The export specialization industries of the U.S., Italy, and France are not distributed evenly across their national territories. Certain parts of the production chains of these industries are largely responsible for the strong trade positions of each country in that industry, and these parts of the national populations of firms and jobs in the industry are even more highly localized than the industry as a whole.

The existing case study literature suggests that the geography of the specific subsectors (i.e., of the specializations of each country) conforms to popular impressions about the “dynamic” regions of these nations. In the U.S., for example, more than 50 percent of employment in the export specialization industries described earlier can be found in eight states (some small and contiguous). California shows location quotients of greater than 1.2 for clusters of 4-digit sectors in electronics/computers, aerospace, instruments, medical equipment, and motion pictures; Texas in electronics and aircraft; Washington in aircraft and electronic instruments; New York and New Jersey in pharmaceuticals and electronics; and Massachusetts and Connecticut in aircraft and machinery, electronics and computers, telecommunications equipment, precision instruments, medical equipment, R&D, and aerospace and armaments (Table 7). These areas are, in other words, concentrations of employment in export-specialized, learning and variety-based, industries.

In Italy, the design-intensive or craft-based goods are highly concentrated in the “Third Italy,” at whose center lie the two regions of Tuscany and Emilia-Romagna. Lesser concentrations can be found at the fringes of the Third Italy in such places as the Marches, the Veneto, and selected locations in Lombardy. While other local concentrations of activities in similar sectors exist in northern Italy, numerous analysts have shown that they do not in general produce goods of the same quality as those of the Third Italy and are probably not as export-oriented as the latter (see Sforzi 1990).

The geography of French specializations is quite complex and less well known than the other two countries. A few observations may nonetheless be put forward. French high technology specialties in aerospace and defense are concentrated in the Ile-de-France (Paris), in the Midi-Pyrenees (centered on Toulouse), and in the Loire (Nantes); and in telecommunications and electronic components, in the Ile-de-France and Rhône-Alpes

Table 7

Technological Districts in the United States

Source: calculated from data in U.S. Census Bureau, Country Business Patterns

(Isere). In design-intensive or craft-based industries, the French specialties in shoes appear to be concentrated in the Rhône-Alpes region (Roannes, Romans, the Mont du Lyonnais) and fashion clothing in the Ile-de-France and the Vendée (Cholet). Certain other isolated specialties, such as porcelain (Limoges) and perfumes (Cannes), also make their appearance. In advanced services, the only important region for export (of engineering and financial services) is that of Paris/Ile-de-France, with Lyon in second place. The Rhône-Alpes, France’s second industrial region, with a long history of industrialization, is highly specialized in metalworking and machinery industries (the Vallée de l’Arve in the Haute Savoie); other concentrations may be found in the northern part of the French Jura.

Index

Networks and Geographical Agglomerations

Vertical disintegration is known to be positively associated with geographical agglomeration; as the level of external transaction in a production system grow, and to the extent that those transactional have geographical cost-structures, the producers caught up in that division of labor tend to cluster in territorial space in order to reduce the time and cost of transacting (Scott 1988).

Most of the theory produced on this subject to date takes as its basic illustration the case of market uncertainty for producers. Where commands are uncertain, for any reason, various forms of supplier and subcontracting relations arise as a way of minimizing unused capacity in the face of fluctuations. Vertical disintegration and agglomeration are, in these cases, cost-minimizing strategies in the face of uncertainty in much the same way they are portrayed in the New Institutional Economics. Thus, both agglomeration and disintegration can be present in principle, even in cases when the production complex is not a PBLT system, including the case where local uncertainty results from rapid technological change (product and process redesign) entering the locality from elsewhere, thereby forcing local producers to hedge their bets by becoming “best-practice imitators”.

The PBTL production complex is a

special case because the relations between producers and users of technologies (i.e., the parties in the division of labor) are subject to uncertainty that is greater than in the other cases and qualitatively different. User-producer transactions take many different forms: between capital goods producers and their users in production of final outputs; between producers of components and consumers of components; between producers of information and users of such information; between producers of materials and consumers of such materials; and between producers of final products and final market consumers. A given transactional relationship in the presence of learning tends to be qualitatively more dense than in the case of simple market fluctuations because it involves new and unstandardized knowledge. In the case of user-producer relations under conditions of technological change, the difficult and not-easily-objectifiable process of interpretation of evidence and opportunities is critical. Moreover, the whole transactional structure may be subject to redefinition as new types of products and new firms enter the structure and as whole new sub-nodes, channels, and codes of transaction are defined. In other words, where rapid learning is taking place, the transactional structure is likely to involve constant negotiation, renegotiation, and dependence on achieved understandings as the basis of achieving common reinterpretations of new evidence and opportunities. This hypothesis applies not only to incremental innovations, but to radical innovations as well, as:

the codes developed to communicate a constant, or a gradually changing technology will become inadequate. Established producers, following a given technological trajectory, will have difficulties in evaluating the potentials of the new paradigm. Users will have difficulties in decoding the communications coming from producers, developing new products, built according to the new paradigm. In this case, geographical and cultural distance might play an even more important role than in the case of incremental innovation. The lack of standard criteria for sorting out what is the best paradigm, implies that “subjective” elements in user-producer relationships ... will become important (Lundvall 1990, 19).

In technologically dynamic production complexes, then, there is a strong reason for the existence of regional clusters or agglomerations.⁵ Agglomeration appears to be a principal geographical form in which the trade-off between lock-in, technological flexibility (and the search for quasi-rents), and cost-minimization can be most effectively managed, because it facilitates efficient operation of a cooperative production network. Agglomeration, in these cases, is the result not simply of standard localization economies (which are based on the notion of allocative efficiency in minimizing costs), but of “Schumpeterian” efficiencies.

Index

Learning and Regional Context: The Qualitative Specificity of Externalities

Not all agglomerations, even those with a deep vertical division of labor and external economies of scope, are technologically dynamic, and the mere existence of a deep division of labor does not guarantee that a PBTL network is in place. We must look to other conditions, having to do with the qualitative nature of interactions between agents, as keys to their technological dynamism.

A voluminous literature has attempted to identify factors supposedly leading to the creation and success of localized,

5. Learning, flexibility, and agglomeration tend to be interdependent, but an important qualification must be noted. There is nothing deterministic about the geographical form of a cooperative production network. Depending on the nature of the transaction and the history, institutional structure, and geographical distribution of key actors in an economy, agglomeration may be more or less pronounced.

innovative production networks. This literature has two main arguments. On the one hand, it attempts to identify sets of discrete factors that underpin the production complex: universities, airports, a good quality of life, the existence of highly trained workers, and so on. On the other hand, it advances the concept of an “innovative milieu,” which in turn often is based on the “synergies” of the factors. The problem is that the former literature’s lists can be cut to fit any circumstance, and that no particular factor or combination thereof corresponds to the cases. Moreover, it offers us no analytical way to specify what is meant by “milieu” or “synergy.” An analytical alternative is needed.

I argue that we must look at the codes, channels of interaction, and ways of organizing and coordinating behaviors that make learning possible. This is an enormously complex theoretical problem, and I limit myself here to a few basic points. If we assume an outer technological frontier (maximum outer limit of existing fundamental or applied knowledge) in a given industry, the question arises as to why some producers act on it more than others, and why they act on it as they do. One obvious reason is that information is imperfect, and some sets of producers have better institutions to reduce the uncertainty associated with this information in such a way that they learn more from it. But this is again just the “weak” version of uncertainty. More positively, the use and development of information so that technological learning takes place has to do with the qualitative behaviors of agents in a network.

Evidence suggests that not all agents are alike when it comes to transactional activity. Economists such as Williamson assume that all actors have universal behavioral principles, such as short-run maximizing and opportunism (the “under-socialized” nature of homo economicu - Granovetter 1985). Implicitly, they consider different organizational forms and transactions costs, holding the objects (products, technologies) of economic activity constant so that the focus is on the optimal arrangements between agents. Our stress on product innovation suggests that these objects are not fixed: indeed, that it is the simultaneous development of objects by agents (subjects) that cuts to the heart of the economic development process (Boltanski and Thevenot 1987). Needs and the possibilities for addressing them are defined simultaneously. This insight opens up a field of investigation of economic organization that goes beyond the transactions costs school in two ways.

First, under conditions of continuous, open-ended product innovation or differentiation, institutional arrangements are the result of this simultaneous subject-object interaction; there is no optimal arrangement because what is to be optimized changes along with the institutional arrangements themselves. Thus, as the recent case study literature on PBTL networks suggests, production systems develop products under different mixes of competition, cooperation, trust, and opportunism of the transactors, and agents (firms, individuals) are motivated by different values about what is good and bad, just and unjust, as well as different localized incentive structures (Sabel 1990). These localized expectations figure importantly in the agents’ short-term choices (time horizons, payoff points, etc.; for a case study of trust relations in French subcontracting networks, see Lorenz 1988). In one well-known critique of standard transactional theory, these features of transactional behavior were aptly described as the social “embeddedness” of economic activity (Granovetter 1985). Here I have gone beyond Granovetter’s critique to center on the definition of economic objects themselves.

If the rationalities of producers and users - their expectations, preference structures, and so on - differ considerably from place to place, some types of rationality and the behavioral routines, rules, and institutions that underlie them seem to be more effective than others at

promoting interactions that lead to technological learning. The potential positive externalities of production networks are only realized according to the concrete qualities of the transactions themselves. Our attention should, therefore, focus on the theoretical and empirical problem of the qualitative basis and differentiation of external economies of networks.

Second, agents (subjects) are produced by their underlying environment. Such agents are the most critical economic resource when considering PBTL. Identifying the structure of participation of these agents in the production system is therefore necessary. What brings agents into mutual engagement in such a way that PBTL occurs? We can call such principles of mutual engagement the conventions of that production network and its agglomeration. Conventions lie beneath the regularized social interactions that sometimes appear as formal rules or institutions, and at other times appear simply as routines or unwritten “rules of the game.” Conventions describe the underlying forms of collective order of the production system, especially the underlying principles of justification (and distribution) of rewards to the various agents in the system.

Specifically, the conventional environment of PBTL systems is likely to rest on rules and practices that: (1) coordinate shared preferences, particularly with respect to growth or product quality; (2) reconcile discordant preferences that the key actors are encouraged to participate in innovative activity; and (3) regulate the buying and selling of goods and services by defining standards of value (e.g., wage/effort bargains). In all, we can say that the analysis of conventions helps us understand why different kinds of resources (skills, capital, etc.) are mobilized and bound together in a division of labor and why the possibilities for doing this differ from place to place.

A final dimension of the problem of learning, networks, lock-in, and conventions is that many kinds of knowledge - once deployed in the context of a specific learning system - are nonetheless not fully appropriable by those who do the learning. Knowledge, whether in the form of capital, products, or theoretical knowledge itself, leaks and is easily imitated. Even a learning system that is qualitatively distinctive and deeply rooted in specific local conditions is viable only by virtue of its true dynamism, not by virtue of its particular outputs in the short run. Yet production networks and their institutionalized conventions and rationalities are themselves subject to the same path dependency and lock-in that they are designed to help individual firms avoid. As with any complex social system, networks may have better or worse capacities for collective adjustment in the face of external events and higher or lower levels of internal dynamism. Such capacities are essential when the technological frontier (i.e., the basic menu of technological possibilities, either within a sector or in the appearance of new product groups) changes in the form of a cluster of basic, radical innovations:

The costs involved in breaking up existing codes and channels of information will tend to cement the existing structure... The force of resistance might be strongest where the interaction has been most effective in establishing strong poles of competitiveness (Lundvall 1990, 21).

Thus we need to know not only what elements are responsible for turning production networks into technologically dynamic learning systems, and what specific conventions select these specialties for each country, but also the mechanisms that make collective adjustments of the learning systems possible or, in contrast, what sorts of conventions promote collective lock-in or blockage of learning in networks (Lorenz 1991).

Index

The Territorialization of Learning: Regions and Countries

Social scientists have recently, rather belatedly, begun to recognize that the

invention and modification of products and processes - i.e., innovation and learning - rest on an extraordinarily complex variety of institutions, social habits, ideologies, and expectations, and that even firm and market structures are to a certain extent outcomes of these underlying social structures. The habits of production cannot be reduced to discrete “factors” such as quantity of R&D expenditures, entrepreneurialism, the availability of capital, and so on.

Most of this reflection has been inspired by the rise of Japanese and German economic power and the growing recognition that these countries (i.e., their firms, labor processes, institutions, and cultures) really differ in fundamental ways from their less successful competitors. To date, however, the debate has been posed almost exclusively in terms of the differences between “national innovation systems.” Processes that operate at the national level undoubtedly have much to do with the selection of national specializations and the differential success levels of countries. I advance the further hypothesis that - at least in some countries - political- economic cultures, rules, and institutions are also highly differentiated at the regional level.

One important dimension of such regionalization is the production of public goods upon which technological learning depends, especially in the form of skilled production and intellectual labor, where no single firm or training institution can possibly produce these resources. The external economies that attach to the training and specialization of such labor have to do not only with the localization of training, but with the fact that interpersonal knowledge is a key ingredient of the formation of each trained cohort of workers, with cognitive and communicative elements that we are only beginning to analyze in social scientific terms.

In any case, since countries are leading innovators in a relatively few industries and these industries tend to be highly concentrated in particular regions, inquiry at the level of industry and region incorporates national-level forces while also permitting focused investigation of specific regional processes, rules, habits, and institutions ⁶This, then, is the field of inquiry into the sources of PBTL systems, and the differences between these systems and others. It involves a structured conceptualization of a broad set of features of a regional political-economic culture, its institutions, and the behavioral routines of its collective agents.

Index

List of Tables

Table 1: Trade Composition and Trade Ratios for Main Industrialized Countries by Typology of Industrial Sectors

Table 2: Top Fifty U.S. Industries Ranked in Terms of World Export Share, 1985

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, United States

Table 5B: The Roots of Export Specialization: Learning/Economics of Variety vs. the Rest, Italy

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 United States

Index

previous page

next page

AAP Homepage