The Competitiveness of Nations
in a Global Knowledge-Based Economy
H.H. Chartrand
April 2002
Michael Storper
The Limitations to Globalization: Technology Districts
and International Trade
Economic Geography
Volume 68, Issue
1
Jan. 1992,
60-93.
Index
Abstract
Six Propositions on Trade, Flexibility, Technology, and
Regional Development
Export Specialization and Technological
Dynamism
The Increase in Trade
Specialization
Trade and Product-Based Technological
Learning
An Historical Divide?
Developmental Effects: PBTL versus the
Rest
Specialization in Three Countries
Technological Learning and the Organization of
Production
Technology, Evolution, and Increasing
Returns
Problems with Path-dependence and Lock-in: The Division
of Labor
Technological Oligopolists and Production
Networks
The Regional Basis of Technological
Learning
The Global Economy as a Mosaic of
Regions
Networks and Geographical
Agglomerations
Learning and Regional Context: The Qualitative
Specificity of Externalities
The Territorialization of Learning: Regions and
Countries
Flexibility, Technology Districts, and the World
Economy
Flexible Production as a Technological
Trajectory
The Technology District as a Particular Form of the
Industrial District
The Limits to Globalization
References
Technological
Learning and the Organization of Production
To understand long-term PBTL specializations, we must
know what gives rise to repeated innovations in a particular sector or cluster
of related sectors, such that the specializer continually redefines best
products.
Technology, Evolution, and Increasing
Returns
Conventional economic theory has had little success in
defining what causes technological change; one of the great embarrassments of
modern economics is “Solow’s residual” 3 When it comes to defining the
direction and shape of technological change, the theory is also rather
unhelpful; it can identify the types of change likely to be inviable, but it
cannot predict the concrete shape and content of change. Indeed, new technologies often violate
classical conditions, in that they do not economize on relatively expensive
factors of production or serve preexisting preferences (they are not
“induced”).
One simple reason is that product and process change are
mutually interdependent, and the former is extremely unlikely to obey classical
conditions. Both types of change,
moreover, have been shown to produce their own paths of development cumulatively
as much as to respond to any given set of signals in the environment. This is the fundamental insight of the
new evolutionary economics. For our
purposes, the argument can be summarized in a few basic propositions (see Foray
1990; Dosi et al. 1988):
(1) Technological change (by which we mean not only
hardware, but skills,
3. This refers to Solow’s (1957) finding that after all
other possible sources of technological change and economic growth were
examined, a huge residual remained, which awakened economists to the difficulty
of understanding technological change, as well as its fundamental importance in
the growth process.
77
cognitive frameworks, and organizational practices) is
subject to increasing, not decreasing returns, and as a result development and
adoption tend to be characterized by positive feedback.
(2) One source of positive feedback is
learning-by-using, which diffuses and deepens the cognitive framework associated
with a given product or process and makes it ever easier to
use.
(3) Another source is external economies of scale in
use; the more independent users (i.e., in the sense of a network of users), the
more difficult it is to change the basic nature of a product or process, for
such a change wipes out the physical capital (and possible add-ons) of users, as
well as their experience in learning-by-using.
(4) External economies in use get built in by
technological complementarities, between base technologies and diverse
applications, and between linked applications.
(5) External economies come from the increasing scale of
production, which lowers costs and which follows from the dynamics described
above.
(6) As a result, after a certain point in the history of
a technology, its production and use tend to become increasingly locked in,
i.e., irreversible or relatively resistant to fundamental
change.
(7) The paradox is that there is no way to tell, at the
outset, what technologies of use or production will be locked in, and no reason
to believe that what does get locked in is the most efficient of all possible
worlds. Efficiency is itself a
property of adoption and use, and together with lock-in, it is path dependent or
endogenous to use and adoption (Nelson and Winter 1982).
(8) Indeed, along the way, the (medium- or long-run)
technology that is locked in, and thus the efficiency (and input history) of
production of the technology, may be “selected” by small events and even
accidents, which have long-run and large-scale consequences: the “first mover”
effect. This implies not that
anything will do, but that multiple acceptably efficient paths are frequently
open and their selection is not preinscribed by any global
rationality.
This reasoning holds for hardware and “soft”
technologies and goods; in the former case, the lock-in has to do with sunk
capital, whereas in the latter, it has to do with reputation and brand name for
a good. In both, intangible capital
in the form of skills and knowledge is subject to strong external economies of
scale.
Problems with Path-dependence and Lock-in: The Division
of Labor
Lock-in is problematical for the same reason that
it may be beneficial to producers and users: it is more difficult to
change from one path to another as external economies in production or use
increase, and it may even become difficult to move further along a given,
narrowly defined product path if production technologies become specific. Yet it is precisely the logic of cost
minimization to make resources increasingly specific and dedicated to the
production of a given output, i.e., to lock in the producers. This subjects them to competition from
other cost minimizers, on the one hand, or to the risk that other producers who
are not locked in will come along with better products or more efficient
technologies, as appears to have happened with the American steel and automobile
industries in the 1970s. Technological dynamism thus involves
compromise between the goals of cost minimization in production and
reversibility in production and product design; on the one hand, to minimize
lock-in to products while keeping production costs down, and on the other, to
maximize the possibility and minimize the cost of mobility to new products and
production technologies.
As a number of recent theoretical works have shown
(Foray 1990; Amendola and Gaffard 1990), the optimal organizational form for a
technologically dynamic production system is that of a production network,
involving a significant degree of vertical (and probably, horizontal)
disinter-
78
gration of the production system, whether
interorganizationally (independent firms) or intraorganizationally (between
units of production within a firm) (Cooke and Morgan 1991). The concept of network is, admittedly,
general; in reality networks may take many different concrete forms. As opposed to the
Galbraithian/Chandlerian command-and-control firm with its dependent
subcontractors, however, the network has a distinctively different economic
logic and organizational form (Chandler 1966; Galbraith 1967; Powell
1990).
The production network reduces the risk of lock-in
entailed by vertical integration and asset specificity. A production network is neither a firm
(hierarchy) nor a system of market transactions (the latter is simply a
“production complex”); it is, instead, a set of units joined in relatively
durable relationships through an “organized market” involving some degree of
cooperation or, at least, symmetry of power relations (Lundvall 1990; Powell
1990; Storper and Harrison 1991). Some units of a network do not have
general-purpose assets that permit them to sell on open markets, nor do they
have assets that are specific (in the sense that they require vertical
integration with other specific assets, as in the reasoning employed by
Williamson ); rather, their assets are specialized to a given range of
activities, involving particular competences on the parts of firms or units
(Salais and Storper 1991). Taken as
a whole, a production network compromises between the need for cost minimization
(either by including some units with high internal scale economies for certain
operations or by aggregating a number of producers in a given sector and giving
them the benefits of external economies of scale in place of internal economies
of scale), and the need to avoid lock-in by substituting externalized and
specialized for internalized and specific resources. On the positive side, specialized
producers can shift along a path of technological evolution, or even between
paths from time to time, by minimizing asset specificity, while forms of
cooperation between specialized producers permit joint technological development
relying on the high levels of competence of the
specialists.
The precise shape of the division of labor varies
according to the particular industry and its possible paths of development.
This is the question of the
coherence of firms and production systems. It is defined by many things, which
include the extent to which product and process technologies are converging or
diverging, whether they are similar or dissimilar, and whether the quasi-rents
from knowledge are appropriable or nonappropriable (Dosi and Salvatore 1992).
At one extreme, some networks will
be highly vertically disintegrated, with many specialized producers, few large
units, and little hierarchy of governance; at another, they might involve some
specialized producers linked into systems of financially integrated,
strategically controlled, and large production units, with a steep governance
hierarchy in evidence. This is the
difference between, say, the networks of design-oriented firms in the Third
Italy and the aerospace industry in
The reasoning employed here differs in two key respects
from that of the New Institutional Economics of Coase (1937) and Williamson
(1985). In the latter body of
theory, the existence of interfirm transactions is a choice of market relations
over internal firm hierarchies in pursuit of a cost-minimizing production
technique. In the line of reasoning
pursued here, transactional networks are likely to come about in the context of
technological
79
innovation, and, because of their skill-mobilizing
qualities and avoidance of lock-in, they are seen as key to growth. These networks are fashioned around
non-cost-minimizing behavior (in the form of technological rents) and rapidly
increasing levels of output, which render short-term optimization impossible
(Amendola and Gaffard 1990). Some
of the quasi-rents on products are distributed to suppliers, who possess what
Asanuma (1989) calls “relation-specific skills” from their repeated interaction
with clients and knowledge of their needs; these “relational quasi-rents” in
buyer-supplier transactions are a key economic basis of the external
accumulation of know-how in these networks. As such, and unlike Williamson’s
conception of market transactions, external economies of the network come into
existence, economies that are, for our purposes, positive-sum outcomes of a
transactional production structure (see Storper 1989). 4
In any case, the process with which we are concerned
cannot easily be placed within the Williamsonian theory. The latter, despite its claims to
predictive accuracy, has great difficulty coming to grips with any given set of
empirical outcomes, for many different forces can combine to tug a production
system in different directions, here toward hierarchies, there toward markets.
This is especially true of PBTL.
Other things being equal,
uncertainty should push producers toward vertical integration, unless it is
simple growth (“Stiglerian”) uncertainty; this is all the more true if asset
specificity is high or appropriability is low, for then uncertainty may be
combined with the risk of hold-up from suppliers. But the long-term costs of integration in
the presence of technological uncertainty are likely to push producers toward
vertical disintegration; this is a consequence of what Knight (1921) called
“true” uncertainty, where one cannot even estimate one’s own risk over a given
time period. Moreover, as
McFetridge and Smith (1988) note, the New Institutional Economics seems to
engage in a “tautology... The degree of asset specificity is jointly determined
with the mode of transacting.”
Two observations may be made here. First,
The network is thus the external equivalent of a
Penrosian firm rather than a neoclassical firm. The irony is that one set of external
economies (of the network) is designed precisely to combat the effects of
another (those associated with technological lock-in). We shall see shortly that this gives rise
to a key dilemma for economic development over time.
4. This conception of externalities shares with the
theory advanced by Perroux (1950) the link between a system of interfirm
transactions and innovation, but Perroux never fully elaborated his growth
theory, and he prematurely closed off theorization of “economic spaces” by
defining them too narrowly as consisting of a propulsive (final output) firm or
industry and its backward linkages.
The best early general explanation of the role of external economies is
Young (1928).
80
Technological Oligopolists and Production
Networks
The notion that PBTL has become a centerpiece of
contemporary capitalist competition has recently been incorporated into the
theory of the firm, most effectively by Best (1990), who synthesizes insights
drawn from Richardson (1972), Penrose (1959), and Schumpeter (1934). The central function of the firm in the
“new competition” is no longer as an optimum agency of resource allocation, but
as strategic coordinator of diverse internal and external resources for the
purpose not only of responding to threats, but of creatively generating markets
that have not, hitherto, existed. The amendment made to Best in the
argument presented here is that rather than seeing the firm as the central
analytical object of the flexible production system, I have focused on the
commodity chain. It is true that in
some commodity chains, individual firms play especially important roles as
strategic coordinators or deployers of resources upon which the rest of the
firms and units in the chain depend heavily, but it is also true for a number of
cases that there is simply no dominant firm and the commodity chain is
coordinated by other means (marketing agents, market relations, contracts,
informal routines, etc.). The more
general way to formulate the problem of the technologically dynamic production
system is as a commodity chain and its different possible forms of
governance. The firm is one
such form of governance among many (for an extended discussion, see Storper and
Harrison 1991).
This notion generates significant controversy. Much attention has focused on what we
might call the “technology-based oligopolists,” i.e., the large, well-known
firms that command significant R&D resources and often operate globally,
like IBM, Phillips, or Boeing. Their capacity to develop new products,
the entry barriers to their markets, and their strategic capabilities lead many
to claim that these large firms are the modal form of production organization,
and that this somehow contradicts the notion of the network described above
(Ernst 1990; Martinelli and Schoenberger 1991). This point about control and
organization, so the story goes, is corroborated by the fact that huge shares of
world trade, and of much intraindustry trade, take the form of intra- and
inter-large firm exchanges (some estimates place this share at greater than 50
percent for the U.S.; for example, see Julius 1990).
The role of such large firms was, of course, the
centerpiece of
Such firms are also increasingly linked to other
technological oligopolists through strategic alliances (Mytelka 1990). Some analysts claim that such
inter-large-firm systems effectively account for the majority of technological
competence in the advanced economies and represent a new “internalization” of
innovation (Gordon 1991). If this
were the case, we would expect large firm networks to be
largely
81
indifferent to place and to exclude significant
exchanges with local, small firms. Yet the technology-based oligopolists
carry out high proportions of their core innovational activities in their home
countries (Tyson 1987). This is
important because it suggests the dependence of these firms on external linkages
upstream of the market, i.e., factors that are contained within neither the
firm’s internal structure nor its relationships to major (especially foreign)
technological partners, such as supplies of technologically competent labor.
In addition, the size profiles of
complexes where these technology-based oligopolists are located are diverse and
have frequently exhibited declines in average firm size, as in
Moreover, it seems increasingly to be the case that
these technology-based oligopolists form strategic alliances for a reason
largely ignored by Vernon; they do it to tap into the technological competences
of other large firms, which are in turn rooted in local and national production
networks, or they carry out foreign direct investment to gain direct access to
such competences (OECD 1991). This
logic suggests that behind the statistically observed trade patterns is a high
level of inter-network (and not simply inter-oligopolist)
exchange.
Table 2:
Top Fifty
Table 3:
Top Fifty Italian
Industries Ranked in Terms of World Export Share, 1985
Table 4:
Top Fifty French
Industries Ranked in Terms of World Export Share, 1985
Table 5A:
The Roots of
Export Specialization: PBTL vs. the Rest,
Table 5B:
The Roots of
Export Specialization: Learning/Economics of Variety vs. the Rest,
Table 5C:
The Roots of
Export Specialization: Learning/Economics of Variety vs. the Rest,
France
Table 5D:
Totals for Three
Countries
Table 6:
The Degree of
Country Specialization in HTO, DIC, and PMM Industries,
1985
Table 7:
Technological
Districts in the
