of horizontal coordination that stand in contrast
with the more familiar mechanism of
hierarchical coordination. I summarize
some results of a comparative analysis of
the relative information efficiency of the
two coordination mechanisms; and finally
I discuss their implications for understanding and interpreting Japanese industrial performance. I am aware that
any attempt to draw general propositions
from specific examples runs the risk of
small sample bias. But the following ex-
amples are drawn from many observations made recently in the course of plant
visits and interviews with engineers and
managers, and I am reasonably confident
that they represent widespread and generic elements of Japanese practice.' Also
I believe that the significance of the comparative analysis to follow is very hard
to grasp without breaking open the economists' black box of the production function and gaining a concrete image of how
firms operate without a rigid hierarchical
order.
A. Examples
1. Production Scheduling-An Automobile Industry Case. A number of
studies based on rather crude accounting
methods claim that the automobile industry represents a case in which Japan
has caught up with the forerunners in
production cost efficiency and is now itself running ahead very fast. (Accounting
studies conducted up to the early 1980s
are surveyed in Robert Cole and Taizo
Yakushiji 1984. Michael Cusmano 1985
provides a more careful estimate of
Japanese producers' comparative performance within the framework of accounting.) Melvin Fuss and Leonard
Waverman (1985) have recently cast
doubt on such assertions and have
claimed, on the basis of elaborate econometric analysis, that even in 1980 Japanese industry was only slightly more
productive "at normal capacity utilization
1 Between September 1987 and August 1989, I interviewed managersand engineers of about 50 Japanese manufacturingfirms and banks. In particular,
I conducted intensive studies, which included plant
visits, of the following firms: Nippon Steel Corporation, MatsushitaElectric IndustrialCo., Ltd., Sony
Corporation,Honda Motor Co., Ltd., Toyota Motor
Corporation,Omron Tateisi Electronics Co., Kyowa
Hakko Kogyo Co., Ltd., Tonen Corporation,IBM
Japan, Ltd., KajimaCorporation,OhbayashiCorporation,TorayCorporation,the SumitomoBank, Ltd.,
and the Sanwa Bank, Ltd. A small portion of interview records have been published in Aoki, Koike,
and Iwao Nakatani(1989).

rates" in spite of the apparent cost efficiency of Japanese producers. Because

the automobile industry is characterized
by significant quasi-fixed factors, such as
capital plant, administrative and designing jobs, and product-specific manufacturing facilities, shifts in consumers'
tastes or an economic downturn can have
severe effects on cost efficiency. Fuss and
Waverman maintain that long-run total
factor productivity growth during the
1970s has been underestimated by an order of 30 to 40 percent for North American producers in accounts that neglect
the effects of undercapacity utilization.
It may be questioned, however,
whether it is appropriate to dismiss inadequate capacity utilization among North
American producers at that time simply
as "short-rundisequilibrium" that had no
bearing on understanding the nature of
comparative cost efficiency. I rather
maintain that the organizational ability
of firms to generate a product mix attuned to unpredictable market demands
and emerging technology without generating an excessive under-utilization of
equipment and a large stock of product
and in-process inventories may be quite
relevant for the comparative assessment
of the organizational efficiency of firms.
In the stable oligopolistic markets for
which fairly standardized products are
supplied, "short-run" market demands
may be reasonably predictable. Under
such a situation, a short-run production
plan may be fixed firmly for a certain
period of time, based on the management's prior knowledge of market demands. The whole production process
may be divided into a series of specialized functions, each standardized, and
each work unit may be required to perform exclusively the specific jobs dictated
by the prior plan. Ex post adaptation to
actual market conditions may be accomplished through the adjustment of product and in-process inventories. Under
the said market condition, such internal

coordination-the H-mode-may be able
to exploit the informational efficiency of
hierarchical coordination and economies
of specialization.2
The efficiency of the H-mode of coordination may become problematical, however, when diverse consumers' tastes
come to demand a variety of products,
when demands shift in a volatile fashion
from one variety to another, and when
the need to deliver ordered products
without delay becomes imperative for
gaining a competitive edge. How have
Japanese auto producers responded to
these challenges and consequently
gained competitive advantage? In the
concrete case of a representative auto
producer, manufacturing coordination
takes place roughly as follows:
The central production planning office
drafts the quarterly and monthly production plans for each factory, based on its
market demand forecasts, and presents
corresponding procurement plans to outside suppliers. These prior plans provide
only a general guideline, however. The
integrated production-delivery plan for
a ten-day period is prepared by the commodity-flow office on the basis of orders
from regional and overseas dealers. In
response to this plan, the engineering
office of each factory prepares a sequence
of daily production schedules. The daily
schedule is then adjusted two days prior
to actual productioxiin response to actual
customers' orders transmitted from dealers to the factory through the on-line network system (the "daily adjustments").
On the final assembly line, wagons, twodoor hatchbacks, and four-door sedans
with red, beige, and white bodies; with
left-hand (for export) and right-hand (for
domestic) steering wheels; with a variety
of transmissions, engines, and options;
2 For a discussion of efficiency of hierarchical coordination, see Oliver Williamson 1985, ch. 9; Aoki
1988a, ch. 2.2


are rolled over seemingly at random. Actually, the sequence of production of varieties of cars is scheduled to minimize
inventory and to respond to the daily customers' orders with the maximum flexibility.
The supply schedule of parts is fine
tuned according to daily production
scheduling through the famed "kanban"
system (Monden 1983; Abegglen and
Stalk 1985, ch. 5). In this system, neighboring shops (or the prime manufacturer
and the supplier) are connected directly
with each other through the chain-linked
circulation of "kanban" (a tag-like card).
Downstream shops, several times a day,
issue order forms called kanban to the
immediate upstream shops (or the supplier) for the provision of particular numbers of parts of various types at a particular time. These kanban in turn are
returned to the original issuer as a check
on the implementation of downstream
The periodical
orders.
automatized
chain-linked circulation of the kanban is
known to be effective for fine tuning the
prior supply plan, while reducing the
time and effort needed for intershop
communications and negotiations without hierarchical intervention by the production planning office and the procurement office.
The firm I described produces about
20,000 kinds of cars, distinguished by the
ways various features are combined, and
about 50 percent of total cars sold are
produced at the rate of fewer than fifty
vehicles per kind per month. Currently,
the number of vehicles produced according to actual customer's orders at dealers
is about 35 percent of total production
(approximately 6,000 vehicles per working day) and these vehicles are delivered
to customers 8 to 12 days after they are
ordered by customers (2 to 3 days for
production scheduling, 2 days for manufacturing, 2 to 3 days for delivery, and
2 to 4 days for holidays). After the cen-
tralized indicative monthly plan has been

formulated, actual production, parts supply, and delivery are coordinated in a
horizontal manner, with the decisive lead
emanating from the customer end. The
computer network system of this auto
firm is akin to the airline reservation system, but it distinguishes itself by integrating the manufacturing process involving thousands of steps and parts into
the system.
2. Quality Control-A Steel Industry

Case. The traditional steel manufacturing method was composed of a sequence
of mutually disjointed work processes,
such as the blast furnace process, the
open hearth process, ingot making, slabbing, rolling, heat treatment, and plating. If products are standardized and sold
for mass markets, such disjointedly sequential processes may be effectively integrated by the H-mode of coordination.
One of the most important characteristics
of process innovation by leading Japanese
steel manufacturers in the last two or
three decades has been, however, to
transform the traditional steel-making
processes into a continuous operation by
the introduction of such techniques as
continuous casting, continuous rolling,
and continuous annealing to obtain savings of energy, time, transportation costs
and in-process inventory. Meanwhile,
demands by auto manufacturers, electrical and other machinery makers, and
construction contractors have become increasingly diverse and specific so that integrated steel plants today produce varieties of final products in small and
medium batches in response to actual
and anticipated orders, but not for anonymous markets. Under such technical and
market conditions, the flexible computerized production control system becomes imperative for transforming actual
orders into continuous process control.
This seems to be clear; however, in order
for continuous processes to be workable,

horizontal coordination among related

processes also seems to be essential. Let
us illustrate this from the quality control
point of view.
Consider, for example, the related processes of casting and rolling. In the Western plant, linkage is patterned after the
H-mode. That is, the engineer at the
quality control office sets engineering
specifications for each process (e. g., a
mineral composition for the casting process and heating temperature for the
rolling process) according to his prior engineering knowledge. Each process manager is required to implement the directive and is relieved of responsibility if
he realizes the given target within a certain error margin. On the other hand,
the engineer would not step into the
workshop. But for continuous processes
there may well be a quality defect problem that can be detected only at the end
of the rolling process, but that originates
from a scar of the equipment at the congealing step of the casting process. Who
can discover this problem, pinpoint its
source, and institute a remedy to prevent
its recurrence? If the engineers isolate
themselves from workshops and if process control is segmented and relegated
to each process manager, it is indeed difficult to cope with such problems.
At a representative Japanese plant, an
"integrated engineering control room"
exists side by side with the engineering
office for each workshop. They are not
hierarchically ordered in terms of status.
In fact, there is often a rotation of personnel between the two to facilitate knowledge sharing among them and discourage
the development of shop-centered interests. The integrated control room and the
workshop office are differentiated only by
primary responsibility. That is, the integrated engineering control room is responsible for locating and solving "crossshop" problems through discussion and
bargaining with offices in charge of indi-
vidual workshops. In the actual solution

of cross-shop problems the control room
acts like a coordinator at the same horizontal level with the workshops. This organizational device is said to be one of
central foci of recent technical assistance
by Japanese steel manufacturers to
American and other steel manufacturers.
3. Product Development. The development of any new product proceeds in
a hierarchical order: It starts with the
basic conceptualization of the new product and its analytic design and proceeds
to the successive phases of detailed design; prototype fabrication, testing, and
redesign; mass production and shipping;
and marketing. Such development processes may be also characterized by various feedback loops. The first conceptualization of a new product may arise from
recognition of a potential market opportunity prompted by marketing activity.
Engineering experience at the testing
phase and service experience at the aftersales maintenance phase may be fed into
the redesign phase. These features may
be thought of as universal phenomena.3
Yet there seems to be a subtle difference
between representative Western manufacturers and Japanese manufacturers in
the ways that feedback mechanisms are
utilized (Aoki and Rosenberg 1989; Imai,
Nonaka, and Takeuchi 1985).
In the case of a representative American computer manufacturer, each developmental phase is made clearly distinct
in such a way that the transition from
one phase to the next is subject to a strict
independent phase-review. The phasereview is carried out from the viewpoint
of engineering as well as business (cost,
marketability, etc.). It is therefore conducted by a group of people drawn from
3 For a model of innovation incorporating the feedback mechanism as an essential element see Stephen
Klein and Rosenberg (1986). For an application of
this model (the "chain-linked" model) to the U.S. Japan comparison, see Aoki and Rosenberg (1989).


various internal divisions, such as the
manufacturing division, the patent section, the sales and services engineering
division. If a development project at one
stage does not pass the review, it cannot
normally proceed to the next stage. It
may happen that representatives from
the manufacturing plant participate in
the early design phase, but in such case
the internal transfer payment for participating personnel is to be made from the
development department to the plant
and the independence of each phase is
kept clear. The feedback loop for redesign is also formalized. If some engineering change occurs at the factory level,
this information is fed back to the design
department, which formally redrafts the
design accordingly and files the revised
design into the "corporate development
and production record system" so that
the same updated engineering specification is automatically available to all factories and service and maintenance divisions all over the world.
At representative Japanese manufacturers as well, phase-review systems of
various kinds are normally instituted.
But the transition from one phase to the
next is in general more flexible and different phases are actually intermeshed and
overlapped. For example, there is more
interaction between design engineers
and plant engineers at the early phase
of design and it is often difficult to say
where the phase of prototype fabrication
and testing actually starts. Very often,
the design laboratory is located at the
site of the plant and there is even a rotation of engineers between the design laboratory and the plant engineering office,
which facilitates "informal" information
sharing between the two. Also the engineer who has been responsible for the
design of a new product, but who possibly has passed the peak of his productivity in design and development may be
transferred to a plant as a line manager
responsible for the manufacturing phase

of the new product (Westney and Sakakibara 1985). This practice motivates the
design engineer to be very attentive to
suggestions and opinions from the plant
level in order to avoid the embarrassment he would face if engineering problems occur at the manufacturing phase
because of a design flaw.
In the case of one auto manufacturer,
the development team is called the SED
team, which is composed of members
from the sales, manufacturing engineering, and development divisions. This
cross-disciplinary team is organized at a
very early phase of product development
and carries out development processes
to the very end as a group. In other
words, feedback loops are internalized
within the team. As much as possible
cross-jurisdictional problems are discussed and solved collectively at every
phase of development. A developed project is proposed for the formal phase review by executive managers, who are also
drawn from SED divisions, only if the
plan is adequately matured. By contrast
with the Western manufacturer's case
described above, the Japanese review
committee is said not to exercise a decisive autonomous judgment. Without any
value judgment at this point, one may
say that feedback loops of development
processes or, more generally speaking,
horizontal coordination among various
organizational units, are more intense,
yet tend to be more informal (undocumented) in the Japanese case.
B. Theory and Interpretations
These examples of industrial practices
in Japan suggest that the coordination
mode that operates within representative
Japanese firms differs from the traditional
modeling of organizational hierarchiesthe H-mode. Let us now try to identify
the fundamental differences between the

two modes by focusing on a few important factors and examine how the relative
cost efficiencies of the two can differ in
various environments. Doing this, one
is bound to commit the sin of oversimplification. But the point here is to make
it clear that the cost function of the firm
is not exogenously and solely determined
by an engineer's blueprint; it also depends on organizational and human factors. I shall, therefore, venture to offer
a sharp theoretical formulation of nonhierarchical coordination and then interpret its performance characteristics in
the context of the Japanese economy.
The H-mode has two essential features: (1) the hierarchical separation between planning and implemental operation and (2) the emphasis on the
economies of specialization. That is,
planning, such as for production scheduling, manufacturing process control, and
commodity development is entrusted to
an office at the top level of each function
(e.g., the production planning office, the
engineering office, the development laboratory) which is supposed to have specialized prior knowledge (on markets, engineering know how, etc.). Let us call
this planning prior planning. Prior planning is fixed for a certain period of time
and implemented by operating units of
the lower levels (e.g., workshops, plant),
each of which is assigned a hierarchically
decomposed special operational function.
Any random event during the implementation period may be coped with by a
priori devices (e.g., buffer inventories,
troubleshooting specialists such as reliefmen and mechanics), and new knowledge that emerges may be used only for
the next round of planning by the higher
office.
Consider an alternative mode reflecting aspects of Japanese firms-let us call
it the J-mode, which has two main features: (1) the horizontal coordination
among operating units based on (2) the
sharing of ex post on-site information

(learned results). That is, prior planning
sets only the indicative framework of operation. As new information becomes
available to operating units during the
implementation period (e.g., customers'
orders at dealers, quality defect problems that become apparent at a workshop, engineering problems associated
with development of a new product that
become evident only at the plant site),
prior plans may be modified. But in order
for on-site information to be utilized in
a way that is consistent with the organizational goal, adaptation must be coordinated among interrelated
operating
units.
In the J-mode, on-site information may
be better ultilized for the realization of
organizational goals (more formally, one
may say that the J-mode can generate
information value by the use of ex post
information). Such a gain, of course, is
not costless. In the J-mode, economies
of specialization of operational activities
are sacrificed, for-some portion of time
and effort of the operating units needs
to be diverted for acquiring new information (i.e., learning) as well as for communicating and bargaining with each other
for coordination. Such costs may be reduced by the development of information
technology: hardware, software, and humanware. Therefore, the comparative
advantage of the H-mode and the J-mode
depends on such factors as the learning
ability of personnel, the ease of communication among operating units, and the
degree of economies of specialization
with regard to the variety and volatility
of market demand. Aoki (1986, 1989) and
Itoh (1987) (see also Jacques Cremer
1989) examined the advantages and disadvantages of the two modes and came
up with the following noteworthy proposition: When environments for planning
(e.g., markets, engineering process, development opportunity) are stable, learn-


ing at the operational level may not add
much information value to prior planning, and the sacrifice of economies of
specialization in operational activities
may not be worthwhile. On the other
hand, if environments are extremely volatile or uncertain, decentralized adaptation to environmental changes may yield
highly unstable results. In both these two
contrasting cases, the H-mode may be
superior in achieving the organizational
goal. In the intermediate situation, however, where external environments are
continually changing but not too drastically, the J-mode is superior. In this case,
the information value created by learning
and horizontal coordination at the operational level may more than compensate
for the loss of efficiency due to the sacrifice of operational specialization.
This result is consistent with the oftenstated suggestion that the hierarchical
mode of coordination based on a highly
developed specialization scheme, which
prevailed in the American steel and auto
industries until the late 1960s, lost its
advantage in the face of product variation
and weakening oligopolistic control (Michael Piore and Charles Sabel 1984).
Economies of specialization may be exploited more favorably for the stable and
large-scale production of standardized
commodities, but not for the small- and
medium-batch production of varieties of
products in a high-volume assembly process where thousands of independent
steps must be coordinated. These markets are precisely the ones where Japanese manufacturers exhibit strong competitive capabilites; however, "in simpler
processes, such as a foundry, where perhaps thirty operational steps are required, the Japanese advantage is slight,
and sometimes non existent" (Abegglen
and Stalk 1985, p. 61).
These propositions are consistent with
the observation that Japanese manufacturers have shown relative strength in
process innovation, as exemplified in the

steel industry, to which intense interactions between engineers in the development laboratoryand engineers, and even
workers, at the factory site may contribute (Edwin Mansfield 1988). On the
other hand, Japanese manufacturershave
not shown a conspicuous advantage in
highly uncertain innovations involving
new conceptualizations of market potential and highly specialized scientific approaches. Nor have they acquired a competitive edge in industries where there
is customized production of newly designed products, such as in the aerospace
industry (David Mowery and Rosenberg
1985).
It is obvious that greater efficacy of
communications and learning at the
opertional level tends to favor the J-mode
relative to the H-mode in a large and
perhaps growing sector of industry. It is
interesting to note, in this connection,
that the J-mode of horizontal coordination based on shared learning at the factory site has emerged and developed in
the last two decades or so by relying on
highly qualified and diligent blue-collar
workers who have formed the core of the
work team. They were mainly recruited
directly out of high schools in the 1950s
and early 1960s when the share of male
persons who found employment right after junior high school was still as high
as 45 to 25 percent depending on business cycle conditions (currently less than
4 percent). Now that the economic obstacles for qualified and motivated youths
to advance to higher education have been
virtually removed, a serious challenge
facing representative manufacturing
firms in Japan today is to recruit highly
qualified blue-collar workers.
The J-mode as practiced by Japanese
firms has one feature in common with
advanced Western firms as exemplified
by the American computer manufacturer
described in example (3) above, that is,

10
knowledge sharing among various organizational units. The subtle difference is,
however, that in the case of the American
manufacturer knowledge sharing is formally instituted based on explicit documentation through the computerized
network system and other technological
means. In the Japanese case, by contrast,
knowledge sharing and the horizontal coordination based on it are often informal
and based on verbal communications
(even tacit understanding), although here
too there is increasing use of the computer network system as indicated by the
integrated marketing-manufacturingnetwork system of the auto manufacturerdescribed in example (1). Such undocumented communications may generate
information value by the finer use of onsite information that is too subtle or cumbersome to document usefully. But the
efficacy of face-to-face communications is
limited by geographical proximity. (It
may be recalled that the research laboratories of Japanese manufacturers are often located at the factory site. The need
for geographical proximity also may explain why there is such a high concentration of business activity in Tokyo.) Also,
the ethnic homogeneity of the Japanese
domestic factory may have been a crucial
factor for the development and effectiveness of the J-mode (Aoki 1988a, ch. 7).
Further, the dual functions of performing
an operating task and learning-communicating-bargaining in contributing to
smooth horizontal coordination may require intense effort on the part of bluecollar workers. It is clear that all these
factors are now subject to serious challenge. On one hand, Japanese firms are
expanding their activities globally to an
ever increasing degree and, on the other
hand, their ability to recruit qualified
blue-collar workers willing to carry out
intense work on the shop floor is becoming problematical as a result of recently
acquired affluence.
II. The Hierarchy of Ranks

Having examined the coordination
mode of the Japanese firm, we now turn
to its incentive system to see how the
characteristics of these two aspects of Japanese firms are related to each other. We
begin by listing some factors required for
the efficient and effective operation of the
J-mode that may have important bearings
on the nature of the incentive system at
Japanese firms.
1. In the H-mode, operating tasks are
separated from the coordinating task and
divided into specific functions. Operating
skills valued in this mode are therefore
specialized skills. In the J-mode, however, operating units are expected to be
engaged in mutually coordinating their
tasks as well. Exclusive attention to the
efficent performance of a particular operating task in isolation may not contribute
to overall efficiency (i.e., the generation
of information value made possible
through efficient horizontal coordination
based on information sharing) and may
not be appreciated much. In addition to
skills in particular operating tasks, the
ability to communicate and work together with peers and others with different functions is required. (Recall the role
of the engineer in the integrated process
control office in the steel industry in example (2) in last section.)
2. As suggested by example (1) in the
previous section, smooth adaptation of
production scheduling to emergent customer demands through horizontal coordination requires each operating unit to
be capable of responding to needed
changes quickly (e.g., in jigs and tools)
and of coping independently with somewhat unusual problems (e. g., breakdown
of a machine, quality defect of an in-process product). Otherwise, smooth operation may be disrupted. Workers in the
final assembly line are entitled to stop
the line whenever they see problems that


would justify doing so. The problem is
handled on the spot, possibly with the
help of neighboring workers, subforemen, and so on, but not by calling in
outside specialists such as mechanists.
The integration of operating skill with autonomous problem-solving capability can
be assured only when the worker has a
good understanding of the relevant work
process as a whole, rather than only a
certain aspect of it. Such general understanding in turn may be nurtured by
making the worker familiar with many
related aspects of the work process. This
is a point that Kazuo Koike has rightly
emphasized in a series of influential papers (Koike 1984, 1988, 1989).
3. In examples (2) and (3) in the last
section, it was pointed out that the practice of job rotation of engineers among
different engineering offices as well as
between engineering jobs and supervisory jobs at the factory facilitates the
knowledge sharing needed for horizontal
coordination among different phases of
engineering and development processes.
For similar and other purposes, whitecollar workers on the lifetime career track
(and sometimes even blue-collar workers) are rotated among various jobs in
different offices and workshops every few
years. Such rotation familiarizes workers
with various jobs and enhances their ability to process and communicate information needed for the efficient operation
of the J-mode. Regular rotation also prevents workers from identifying themselves strongly with the interests of specific jobs, workshops, plants, and offices
so that the development and assertion
of local interests inconsistent with the organizational goal are restrained.
These factors point to two important
needs: (1) the design of incentives not
tightly related to a specific job category,
but that motivate wide-ranging job experience among employees; and (2) the development of a personnel office that ad-
11
ministers such incentives and is also responsible for personnel posting, including interjurisdictional rotations, with an
eye to the firm's long-run organizational
goals.
As for the first, Japanese firms have
developed rank hierarchies as a primary
incentive device, which Aoki (1988a, ch.
2) describes in some detail (also see Ronald Dore 1973 for a classical description).
The essential idea may be summarized
as follows. There are usually separate
rank hierarchies for blue-collar workers,
white-collar workers, and engineers, as
well as one for the supervisory and managerial employees above them. Each rank
carries a certain level of pay, but not a
specific job. Therefore employees in the
same rank may be doing different jobs.
For instance, an engineer at the integrated process control office, an engineer
at development laboratory, and an engineer at the plant site may well be in the
same rank and receive identical pay (possibly with minor allowances for particular
jobs). New entrants to the firm who are
just out of school are placed at appropriate ranks in the nonmanagerial rank
hierarchies determined by their years of
education.
After entry at an identical starting
point for a certain educational credential,
employees compete for promotion in
rank throughout their careers. The criteria for promotion are years of service and
merit, with the latter not specifically related to particular jobs but to broadly defined problem-solving abilities, communication skills, and so on. The speed of
promotion for all employees is the same
early in their careers, however, as young
employees are considered to be in training and their aptitude for the firm's specific implementation of the J-mode is being tested. Differences in speed of
promotion among employees becomes
more apparent, however, in midcareer
(say, after employees reach their mid-

12
30s). So, the fast flyer among blue-collar

workers may reach the top rank in his
later 30s (and proceed to supervisory
ranks afterward), while the slow mover
may reach the top rank only a year before
mandatory retirement at age 60. Promotion criteria become stricter particularly
for white-collar employee as his career
advances and if an employee does not
exhibit continual progress he or she may
be separated in mid to late career, although an honorable exit is usually arranged by the employing firm by posting
the employee in a less promising quasioutside job at a minor subsidiary or other
related firm. Thus the mystifying notion
of "life-time" employment and the "seniority" system tells only half of the truth.
Also it may be noted that in the late
1980s, the personnel departments of
some large Japanese firms have begun
official recruiting of midcareer specialists
and skilled workers from other firms as
the shortage of such staff has become
more pronounced.
The ways in which rank hierarchy
works as an incentive, that is, the ways
in which it copes with the problems of
moral hazard (the possibility of employees' shirking in the absence of proper
monitoring), adverse selection (the difficulty of hiring the right workers, whose
qualifications cannot be known with certainty prior to employment), and the
provision of motivation for wide-ranging
firm-specific learning and teamwork can
be rigorously analyzed in the light of recent development of the incentive literature, and the theories proposed may be
tested (see Aoki 1988a, ch. 3 for the theoretical analysis of Japanese employment
practices and the relevant literature).
From the theoretical point of view, one
point should be stressed here: The existence of a credible threat of discharge
when the employee does not meet the
criteria for continual promotion plays an
important role in enabling the rank hier-
archy to operate as an effective incentive

to curb shirking. A discharge in midcareer may point to some negative attributes of the discharged so that he or she
may not be able to gain equivalent rank
outside, when information about him or
her is not perfect. So an employee must
compare short-run gains from shirking
with potential losses in wealth due to discharge and consequent demotion. As
stated above, in fact, "lifetime" employment and "seniority" advancement are
not automatic. Otherwise, they would
not be effective as incentives.
Theoretical analysis (e.g., Bentley
MacLeod and James Malcomson 1986)
shows, however, that actual dismissal as
a disciplinary measure seldom needs to
be observed, insofar as the rank-hierarchy system operates well as an effective
monitoring mechanism. Also, the possibility of promotion gives employees a
positive incentive to learn within the context of their employing firms, and the
potential loss of seniority and of retirement compensation related to duration
of employment discourages the midcareer exit of trained employees. As a result, the duration of employment tends
to be relatively long for Japanese workers
(e.g., Hashimoto and Raisian 1985;
Mincer and Higuchi 1988). But, how
much of the seniority rise in employee's
income is due to learning achievement
and consequent productivity increase
(which is explained by human capital theory) and how much to "bonding"for diligence (which is explained by monitoring
theory) is still to be empirically settled.
In order to administer rank hierarchies, Japanese firms have developed the
personnel department as an important
institution. This department has full control of the recruitment of new employees
for career tracks out of school, designs
and runs rank hierarchies (pay scale and
promotion criteria), and rotates whitecollar workers with an eye to the wider


interests of the organization. (The rotation of blue-collar workers may be delegated to a subsection of a personnel department at the plant level.) Since the
personnel department potentially has excessive power because of its control of
promotion and rotation, managers of the
department are usually themselves subject to rotation. To avoid demoralizing
employees by unfair treatment, the criteria for promotion and rotation are designed to be as objective as possible. Also
the supervisory assessment of an employee by multiple supervisors through
the rotation of the employee and supervisors may make actual personnel decisions
consistent with "public opinion" within
the firm. Yet mistakes and personal complaints, reasonable and unreasonable, are
unavoidable. But protest by "exit" is
costly; hence the enterprise union, the
institution through which employees can
"voice" their complaints and grievances,
has developed as a counterpart to the
personnel department.
The enterprise union covers all regular
employees, blue-collar and white. It does
not negotiate about "the" wage rate for
each job category; it does negotiate about
the base pay for the bottom rank, pay
differentials among ranks, and the admissible range regarding the speed of
promotion. Within the negotiated agreement, however, the personnel department has discretion over the ranking and
job assignments of employees. One of the
important roles of the enterprise union
is to absorb employees' grievances about
personnel decisions and to monitor the
fairness of the personnel administration
of the firm. It is not accidental that
unions take an enterprise-based form at
Japanese firms rather than that of industrial or craft unionism as in those economies where workers' careers are more
oriented to a broader market rather than
to the individual firm (Aoki 1984a, part
3; Koike 1984; Shirai 1984).
13
Thus there seems to be an interesting

asymmetry between typical Western
firms and typical Japanese firms regarding ways that incentive and coordination
modes are interrelated. In the Japanese
firm, rank hierarchy is used as a primary
incentive device, while the coordination
mode is less hierarchical. In contrast, the
Western firm combines a relatively more
hierarchical approach to coordination
with the relatively decentralized market
approach to incentives (i.e., clear employment contracts relating specific jobs
to competitive wages may be written).
In the Japanese firm, management may
feel secure in delegating tasks of coordination to lower levels where relevant onsite information is available, because employees are aware that they are being
evaluated by their own long-term contributions to organizational goals. Thus they
are induced to comply with management
authority without explicit hierarchical direction over daily operation. On the
other hand, in the societal environment
where more individualistic values prevail, management authority is not automatically assured within the firm organization and needs to be asserted by a
contractually agreed, hierarchical structure of decision making.
Keeping in mind that the alleged
asymmetry is an overly simplified stylization,4 let us dare to summarize it in the
following manner;
The First Duality Principle: In order for

firms to be internally integrative and or4 Organizational practices usually attributed to
"Japanesemanagement,"such as life-time employment (actuallylong-term employment), internal career advancementbased on seniority and merit, the
rotation of personnel over jobs and broaderjob assignment, bonuses, have also been widely observed
with respect to the personnel administrationof whitecollar employees of many "well-run"Western firms
for quite some time (e.g., Fred Foulkes 1980). Koike
(1984) even described the position of blue-collar
workers in Japanese firms, albeit somewhat controversially, as "the white collarizationof blue-collar
workers."

14
ganizationally effective, either their coordination or their incentive mode needs

to be hierarchical, but not both. Comparatively speaking, Japanese firms
tend to be less hierarchical in the coordination mode, while they rely upon
rank hierarchies in their incentive system.
III. The Nature of Bank-oriented
Financial Control
In this section, I take up the financial
aspect of Japanese firms and examine its
characteristics. I argue that there is a
close and logical symmetry between the
internal organizational aspect and the financial control aspect of Japanese firms,
and I summarize this symmetry as the
Second Duality Principle. As regards the
financial control aspect, I need to explain
the following two stylized facts. I regard
the two as parts of a unified and consistent system, even if they do not appear
to be necessarily consistent internally
and mutually from the orthodox neoclassical (contractual) view of financial control (for detailed data supporting the stylized facts summarized here, consult Aoki
1988a, ch. 4).
.1. In Japan, banks (city banks, trust
banks, local banks, and others-each
specialized in different business activities
subject to regulation and administrative
guidance by the Ministry of Finance) are
allowed to hold stocks of nonfinancial
companies up to the maximum of 5 percent of the shares of each stock. Financial
institutions as a whole (including insurance companies) own about 40 percent
of the total stock outstanding of listed
companies, but there is usually one or a
few influential city banks for each listed
company that own up to, or close to, the
maximum. There is also one city bank
for each company among its major stockholders, called the main bank, which has
the closest tie in terms of cash manage-
ment, as well as short-term (and longterm) credits. The main bank plays the
role of manager of a loan consortium
when a group of banks extends major
long-term credit to the company, and it
is responsible for closely monitoring
the business affairs of the company. If
the company suffers a business crisis, the
main bank assumes major responsibility
for various rescue operations, which include the rescheduling of loan payments,
emergency loans, advice for the liquidation of some assets, the facilitation of
business opportunities, the supply of
mangement resources, and finally reorganization, to secure the claims of the
consortium (Sheard 1989). In the normal
course of events, however, the main bank
exercises explicit control neither in the
selection of management nor in corporate
policy making. Well-run companies that
incur little or no debt from banks appear
to be virtually free from banks' intervention, and their managements enjoy the
highest degree of autonomy.5
2. Because a large proportion of the
equities held by banks and other corporate entities are extremely stable, the
managements of Japanese firms are insulated from takeover raids through the
open market. Individual stockholders,
who own only about 30 percent of outstanding total equities of listed companies, do not have any effective voice in
the corporate governance structure. As
long as a company is well run, top management is selected by the outgoing president (or chairman)from among the firm's
own managerial corps. Thus top management is considered to be the highest rank
for the career advancement of employ5For instance, 15 percent of the stock of Toyota
Motor Corporation is owned by three city banks,
Tokai, Mitsui, and Sanwa, but there is no collusion
on the part of these banks to influence the corporate
direction and management selection by Toyota. They
do, however, woo Toyota management competitively
for commercial dealings.


ees. From this observation, some authors
(e.g., Abegglen and Stalk 1985; Komiya
1987, 1989) infer that Japanese firms are
run virtually in the interests of employees subject to some degree of financial
constraint. At the same time, however,
individual stockholders have enjoyed the
high annual after-tax real market rate of
returns to stockholding of 11.7 percent
(the standard error 18.5 percent) over the
period 1963-86 (see Aoki 1988a, pp. 11316). The economic interests of individual
stockholders do not seem to be entirely
neglected at least ex post.
Thus there seem to be paradoxes. On
the one hand, banks are major stockholders of companies, but they do not appear
to exercise vertical control over management of these companies and they remain
as silent business partners in good profit
states. Their power becomes visible only
in bad states. On the other hand, even
though the individual stockholders have
only apparently weak control over firms,
they are de facto beneficiaries of Japanese
firms. But these are paradoxes only from
the viewpoint of orthodox financial theory in which stockholders are assumed
(or at least ought) to control management
through the corporate governance structure as well as the threat of market discipline (i.e., takeover raids). But this does
not mean that Japanese firms are completely free of financial control (monitoring).
6 For a more detailed estimate of the market rate
of stockholding based on firm-based data, see Ando
and Auerbach (1988). Taking into account both taxes
and risk premia, Bernheim and Shoven (1989) attempt to estimate the cost of capital for Japanese
firms. Whether the ex ante cost of capital facing Japanese firms is relatively lower or higher than the
American counterpart has not been empirically settled yet and is still controversial in spite of an often
made allegation that the cost is relatively lower (e.g.,
George Hatsopoulos, Paul Krugman, and Lawrence
Summers 1988). See Aoki (1988a, pp. 112-13) for a
criticism of the casual comparison of costs of capital
between different economies based on raw data unadjusted for different tax and accounting conventions.
15
First, let us note that main bank's control manifests itself in bad profit states.
If the profit of a company starts to decline, the main bank is able to detect
the problem at a rather early stage
through information gained from the
management of commercial accounts,
short-term credits, long-term personal
contacts with top management of the
company and its business partners, and
so on, because of its special position.7
Tacit and explicit pressure for the internal overhaul of management would be
initiated in exchange for various types
of rescue operations as noted before. If
bad states continue, the main bank may
decide to take over management through
the governance structure of the company
(stockholder's meeting, board of directors). Recent experiences indicate, however, that banks do not change the fundamental nature of internal management,
but rather hand over top management
after recuperation to internally promoted
employees (see Richard Pascale and
Thomas Rohlen 1983 for an interesting
case study of the bank takeover of
Mazda).
It is because of this possibility of bank
takeover that I have adopted a definition
of the main bank, which focuses on its
role as a major stockholder as well as a
main lender (or manager of a loan consortium). I would argue that although in the
normal state bank control is not visible,
the potential threat of bank takeover may
play an important monitoring function
when the financial system is viewed as
a whole. A recent contribution by Philippe Aghion and Patrick Bolton (1988)
is suggestive in this regard. Aghion and
7 In the bank-orientedsystem in Japan, a single
bank, that is, the main bank, monitorsthe company
closely as the major cash manager-creditor-stockholder so that it is .in a position to detect a problem
early. In contrast, at the time of near bankruptcyof
Chrysler, about 400 banks had extended credit to
Chrysler and no bank was aware of the magnitude
of the problem until the real crisis became evident.

16
Bolton argue that the essence of debt

contracts lies neither in its return characteristics (such as fixed-interest claims
combined with bankruptcy risk) nor in
defining a liquidation point, but in the
following arrangement regarding the
transfer of control: As long as good profit
states continue, the outside investor
(bank) does not intervene and inside
management continues to manage; otherwise the outside investor (bank) takes
and reorganizes.
over management
Within the specific context of their
model, they showed that such a financial
system can achieve ex ante a Pareto superior outcome under certain conditions.
The external investor with control rights
may be excessively risk-prone in good
profit states, while inside management
may be excessively wasteful in bad states
by trying to survive at the creditors' expense.
Needless to say, the Japanese situation
is much more complicated than the
Aghion-Bolton model. Also Aghion and
Bolton's analytical result does not imply
that bank-oriented financial control is always efficient. But one cannot deny that
in Japanese practice there is a close positive correlation between the degree of
management freedom from bank control
and the level of corporate profits. Because "internal" management naturally
abhors "external" interference by the
main bank, and bank control is a humiliating blow to the failed manager, management aspires to pursue high profits.
It may thus be considered that corporate
ranking by profit is an effective discipline
exercised over the apparently autonomous management of Japanese firms,
even though its form is different from
that envisioned in the orthodox AngloAmerican financial literature (hierarchical control of stockholders through the
corporate governance structure combined with market-oriented incentive
contracting for managers and the threat
of takeover).8 This observation suggests

that the nature of the financial control
over Japanese firms is symmetric to that
of the internal control within them:
There is no clear hierarchical division between corporate control and managerial
operation. Financial control by bank cum
stockholders concerning corporate direction is exercised only in a business crisis.
Otherwise corporate decision is delegated entirely to management "on-site."
Management, however, is indirectly disciplined through competition over ranking, this time, ranking of firms by corporate profits. A question may then be
raised. Is there any logical reason for the
said symmetry: weak-decision hierarchy
cum incentive-ranking hierarchy (WDIR)
both in the internal and financial aspects.
In other words, can a WDIR-type internal structure be combined effectively
with a non-WDIR-type financial system
as envisioned in the orthodox finance literature?
First, note that in order for horizontal
coordination within internal organization
to operate most effectively (in the sense
of being able to achieve organizational
8
It is doubtless true that North American managers are potentially more vulnerable to takeover raids.
There are, however, some interesting developments
that may suggest a convergent phenomenon. First,
the growth of institutional stockholding such as by
pension funds in the U.S. tends to stabilize a large
proportion of the ownership of major firms' equities.
As a result, a closer relationship between management and major institutional investors based on the
sharing of business information may emerge. Also,
in the U.S. the spread of leveraged buyouts has
brought in many of the features associated with bankoriented financial contracting, such as a high debt
ratio and combined debt and equity holdings (strip
financing). In the case of a management leveraged
buyout, management becomes more insulated from
possible raids by outsiders, but vulnerable to banks'
controls when the firm's performance deteriorates
(Erik Bergl6f 1989). Whether these phenomena are
temporary or not (except for the growth of pension
funds and its implications) is not yet clear, but, together with increasing reliance of Japanese firms on
bond financing abroad, the distinction between the
bank-oriented system and the market-oriented financial system is becoming somewhat blurred.


goals), management decision needs to be
continually adapted in response to onsite information available at all levels of
the functional hierarchy. Further, organizational goals themselves may need to
be adjusted in response to employees'
voice. In this way, a sense of joint effort
is created so that employees' active cooperation in horizontal coordination may be
elicited. This amounts to a set of mutual
vertical commitments in which management recognizes the interests of employees and, in return, employees exert
greater effort. Commenting on my previous work (Aoki 1986) Leibenstein wrote:
"We are likely to find that a high degree
of mutual vertical commitment creates
a sense of common objectives and a concern for the results beyond one's own
job, which in turn is likely to support
horizontal coordination on production
goals. . . . Thus while I agree with Aoki
on the importance of horizontal cooperation I would argue that where horizontal
cooperation on production goals flourishes it is likely to be a consequence of
a high degree of mutual vertical commitments" (p. 170).
Toward the end of the next section, I
inquire more precisely into the logical
structure of the mutual vertical commitments which can yield a Pareto superior
outcome for financial and employees' interests. Anticipating that, I only maintain
here that the sharing of knowledge, as
well as the partial identification of interests, across various levels of internal organization is essential for facilitating
vertical interactions, especially for management to absorb information and demands at the lower level, for corporate
decision making. The internal selection
of management through promotional
ranking may serve as an effective mechanism by which such knowledge sharing
and interest identification are nurtured
on the basis of the sharing of experiences.
Further, rank hierarchy as an incentive
17
device becomes fully operative only if the

internal promotion ladder for employees
extends to as high as the top executive
position. And, the autonomy of internal
management from external financial interests is guaranteed under the WDIRtype financial control as long as the profit
state is satisfactory.
From the perspective of incentive as
well, the WDIR symmetry seems to be
logical. Under an employment contract
established in the context of a rank hierarchy, operating employees tend to be
associated with a particular firm on a
more or less permanent basis because
midcareer mobility (exit) may be costly
(Hashimoto and Raisian 1985; Mincer
and Higuchi 1988). Their lifetime wealth
is at stake with the employing firm in
the form of future claims for seniority
payments and retirement compensations
and this wealth is not salable. Opportunities for risk diversification are thus limited. There needs to be a mechanism,
therefore, that insures employees to
some degree against firm-specific risk.
Because of the stable concentration of
stockholding, as well as continual information flow between firms and banks in
the context of the WDIR financial system, the reorganization of a troubled firm
by the main bank may occur sooner than
under the market-oriented financial system and this may mitigate the risk of
bankruptcy (Berglof 1989). In fact, the
bankruptcy of companies associated with
a reputable main bank rarely occurs in
Japan. Horiuchi, Parker, and Fukuda
(1988) tried to refute empirically the insurance hypothesis regarding the main
bank. But the function of a monitor and
that of an insurer against bankruptcy risk
by the main bank are not necessarily mutually exclusive. At the time of the rescue
operation, the main bank normally bears
more than proportional costs in comparison to other members of the consortium
and this may be explained by the suppo-

18
sition that the main bank's reputation as

a good and responsible monitor is at stake
(Aoki 1988a, pp. 142-49, 232-33). On the
other hand, managers and operating employees are relatively mobile across firms
under Western-type employment consecurities
tracting, but well-developed
markets can provide an efficient mechanism through which individual wealthrisk is diversified and thereby reduced.
Therefore I claim the following:
The Second Duality Principle: The internal organization and financial control of
the Japanese firm are dually characterized by weak-decision hierarchy and incentive-ranking hierarchy. This duality
is not accidental.
IV. Dual Control over Corporate
Management Decision
In this section I discuss questions concerning the goals and purposes of Japanese management. We have seen in the
last section that Japanese management
is relatively independent of external financial control in making corporate decisions. This freedom, however, exists only
so long as a satisfactory state of profits
is maintained. Should that state be seriously compromised, the external power
of the main bank, which stands in the
background, will be exercised. This substantial, if constrained, freedom of Japanese management poses a question: Is
the conventional profit-maximizing objective itself routinely qualified by an admixture of other goals? Does it apply only
as a subsidiary constraint on the pursuit
of other goals? Some influential economists argue in that way, for example, Komiya, who advances the hypothesis, as
a first approximation, that a Japanese firm
"chooses the amount of output and the
amounts of labor and capital inputs so
as to maximize income per employee
. . . after the payment of a fixed share
of profits to stockholders" (1989, p. 115).
Such a presumption essentially boils

down to the model of a worker-controlled
firm that postulates the maximization of
income per worker since the profit motive is assumed not to have any direct
impact on the decisions of corporate management. I predict, however, that it will
be difficult to sustain such a hypothesis,
if its implications for corporate behavior
are empirically tested. For example, the
hypothesis implies that to protect the interests of incumbent employees the
growth rate of the worker-controlled firm
would be slower than that of a profit-maximizing firm (Anthony Atkinson 1973), an
unlikely situation in Japanese firms.
On the other hand, there are reasons
to believe that employees as a group constitute assets specific and internal to the
firm. We have seen that Japanese firms
rely upon a system of horizontal coordination in which employees at the operational level actively participate. Also I
have argued that the information-processing and communicative abilities of
participating employees are nurtured
largely through learning by doing in the
context of a firm-specific coordination
network. Such abilities cannot be acquired in ready-made form prior to membership in the network and their values
cannot be thoroughly realized in isolation
from it. In other words, skills effective
for the creation of information value in
the context of horizontal coordination
may not be classifiable along well-defined
job categories, for which market contracts transferable between firms can be
unambiguously written. One may reason
then that employees of Japanese firms
as a group become assets specific to the
internal network and that rewards for
them are internally determinable (subject to possible external constraints) and
payable out of the value generated by
the network net of costs due to the training of employees, the sacrifice of economies of specialization, and so on.


If so, however, the following hypothetical question may be raised from another
angle: Why do employees not purchase
the physical assets necessary to maintain
the network through debts and replicate
the network, guaranteeing to themselves
its whole value? In short, why would an
employee-controlled
firm not be created? This question arises not only out
of intellectual curiosity alone, for some
authors do argue that Japanese firms are
in effect managed on behalf of their employees. Yet there are reasons that make
it difficult for employees to control their
firms explicitly and entirely. One is the
obvious limitation of the availability of
finance for the purpose of creating the
employee-controlled firm. It may be recalled that monitoring by the main bank
is particularly effective because of its dual
position as a major creditor and a main
stockholder. Since potential creditors are
excluded from equity ownership of the
employee-controlled firm, however, they
may not feel as secure in providing credit
and reluctant to do so.
Further, to convince all employees to
move to the new "clone" firm, they must
agree on how to divide up the value appropriated by the firm. This division involves costs of collective choice which
may be prohibitively high, specifically
when the employees are not homogeneous. Also, George Mailath and Andrew
Postlewaite (1988) argued that if there
are intangible gains that employees get
from the firm (as distinguished from the
network as such) and that they cannot
each verify, they may be induced to exaggerate their reservation wages (that is,
the wage at which they would move to
the proposed clone firm) and that an attempt to induce truthful revelation may
make a proposal for a new viable emfirm untenable. This
ployee-controlled
would be especially true when the size
of the network becomes very large. Such
private reservation wages, for example,
19
may take the following forms. The performances of employees of the Japanese firm
are evaluated and rewarded in the long
run by the elaborate and admittedly impartial personnel administration system
crystallized in the hierarchy of ranks, and
this may provide to workers the long-run
security and the sense of fair treatment
they desire. It does not seem obvious,
however, how the egalitarian idea of the
firm and the cenemployee-controlled
tralized management of hierarchy of
ranks can be made mutually compatible.
Also, the loss of the main bank's services
in monitoring management may impose
costs of monitoring management on each
employee in terms of time, effort, and
resources. Such costs may be private information.
The discussion in the preceding two
paragraphs is admittedly hypothetical,
but it may help us understand that the
impact of financial control over Japanese
firms cannot be neglected entirely even
if employees are network-specific assets.
A portion of the value created by the
network thus accrues to financial investors who supply finance and monitor
management. On the other hand, if employees' reward cannot be entirely determined by external market competition,
but is negotiated internally, employees,
too, would be interested in how corporate decisions are made. Corporate decisions would have an impact on workers'
short-term and long-term positions in the
rank hierarchies that define their lifetime earnings. Employees are not only
interested, they are also able to exercise
influence on corporate decisions. When
employees are promoted to be executive
managers, their motives may well remain
mixed and contain a carry-over from their
longer careers as employees in the lower
ranks. It is true that, as executive managers, they must give attention to profit
making in order to maintain their own
position and autonomy. Yet they may re-

20
tain a degree of identification with the

interests of employees and a degree of
freedom to support them.
Further, for financial interests as well,
it may be reasonable to leave open the
possibility of mutually beneficial exchange between the levels of employees'
earnings and effort, on one hand, and
the direction of corporate decision making, on the other. For example, employees may be willing to trade off current
earnings and expend more effort for
higher job security, which may also raise
the profit level. Further, such exchange
may help preserve the network-specific
assets as well, which would be mutually
beneficial in the long run. Therefore, we
propose the following hypothesis, making its behavioral implications subject to
future empirical testing.
The Third Duality Principle. The corporate management decisions of Japanese
firms are subject to the dual control (influence) of financial interests (ownership)
and employees' interests rather than to
unilateral control in the interests of ownership.
It may be noted that this proposition
departs not only from the usual neoclassical presupposition, but also from the
"share system" view of Martin Weitzman
(1984), although Weitzman claims that
the Japanese economy is "the only industrial economy in the world with anything
remotely resembling a share system" (p.
76). The essential difference lies in the
fact that in the model I propose, not only
distributive shares but also corporate decision making is implicitly or explicitly
subject to "sharing."In Weitzman's theoretical design, only the share parameter
defining the division of value-added between profit and workers' earnings is subject to bilateral agreement. After such
agreement, corporate decision making,
such as on the size of employment, is
the prerogative of management who are
driven exclusively by the profit motive.
If the marginal return from a worker is

diminishing, yet the share parameter is
fixed ex ante at less than one, income
per worker is ever diminishing as the size
of employment increases. On the other
hand, when marginal returns are positive, management is induced to expand
the size of employment. I find it unrealistic to imagine that unions (at least the
Japanese enterprise union) fail to recognize the subsequent outcome of agreeing
with the Weitzman's share contract,
namely that they lose control over remuneration per worker. I would maintain that the hypothesis of dual control
by financial and employees' interests
over corporate decision making is a more
reasonable one, once the network specificity of employees' skills is accepted.
Questions to be asked next are as follows: How do firms behave under dual
control? Is there any qualitative or quantitative difference between the implications of dual control and unilateral ownership control in how firms are run? How
is dual control exercised? What role do
management and employees each play?
First we note that employees as a
group can withdraw cooperation in horizontal coordination if they feel that they
are not treated "fairly"by management
in pecuniary rewards and corporate decision making. On the other hand, the
main bank as the major creditor cum
stockholder can threaten management
with the discipline of bank takeover if a
sufficient level of profits over time is not
assured.9 Further, management's social
9 Strictly speaking, there is a question as to
whether there is any conflict of interest between the
main bank as a creditor cum stockholder and the
individual stockholders. This problem is investigated
in Aoki (1984b; 1988a, pp. 127-38) within the framework of a miniature general financial equilibrium
model incorporating features of taxes and financial
regulation in Japan. This analysis indicates that the
bank prefers its portfolio company to rely more on
debt financing than on equity financing than individual stockholders would do and that the conflict has
been resolved in favor of the bank, although less so
since the mid-1970s.


prestige and autonomy are enhanced if
the managed firm stands higher in corporate profit ranking. We may therefore
suppose that the distribution of a firm's
revenue between employees' earnings
and profits reflects either the relative
bargaining power of the enterprise union
vis-a-vis management acting in the interests of profits or the notion of fairness
by management.'0 Further, let us imagine that management strikes a balance
between employees' interests and financial interests in making corporate policy.
Finally, suppose that, to the degree that
employees trust management corporate
policy making to be fair, they supply
more effort in operating activities and
horizontal coordination than would be
expected under the competitive wage
system (i. e., more effort than maximizing
"individual" labor surplus obtained by
equating the marginal value disutility of
effort with the wage rate).
Such mutual commitments by management and the employees are expected
to yield a Pareto superior outcome under
the assumption of network specificity of
human assets. " The following are some
10 If the
management'snotion of fairnessis represented by Nash'sformulationof symmetryin his axiomatic approach to bargaining (Nash 1950), on one
hand, and if the relative bargaining power of the
union is measured in terms of "boldness"as formulated in cooperative game theory, the distributive
outcomespredictedby the two approachesare identical (see Aoki 1984a, ch. 5).
11Let us imagine as a thought experiment that
managementformulatescorporatepolicy by weighting the policy optimalto the representativeemployee
and the policy optimalto long-runprofitmaking(i.e.,
the present value of the future stream of profits),
with weights being given by each distributive share
in firm-specificquasi rent. I call this policy making
the "weighting rule" (Aoki 1984a, pp. 74-80). Further, assume that the employees supply the level of
effort that maximizes "collective"value surplus by
equating the marginalvalue utility of effortwith the
marginalvalue product (not the wage rate). When
the utility function of the profit claimantand that of
representativeemployee are both the "constantpure
boldness type" in the sense defined in Aoki (1984a,
pp. 74-77), this idealized constellation of mutual
commitments can be proven to yield the outcome
known as the generalized Nash bargainingsolution,
21
of their behavioral implications, in which

the "stockholder-controlled" firm refers
to a firm that maximizes the stock value
(the present value of a stream of future
profits) of the firm under the competitive
wage system.
1. The dually controlled firm pursues
a higher growth rate (or somewhat more
loosely speaking, tends to have a longer
horizon) in investment decision making
than the stockholder-controlled firm facing the same level of employees' current
earnings, because the former takes into
account employees' extra benefits from
the growth of the firm in the form of
enhanced future promotion possibilities
in their rank hierarchies (Aoki 1988a, pp.
164-66).
2. The dually controlled firm sets the
amount of employment at the level at
which the marginal value product of an
additional worker is equal to a worker's
earnings minus the marginal rate of an
implicit unemployment insurance premium. Thus if the employees' fear of unemployment is positive, the dually controlled firm provides a higher degree of
job security than the stockholder-controlled firm (Aoki 1988a, pp. 174-76).
3. In order to protect the interests of
incumbent employees, the dually controlled firm tends to limit the expansion
of the work force relative to the growth
of value-added by spinning off relatively
labor-intensive
activities to relatively
with weights given by distributive shares (see Aoki

1984a, ch. 6; 1988a, ch. 5 for a proof. As is well
known, the Nash bargaining solution is the only one
outcome that satisfies the set of axioms that John
Nash (1950) imposed on the efficient and "fair" (symmetric) arbitration to fulfill. Also, the recent development of game theory showed that, under certain conditions, an equilibrium of a noncooperative twoperson bargaining game, known as the "perfect subgame equilibrium," exhibits qualitatively equivalent
characteristics with the generalized Nash bargaining
solution (Ken Binmore, Ariel Rubinstein, and Asher
Wolinsky 1986). Although the idealized construct
given above may appear arbitrary at first, its behavioral implications may well stand up to variant institutional assumption.

22
lower wage subsidiaries or outside suppliers, 12 as well as leaning more toward capital-intensive technology than the stockholder-controlled firm (Aoki 1988a, pp.
166-74; Hajime Miyazaki 1984).
4. If the implicit unemployment insurance premium payable by employees is
high, the dually controlled firm chooses
work sharing rather than layoffs as a first
response to bad business conditions (Aoki
1988a, pp. 176-81).
5. The dually controlled firm seeks innovative opportunities by developing an
in-house knowledge base rather than
pursuing breakthrough innovation requiring an entirely new organization of
its research and development team (Aoki
1988a, pp. 237-52; Aoki and Rosenberg
1989).
Theoretically speaking, the dually controlled firm may be viewed as a mixture
of the conventional neoclassical model
(the N-model) of the stockholder-controlled firm and the model of the workercontrolled firm (the W-model) in the
manner of Domar-Ward (Aoki 1984a, ch.
5). It is well known that the worker-controlled firm tends to limit the size of the
labor force in order to increase the probability of job security in comparison to the
conventional N-model (Miyazaki 1984).
The behavioral characteristics of the Wmodel may help us understand propositions (2) through (4) intuitively. Proposition (1) appears to run counter to this
characteristic, but note that this proposition is stated in comparison to a firm that
chooses a growth rate solely to maximize
its stock price after having made its wage
bargain. Efficiency requires the conjoint
decision of wage rate and growth rate
when employees become assets internal
to the network, because employees may
12
See the Appendix for the structure of differential
earnings observed between the parent firm and its
satellite firms. As described in the Appendix, however, the use of a lower wage is only one aspect of
subcontracting among many others.
be willing to forgo the current earnings

level for future benefits made possible
by promotion in the rank hierarchy.
Proposition (5) may be understood by
considering that engineers and researchers who have firm-specific knowledge are
constituent members of the dually controlled firm.
The characterizations above may seem
to imply that employees are the only beneficiaries of the dually controlled firm.
But this is not so. As employees in the
dually controlled firm may be induced
to trade off the level of current earnings,
make investment in training, and commit
to the higher level of effort for those benefits indicated above, the profit level is
expected to rise as well. In other words,
once employees become network-specific assets, mutual commitments of employees and management would yield a
Pareto superior outcome.
V. Concluding Remarks
In preceding sections, I described a
model of a Japanese firm based on stylized facts, the essence of which is summarized in the three Duality Principles.
This model-the J-model-is in many respects different from models of the firm
constructed by Western economists. Archibald described the current state of the
theory of the firm in his contribution to
The New Palgrave: A Dictionary of Economics: "It is doubtful if there is yet general agreement among economists on the
subject matter designated by 'theory of
the firm,' on, that is, the scope and purpose of the part of economics so titled"
(1987, p. 357). It would be fair to say,
however, that agency theory is currently
one of the most influential theories on
the firm, especially among Anglo-American theoretical economists.
According to this theory, the firm is
conceived as a "nexus of (agency) contracts" (Michael Jensen and William


Meckling 1976). In an agency contract,
the entity called the principal delegates
decision making for realizing its own objective to the agent who may have superior on-site information, but different
preferences. The principal tries to control the latter's action by the design of
an appropriate incentive contract. The
ultimate principal of the firm is its owner,
or the stockholders in the context of the
modern corporate firm, and its agent is
management. Management is then conceived to operate hierarchically through
a chain of incentive contracts, with management of the higher level acting as the
surrogate for the ultimate principal and
that of the lower level as the agents of
higher-level management. At the bottom
of the hierarchy, management controls
operating employees through incentive
contracts.
An agency contract may be written in
many ways, but its design is conditioned
by outside markets in one important way:
the principal cannot induce the agent to
enter a contractual relation unless the
principal guarantees the agent at least
the level of its reservation utility determined by outside opportunities. Finally,
the stockholders' rights to control management are market-transferable so that
the "bidding" among investors will ultimately lead to the maximization of firm's
value subject to the inevitable costs incurred in the chain of agency contracts
("the agency loss").
Thus the essential factors of the agency
model of the firm are summarized as: (1)
hierarchical decomposition
of control
originating at stockholders (H-mode); (2)
market-conditioned incentive contracting; and (3) the control of the management decision according to the value
maximization criterion. Compare these
with the three Duality Principles for the
J-model. Between them, clear differences are evident. Why are there the
differences? Is that because the J-model
23
is culturally unique and useful only as a

tool for a microanalytic understanding of
the Japanese economy?
One of reasons why many Anglo-American economists are comfortable with the
agency model as the "model of the firm"
and why I am presenting the J-model
as a tool for understanding the workings
of the Japanese economy is doubtless that
there are differences in the ways that
firms are run in the West and in Japan
and that the models reflect some aspects
of those real differences. But, are these
differences absolute? Are they more important than the possible commonality
that may not be taken into account by
either of the models? If there is a convergent trend between the West and Japan,
does it not mean that the J-model and
the agency model represent only prototypes to be absorbed into a more general
hybrid model of the firm?
The primitive comparative analysis of
the H-mode versus the J-mode of coordination summarized in Section I indicated
that the relative efficiency of these two
prototype models depends on various environmental parameters such as defining
the nature and volatility of consumer demands, the degree of market concentration, the technology involved in the
production process, and possibly government regulation. Therefore if only efficiency matters (and if relevant government regulations are alike across national
economies), we would observe different
coordination patterns across markets. In
spite of the increasing globalization of
markets the fact that we have been observing a relatively similar coordination
mode within each economy, but relatively dissimilar patterns in the West and
Japan, may have to do with historical,
cultural, and regulation factors. As indicated just before the description of the
first Duality Principle, the maintenance
of organizational integrity in the context
of individualistic values in the West (par-

24
ticularly in North America) may have necessitated contractual agreement on the

more hierarchical structuring of internal
coordination. On the other hand, in Japan, respect for differentiated status by
attributes (sex, age, seniority, family
background, etc.) and level of training
has been a dominant traditional social
value. A superior in Japan may therefore
be more comfortable in delegating actual
decision making to his subordinate. Also
within a small group, horizontal coordination rather than clear job demarcation
tends to emerge spontaneously in Japan,
possibly because of the collective memory of the traditional agrarian customs
and values (Aoki 1988b).
Having admitted that there are some
cultural and historical traits in the ways
that firms operate in each economy and
that the efficiency criterion is not the only
factor shaping the ways that firms are
run, however, I would maintain that
there is an important element of conscious design in viable business organizations. For example, small group dynamism per se, to which cultural
anthropologists attribute the role of a
driving force in Japanese organization
(e. g., Chie Nakane 1970), cannot be effective in the context of large organizations. A coherent, self-centered group
may develop and assert its own interests
at the sacrifice of organization goals.
Managements of Japanese firms have
taken pains to combat such tendencies
by consciously designing intergroup coordinational mechanisms (the kanban
system is but one example), shifting the
emphasis from seniority to merit acquired by experience as a promotion criterion. They have transformed the senrank hierarchies into
iority-oriented
forms compatible with an organizationwide competitive drive, and so on. The
sharing of rents and the commitment to
employees' interests in corporate policy
making are no longer considered an ex-
pression of paternalistic benevolence of

the management or owner, but can be
regarded as a means to elicit employees'
cooperation and diligence. Many elements of the J-model should now be regarded as serious objects of economic
analysis, particularly in view of Japanese
industrial and technological challenges
on the global scale.
As indicated in preceding sections, the
relative merits of horizontal versus hierarchical coordination, market-oriented
incentive contracting versus rank hierarchy, bank-oriented versus marketoriented financial control are not yet so
clear-cut, however, and comparative
analysis dealing with such issues has only
just begun. Meanwhile, there is a greater
tendency toward a convergence of organizational form and practice because of
the strong force of natural selection operating through international market competition as well as deregulation within
and across national boundaries.'3 Phenomena similar to some aspects of Jmodel have emerged in the West spontaneously or as a result of conscious design,
while some elements of the agency and
other contractual modeling are becoming
ever more visible in Japan. From this
angle, the J-model may provide a new
analytical insight into the working of
newly emergent-or latent-phenomena
in the Western economy. And similarly,
the agency model may be helpful for understanding some aspects of Japanese organization. But, in the future the J-model
is perhaps fated to be subsumed under
the yet to be developed general theory
of the firm, and so is the agency model.
APPENDIX
Quasiintegration:The SupplierRelation In the manufacturingindustry, which involves many steps in a

production process, Japanese firms tend to spin off
subsidiariesand rely upon subcontractorsfor the sup13 See footnotes 4 and 8 for some details concerning
this trend.


ply of parts and intermediate products so that the
degree of operating integration is relatively lower.
Because the prime manufacturer relates to these satellite firms in various forms of ownership ranging
from full ownership of subsidiaries to the minority
holding of independent firms, their transaction relationship is long-lived. Also, the prime manufacturer
tends to procure its supplies directly from a relatively
small number of suppliers, while indirectly related
to a larger number of suppliers through a hierarchy
of suppliers. 14
It has been asserted that this lower degree of integration is due to the late development of industrialization and the consequent shortage of capital at the
disposal of primary firms. But this assertion is refuted
by the fact that the degree of outside sourcing has
been steadily increasing in the last two decades when
capital supply has become more abundant. It has
also been asserted, and is still widely accepted outside of Japan, that the large prime manufacturer utilizes outside suppliers of weaker market power as a
business cycle buffer, withdrawing supply contracts
when demands for final products slacken. This theory
is referred to as the "dual structure" hypothesis, as
it presupposes an industrial structure of two layers,
with exploiting, monopsonic firms on the top and
exploited, weak subcontractors at the bottom (e.g.,
Takafusa Nakamura 1981, ch. 5).
Recent articles by Asanuma (1985, 1989) on the
automobile industry and electric machinery industry
have shown convincingly, however, that there is a
clear demarcation of job tasks between the prime
manufacturer and the suppliers and that the supply
contract for a particular part of a particular model
of the final product is given exclusively to a particular
supplier and normally extends over the entire life
there is usually more
cycle of the model-although
than one supplier that has contractual relations with
the same prime manufacturer for the same parts but
for different models. Thus the suppliers that have
long-run relations with the prime manufacturer act,
in this regard, like in-house divisions of the integrated firm of the West. But, the suppliers have considerable autonomy in other respects, for example,
in personnel administration, an important aspect of
Japanese firms. We may therefore describe the longterm supplier relation simply as "quasi integration."
The supply price is revised at regular intervals,
say twice a year, by bilateral negotiation, incorporating both risk sharing and incentives for innovation
and effort (Asanuma 1985). There are trade-offs between these two functions: Complete insurance for
the risk of cost fluctuation requires full-cost pricing,
whereas fixed-cost pricing yields gains from cost reduction effort to the innovating supplier. Kawasaki
and McMillan (1987) have shown econometrically
that the smaller the supplier, the more risk is absorbed by the prime manufacturer in the automobile
industry. This refutes the orthodox dual structure
hypothesis, but seems to be consistent with the predictions of economic theory, because the larger man14
See Aoki 1988a, ch. 6.1, for detailed data.
25
ufacturer seems to have a better capacity for risk

absorption through product diversification and financial capability. Obviously the prime manufacturer
would perform an insurance function only in exchange for an insurance premium. As the supplier
gains more unique technological expertise, therefore,
he tends to share more business risk and appropriates
the effects of his own cost reduction effort.
Reasons for quasi integration may be many, but
one of them is related to the coordination characteristics of Japanese firms discussed in Section I. Costs
of intrafirm horizontal coordination would become
greater as the size of a network gets larger and the
technical heterogeneity of constituent units of the
network increases. Hence Japanese firms are motivated to spin off heterogeneous activities and keep
the internal network simpler and smaller. In this
way, Japanese firms can minimize the need for administrative personnel for planning and monitoring.
Direct horizontal coordination for the daily supply
of parts between the relevant workshop of the prime
manufacturer and the supplier as well as the assumption of responsibility for routine problem solving by
the supplier may contribute to the kind of informational efficiency attributed to horizontal coordination
as discussed in Section I. That is, quasi integration
can save inventory costs in continually changing market conditions, but may not be particularly effective
in highly stable markets and may be vulnerable to
shocks in drastically changing markets.
It is noteworthy that the supplier of important parts
participates in the development processes of the
prime manufacturer at an early phase and the design
and pilot fabrication of new parts are done in parallel
to corresponding activities at the prime manufacturer. This is considered to shorten the lead time
necessary for the development of a new model (Imai,
Nonaka, and Takeuchi 1985). In the development
case referred to in example (3) in Section I, engineers
from important supplier firms are housed in the development laboratory of the prime manufacturer as
"guest engineers" and participate in the SED team
on a permanent basis.
Technical and technological expertise of the supplier connected to the prime manufacturer through
horizontal operational coordination based on the kanban system and joint development effort is at least
partially relation-specific for the same reason as skills
of employees are network-specific. It is then anticipated and in fact observed that a group of suppliers
related to one prime manufacturer directly or indirectly are hierarchically layered according to the degree of the supplier's relation-specific (network-specific) expertise and the length of their tenure with
the prime manufacturer. The suppliers, hierarchically layered, are likely to participate in differential sharing in the value created by the group as a
whole according to their ranks. This aspect of grouping may appear reminiscent of the orthodox dual
structure hypothesis, but distributional differentiation is now gradational rather than dually layered.
Suppliers that supply unique products to the prime
manufacturer seem to have considerable bargaining
power over the distribution of group rents through

26
price negotiation. Also this is the class of suppliers

that participates in joint development efforts with
the prime manufacturer. Thus the supplier relation
of Japanese firms seems to be characterized by weakdecision hierarchy cum incentive-ranking hierarchy
(WDIR) just like other aspects of Japanese firms.
ANTHONY. "Worker Management and the

Modern Industrial Enterprise," Quart. J. Econ.,
Aug. 1973, 87(3), pp. 375-92.
BERGL6F,
ERIK. "Capital Structure as a Mechanism
of Control-A Comparison of Financial Systems,"
in Thefirm as a nexus of treaties. Eds: MASAHIKO
ATKINSON,
Bo GUSTAFSSON,
AOKI,
SON.
REFERENCES
E. WILLIAM-
AND OLIVER
London: Sage Publications, 1989, pp. 237-
62.
ABEGGLEN,
JAMES C. AND STALK, GEORGE,
JR.
Kai-
sha, the Japanese corporation, NY: Basic Books,

1985.
"An 'Incomplete Contract' Approach to Bankruptcy and the
Financial Structure of the Firm," MIT Department
of Economics Working Paper, No. 484, 1988.
ANDO, ALBERT AND AUERBACH, ALAN J. "The Cost
of Capital in the United States and Japan: A Comparison," J. of the Japanese and International
Economies,June 1988, 2(2), pp. 134-58.
AGHION, PHILIPPE AND BOLTON, PATRICK.
AOKI,
MASAHIKO.
The Co-operativegame theory of
thefirm. Oxford: Oxford U. Press, 1984a.

. "Shareholders' Non-Unanimity on Investment Financing: Banks vs. Individual Investors,"
in The economicanalysisoftheJapanesefirm.. Ed.:
M. AOKI.Amsterdam: North-Holland, 1984b, pp.
193-224.
. "Horizontal vs. Vertical Information Structure of the Firm," Amer. Econ. Rev., Dec. 1986,
76(5), pp. 971-83.
. Information, incentives, and bargaining in
the Japanese economy. NY and Cambridge: Cambridge U. Press, 1988a.
. "Decentralization-Centralization in Japanese
Organization: A Duality Principle," Mimeo., U.
of Kyoto, 1988b. (To appear in The political economy of Japan, Vol. 3. Eds.: SHUMPEI KUMON
AND HENRY
Stanford: Stanford U.
RoSOVSKY.
Press.)
. "The Participatory Generation of Information Rents and the Theory of the Firm," in The
firm as a nexus of treaties. Eds.: MASAHIKO AOKI,
Bo GUSTAFSSON, AND OLIVER E. WILLIAMSON. London: Sage Publications, 1989, pp. 26-51.
KAZUO AND NAKATANI,
AOKI, MASAHIKO; KOIKE,
IWAO. Nihon KigyoGurobarukano Kenkyu:Jyoho-
Sisutemu, R&D, Jinzai Ikusei (The globalization

of the Japanesefirm). Tokyo: PHP, 1989.
NATHAN. "The Japanese Firm as an Innovating Institution," in Economicinstitutionsin a dynamicsociety. Eds. TAKASHI SHIRAISHI
AND SHIGETO
TSURU.
London:
Macmillan Press, 1989, pp. 137-54.
ARCHIBALD, G. C. "The Theory of the Firm," in The
new Palgrave:A dictionary of economics, Vol. 2.
London: Macmillan, 1987, pp. 357-63.
ASANUMA,
BANRI. "The Organization of Parts Purchases in the Japanese Automotive Industry," Japanese Econ. Studies, Summer 1985, 13(4), pp. 3253.
. "Manufacturer-Supplier Relationships in Japan and the Concept of Relation Specific Skill,"
AOKI, MASAHIKO AND ROSENBERG,
J. of the Japanese and International Economies,

Mar. 1989, 3(1), pp. 1-30.
B.
BERNHEIM,
AND
DOUGLAS
SHOVEN,
B. "Com-
JOHN
parison of the Cost of Capital in the U. S. and Japan:

The Role of Risk and Taxes," Mimeo., Stanford
U., 1989.
BINMORE,
KEN;
ARIEL
RUBINSTEIN,
AND
WOLINSKY,
"The Nash Bargaining Solution in Economic Modelling," RandJ. Econ., Summer 1986,
17(2), pp. 176-88.
ASHER.
RICHARD E. AND UEKUSA, MASU. Industrial

organizationinJapan. Washington, DC: Brookings
CAVES,
Institution, 1976.
ROBERT AND YAKUSHIJI, TAIZO. TheAmnerican
and Japanese auto industries in transition. Ann
COLE,
Arbor: Center for Japanese Studies, U. of Michigan, 1984.

CREMER, JACQUES. "Common Knowledge and the
Coordination of Economic Activities," in Thefirm
as a nexus of treaties. Eds: MASAHIKO AOKI, Bo
GUSTAFSSON, AND OLIVER E. WILLIAMSON. London:

Sage Publications,
1989, pp. 53-76.
CUSMANO, MICHAEL A. TheJapanese automzobile
in-
dustry:Technologyand managementat Nissan and

Toyota. Cambridge: Harvard U. Press, 1985.
DERTOUZOS,
SOLOW,
L.;
MICHAEL
ROBERT
LESTER,
K.
RICHARD
AND
Made in America. Cambridge:
M.
MIT Press, 1989.

RONALD. British factory, Japanese factory:
The origins of national diversityin industrial relations. Berkeley: U. of California Press, 1973.
FOULKES, FRED. Personnelpoliciesin large nonunion
companies. Englewood Cliffs, NJ: Prentice-Hall,
DORE,
1980.
FREEMAN,
RICHARD
B.
AND
WEITZMAN,
L.
MARTIN
"Bonuses and Employment in Japan," J. of the

Japanese and InternationalEconomies,June 1987,
1(2), pp. 168-94.
Fuss,
MELVYN
A. AND
WAVERMAN,
"Pro-
LEONARD.
ductivity Growth in the Automobile Industry,

1970-1980: A Comparison of Canada, Japan and
the United States." National Bureau of Economic
Research Paper, No. 1735, 1985. (Forthcominig in
Productivitygrowth in the United States and JaAND

pan. Eds.: C. HULTEN
J. R. NORSWORTHY.
NBER Studies in Income and Wealth, Chicago:
U. of Chicago Press.)
HASHIMOTO,
MASANORI
AND
RAISIAN,
JOHN.
"Employ-
ment Tenure and Earnings Profiles in Japan and

the United States," Amer. Econ. Rev., Sept. 1985,
75(4), pp. 721-35.
HATSOPOULOS,
N.;
GEORGE
KRUGMAN,
PAUL
R.
AND
H. "U.S. Competitiveness:
Beyond the Trade Deficit," Science, July 15, 1988,
241, pp. 299-307.
SUMMERS,
HORIUCHI,
SHINICHI.
LAWRENCE
AKIYOSHI;
PACKER,
FRANK
AND
FUKUDA,
"What Role Has the 'Main Bank' Played

27

in Japan?"J. of the Japanese and International
Economies, June 1988, 2(2), pp. 159-80.
HiAND TAKEUCII,
NONAKA,
IKUJIRO
"Managing the Product Development Process: How Japanese Companies Learn and Unlearn," in The uneasy alliance managing the
IMAI,
KENICHI;
ROTAKA.
dilemma.Eds: KIM
productivity-technology
CLARK
Boston: Harvard Business School Press,

1985, ch. 8.
ET AL.
ITOH,
Processing
HIDESEII. "Information
Capacities
of the Firm," J. of the Japanese and International

Economies, Sept. 1987, 1(3), pp. 299-326.
JENSEN,
MICHAEL
AND
MECKLING,
of the Firm: Managerial
"Theory
WILLIAM.
Behavior,
Agency
Costs
and Capital Structure," J. Finan. Econ., Oct. 1976,
3(4), pp. 305-60.

JOHN.
"The DeSEIICEII
AND MCMILLAN,
sign of Contracts: Evidence from Japanese Subcontracting," J. of the Japanese and International
Economies, Sept. 1987, 1(3), pp. 327-49.
KAWASAKI,
STEPIHEN
KLEIN,
AND
ROSENBERG,
"An
NATHAN.
Overview of Innovation," in The positivesum stratNATHAN

AND
ROSENegy. Eds.: ERALPH LAUDAU
DC: National Academy Press,
BERG. Washington,
1986, pp. 275-305.

KOIKE, KAZUO. "SkillFormation Systems in the U.S.
and Japan: A Comparative Study," in The economic
analysis of the Japanesefirm. Ed.: M. AOKI. Amsterdam: North-Holland, 1984, pp. 47-75.
. Understandingindustrial relations in modern Japan. London: Macmillan Press, 1988.
______"Intellectual Skill and the Role of Employees
as Constituent Members of Large Firms in Contemporary Japan," in Thefirm as a nexus of treaAND
ties. Eds.: MASAHIKO AOKI, Bo GUSTAFSSON,
E. WILLIAMSON.
London: Sage PublicaOLIVER
tions, 1989, pp. 185-208.

KOMIYA, RYUTARO. "Japanese
Firms, Chinese
Firms:
Problems for Economic Reform in China. Part II,"
J. of the Japanese and International Economies,

June 1987, 1(2), pp. 229-47.
. "Structural
of
and Behavioristic
Characteristics
the Japanese Firm" (in Japanese in Gendai Chu-
goku Keizai:Nicchu no Hikaku Kosatsu (The contemporaryChinese economy:A comparativestudy

of China and Japan). Tokyo: U. of Tokyo Press,
1989, ch. 3, pp. 97-145.
LEIBENSTEIN,
Inside the firm. Cambridge:
HARVEY.
Harvard U. Press, 1987.

LINCOLN,
JAMES;
HANADA,
MITSUYO
AND
MCBIRD,
KERRY. "Organizational Structures in Japanese and

U.S. Manufacturing,"
Administrative Science
Quarterly, Sept. 1986, 31(3), pp. 334-64.
MACLEOD,
BENTLEY
W.
AND
MALCOMSON,
JAMES
M.
"Reputation and Hierarchy in Dynamic Models of

Employment." Mimeo., U. of Southampton, 1986.
MAILATII,
GEORGE
J. AND
Information
"Asymmetric
lems with Many Agents."
vania, 1988.
POSTLEWAITE,
ANDREW.
and Bargaining
ProbMimeo., U. of Pennsyl-
EDWIN. "Industrial R&D in Japan and

the United States: A Comparative Study," Amer.
Econ. Rev. May 1988, 78(2), pp. 223-28.
MANSFIELD,
MINCER,
JACOB AND HIGUCHI,
YOSHIO. "Wage
Struc-
tures and Labor Turnover in the United States

and Japan," J. of the Japanese and International
Economies,June 1988, 2(2), pp. 97-133.
MIYAZAKI, HAJIME. "Internal Bargaining, Labor Contracts, and a Marshallian Theory of the Firm,"
Amer. Econ. Rev., June 1984, 74(3), pp. 38193.
MONDEN,
YASUHIRO. Toyota production system. Atlanta: Industrial Engineering and Management
Press,
MOWERY,
1983.
DAVID
C.
AND
ROSENBERG,
NATHAN.
"Com-
mercial Aircraft: Cooperation and Competition Between the U.S. and Japan," Calif. Manage. Rev.,
Summer 1985, 27(4), pp. 70-92.
NAKAMURA,
TAKAFUSA.
The postwarJapanese econ-
omy. Translated by JACQUELINE KAMINSKI. Tokyo:

U. of Tokyo Press, 1981. (Originally published as
Nihon keizai: Sono seicho to kozo, 2nd ed. Tokyo:
U. of Tokyo Press, 1980.)
CHIE.
NAKANE,
Japanese society. Berkeley and Los
Angeles: U. of California Press, 1970.
NASH,
JOHN.
"The Bargaining Problem," Econometrica, Apr. 1950, 18(2), pp. 155-62.
PASCALE,
AND
RICHARD
ROlILEN,
THOMAS.
"The
J. of Japanese Studies, Summer 1983, 9(2), pp. 219-63.

Mazda Turnaround,"
E. Thesecond
MICHAEL
AND SABEL,
CHARLES
industrial divide. NY: Basic Books, 1984.
PIORE,
PAUL. "The Main Bank System and Corporate Monitoring and Control in Japan," J. Econ.
Behav. Organ., May 1989, 11(3), pp. 399-422.
SHEARD,
SHIMADA,
HARUO
AND
MACDUFFIE,
JOHN
PAUL.
"In-
dustrial Relations and 'Humanware'." Alfred P.

Sloan School of Management Working Paper, No.
1855-88, 1986.
SHIRAI, TAISEIIRO. "A Theory of Enterprise Union,"
in Contemporary industrial relations in Japan.
Ed.: TAISHIRO SHIRAI. Madison: U. of Wisconsin
Press, 1984, pp. 117-43.
TACIHIBANAKI, TOSIIAKI.
"The Determination
of the
Promotion Process in Organizations and of Earnings Differentials," J. Econ. Behav. Organ., Dec.
1987, 8(4), pp. 603-16.
TAYLOR, JOHIN B. "Differences in Economic Fluctuations in Japan and the United States: The Role of
Nominal Rigidities," J. of the Japanese and International Economies, June 1989, 3(2), pp. 12744.
MARTIN.
The share economy. Cambridge:
WEITZMAN,
Harvard U. Press, 1984.
D. ELEANOR AND SAKAKIBARA,
WESTNEY,
KIYONORI.
"Comparative Study of the Training, Careers, and
Organization of Engineers, in the Computer Industry in Japan and the U.S." Mimeo., MIT, 1985.
OLIVER E. The economic institutions
of capitalism. NY: Free Press, 1985.
WILLIAMSON,


Aoki Firma Japoneza

Uploaded by

Copyright:

Available Formats

Aoki Firma Japoneza

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Aoki Firma Japoneza

Uploaded by

Copyright:

Available Formats

American Economic Association

Toward an Economic Model of the Japanese Firm

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Journal of Economic Literature

THE LAST DECADE

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Journal of Economic Literature, Vol. XXVIII (March 1990)

Stalk (1985, ch. 8), Aoki (1984a, 1988a),

because the Japanese model is culturally

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Aoki: Toward an Economic Model of the Japanese Firm

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Journal of Economic Literature, Vol. XXVIII (March 1990)

rates" in spite of the apparent cost efficiency of Japanese producers. Because

the said market condition, such internal

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Aoki: Toward an Economic Model of the Japanese Firm

tralized indicative monthly plan has been

2. Quality Control-A Steel Industry

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Journal of Economic Literature, Vol. XXVIII (March 1990)

horizontal coordination among related

vidual workshops. In the actual solution

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Aoki: Toward an Economic Model of the Japanese Firm

responsible for the manufacturing phase

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Journal of Economic Literature, Vol. XXVIII (March 1990)

sharing of ex post on-site information

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Aoki: Toward an Economic Model of the Japanese Firm

process innovation, as exemplified in the

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Journal of Economic Literature, Vol. XXVIII (March 1990)

II. The Hierarchy of Ranks

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Aoki: Toward an Economic Model of the Japanese Firm

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Journal of Economic Literature, Vol. XXVIII (March 1990)

30s). So, the fast flyer among blue-collar

archy to operate as an effective incentive

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Aoki: Toward an Economic Model of the Japanese Firm

Thus there seems to be an interesting

The First Duality Principle: In order for

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Journal of Economic Literature, Vol. XXVIII (March 1990)

ganizationally effective, either their coordination or their incentive mode needs

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Aoki: Toward an Economic Model of the Japanese Firm

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Journal of Economic Literature, Vol. XXVIII (March 1990)

Bolton argue that the essence of debt

of takeover).8 This observation suggests

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Aoki: Toward an Economic Model of the Japanese Firm

device becomes fully operative only if the

This content downloaded from 86.35.136.233 on Mon, 21 Oct 2013 04:00:52 AM

Journal of Economic Literature, Vol. XXVIII (March 1990)