A body of theory may be recognized as defective if it fails to elucidate the reality that it claims to represent. This may be the case if the theory is logically flawed, if it assumes away some of the critical variables, or if the postulated relationships or outcomes predicted are shown to be false. The conduct of inquiry in any field involves the critical scrutiny of existing theory and its refinement, reformulation, are replacement by an alternative perspective. The past two decades have seen a sustained assault on classical location theory, along with various attempts to broaden the perspective of spatial economic analysis.

The main thrust of the critique may be summarized as follows. First, the preoccupation with optimality characteristic of conventional economic theory fails to recognize the evidently suboptimal nature of much plant location practice, when judged against such criteria as cost minimization or the maximization of profits. Second, and related to this, is the fact that theory of the kind explored in the previous chapter, assuming as it does some idealized conception of omniscient "economic man," neglects to explore actual human behavior. Third, the nature of the industrial organizations assume din conventional economic theory, and uncritically adopted in much location theory, bears little relationship to those actually existing in the contemporary world. Fourth, reliance on neoclassical economics has prevented the development of a theory of industrial location adequately grounded in political economy. Such shortcomings will obviously limit the validity of location theory, both as a means of explaining our sorry reality and as a guide to improving it.

This chapter elaborates various elements of the critique of industrial location theory in the classical tradition and explains some of the reformulations that have teen proposed. But before proceeding, it is important to recognize that we are not about to demolish the work summarized in Chapter4_far from it. Some of the lines of critique are not as persuasive as is generally supposed, so we shall have cause to come to the defense of certain elements of classical theory. We shall see, as we proceed, that new perspectives are seldom a dramatic break from the past; more often they comprise incremental developments rooted in what has gone before and, in some respects,captive of past approaches or intellectual traditions. Revolutionary change in theory is as rare as in society.


The basic assumption in conventional economic theory is that firms seek to maximize profits. This is done by achieving optimality with respect to the scale and technique decisions (explained in Chapter 2), associated with which is optimality with respect to location. The operating environment assumed is that of the perfectly competitive market, as held to exist within agree-enterprise or capitalist economic system.

While neoclassical economic theory does provide a guide to the conditions required for efficiency in resource allocation, it is obviously limited in its capacity to explain actual (location) decisions. Firms do not necessarily find the optimum, whether with respect to scale, technique, or location. All manner of individual circumstances can be evoked to account for suboptimal economic behavior, such as the "golf-course effect" mentioned in chapter 3. Greenhut recognized the importance of personal factors in his theory of plant location, but it was the accumulation of empirical evidence rather than a priori reasoning that finally brought the problem of suboptimal decision-making to the surface. A more inductive approach thus began to replace deductive economic theory as the prevailing paradigm for location analysis.

Empirical investigations of the plant location process using questionnaires or interviews had been suggesting the importance of personal factors for some time before the emergence of the behavioral perspective in geography. For example, in a survey of new industrial development or expansion schemes in Atlanta, Chapman and Wells (1958) found personal reasons cited by eleven of forty-nine firms as factors influencing the choice of this particular city as a location. Overall, personal factors ranked fourth behind the market, transportation facilities, and labor. Katona and Morgan (1950,1952) interviewed executives of 188 Michigan plants and found that enough of these to represent 51 percent of the state's employees listed personal rea

sons of Some kind for their original location in Michigan. The most frequent response to the question of how the firm happened to locate in Michigan rather than another state was that the founder lived in Michigan when the plant was first opened. Malinowski and Kinnard (1961) took a random sample of small firms in the Hartford, Connecticut, area and found that of a total of 359 reasons mentioned by executives for location at their specific site, 44 ~A~ere purely personal. The personal reasons included proximity to home an~d family, personal attachment to the area, and important contacts.The general conclusion to emerge from these and similar studies was that

tradition?nal cost and demand factors have indeed an important influence on location,~tion decisions, but . . . other economic as well as non-economic factor! may play a significant role under certain circumstances. Thus the assumption_n of profit maximization is too restrictive for an analysis of location decisions. The process of industrial location is clarified if we recognize other preferences in addition to the desire to maximize profits[Mueller and Morgan, 1962, p. 204].

It is impo_rtant to make a distinction between so-called personal factors and SubOptimizing behavior. The frequently observed association of plant location with the home town of the founder (a "personal" factor) could very well be a profit-maximizing choice, as Webber (1972, p. 100) points out. The Actual practice of building up a business, with its heavy reliance on local contactS, knowledge, credit, and so on, may not involve a locational choice at all: there was only the one possibility. Suboptimality refers to decisions that are not the best, whether the criterion is profit maximization or a''noneco0_nomic objective. Decisions made on the basis of personal factors need not be suboptimal.

As in the development of conventional location theory, the conceptualization of suboptimal decision-making was initially derivative, depending on an ad`'vance in another field of inquiry. The breakthrough came from administration_n theory, in the form of the notion of satisficing behavior. Optimization requires a level of information and decision-making ability that is beyond the capacity of both the human mind and the resources of the industrial organization. Simon (1957, p. xxiv) refers to "the theory of intended and bounded rationality_of the behavior of human beings who satisficebecause theY have not the wits to maximize. " Everyday experience confirms that we co()ntinually make decisions that are satisfactory in the circumstances though n~not necessarily optimal; so it is with industrial location. Thus a "satisficing plan could replace the all-knowing, perfectly able and rationalhomo economicus of traditional theory.

How n~right conventional location theory accommodate this unfamiliar character, capable of such odd behavior? A casual remark by Losch (1954,p. 16) pr~provides an initial clue: "as long as such a capricious choice costs no more than the entrepreneurial profit, it is still consistent with theory."Simon thought in terms of bounded rationality; we may regard freedom to indulge in suboptimal behavior as spatially constrained. Economic circumstances impose limits to how far plant location can depart from the optimum and still survive.

The formal integration of suboptimal location choice and the traditional focus on optimization was originally suggested by Rawstron (1958). He was concerned with the extent to which choice of location is restricted if economic viability is to be achieved and with how this restriction is imposed.Variations from place to place in the cost of production generate what he termed spatial margins to profitability_one of the few really original concepts introduced into spatial economic analysis by a geographer. Beyond the margin, total cost exceeds total revenue, pr.profit is negative (a loss), and the location is not economically viable. Within the margin some profit can be made anywhere, though not necessarily maximum profits. The spatial margin to profitablity thus bounds the area within which there is freedom of locational choice if the assumption of profit-maximizing behavior is abandoned.

The spatial margin to profitability can be explained in a simple three-dimensional graphic model (Figure 5.1). Spatial variations in total cost are shown as a cost surface, in the simplified form of an inverted cone. Total revenue is assumed to be the same everywhere, so the revenue surface is a horizontal plane. The intersection of the cost and revenue surfaces define the spatial limits to profitability, or margin, which is projected onto the two dimensions of the earth's surface at the bottom of the diagram. The optimum location (O) is where total revenue exceeds total cost by the greatest amount_the minimum-cost location in this model, where revenue is a spatial constant.

Smith (1966) took up and developed the idea of the spatial interaction of cost and revenue as a constraint on locational choice in a series of graphic models (Figure 5.2). As in Figure 5.1, the situations depicted are very much a simplification of reality and are based on the assumption that cost and price at any place are fixed and cannot be altered by the individual firm by large-scale production, changes of manufacturing techniques or input combination, or management skill. For simplicity of presentation it is also assumed that the output any firm can attain is constant in space and that variations in demand, if they exist, are reflected in variations from place to place in price. In fact, location is determined solely by the interaction of unit cost and price (or revenue), all other influences having been assumed away.

The effect of spatial variations in cost and revenue can best be illustrated by holding one constant and allowing the other to vary. In Figure 5.2a,costs are variable in space while demand is constant and the price obtainable is the same everywhere

(p). Cost and price (dollars) are plotted on the vertical axis, and distance along the horizontal axis representing linear or one-dimensional space. The average cost per unit of production at any point in space is indicated by the appropriate value on the line AC, which rises in both directions from a point 0. This line is the space cost curve (not to be confused with the conventional cost curve of production theory where the horizontal axis measures quantity instead of distance). The point of minimum cost is represented by 0, and Ma and Mb show where the average cost is just equal to the price obtainable. The vertical distance between p andAC, where price exceeds average cost (that is, between Ma and Mb), indicates the average profit on each unit of production.

Because of the assumption that output does not vary from place to place or firm to firm, the spatial average cost/revenue situation can be taken as


The main defect of an approach based on recognition of spatial limits to freedom of choice is that it fails to inform about how decisions (within the margin) are actually made. The past two decades have seen the ascendancy of the so-called behavioral school in locational analysis, as in human geography as a whole, with its emphasis on the elucidation of real-world decision-making. Rejecting the unduly deterministic framework derived from economic theory, the behavioralists have sought guidance in such fields as organization theory and psychology. But a convincing general theory capable of incorporating the variety of human motivations and actions found in practice has eluded even its most ardent pursuers. Indeed, the behavioral movement as a whole may fairly be characterized as that of empirical findings in search of a theory. In this section we explore a few of the alternative frameworks have been proposed, so as to convey something of the nature of the behavioral approach.

The fundamental problem is that of actually explaining the evidently suboptimal location decisions made in the real world. One possibility,recognized by Losch (1954, p. 260), is that an individual businessman may choose a location other than the economic optimum because he is concerned with the greatest total utility, rather than with money profits alone.

The mathematical determination of the optimum transport point, for example, is infinitely more impressive as a solution to the location problem but also incomparably less accurate than the statement that an entrepreneur, all things considered, will establish his enterprise at a place that helices best [Losch, 1954, p. 224].

Greenhut (1956, pp. 175-176) developed a more detailed argument along similar lines: the non pecuniary satisfaction which an entrepreneur may derive from setting up in a particular location can be regarded as psychic income, and each entrepreneur will choose the location at which total satisfaction in terms of both financial and psychic income is maximized. The logical extension of this argument is that the location theorist should regard the maximization of personal satisfaction as the entrepreneur's goal rather than the maximization of financial profit. But as Greenhut points out, this creates difficulties. To say, in effect, that man puts factories where it pleases him is clearly an inadequate basis for a theoretical framework aimed at assisting both the interpretation of actual industrial location patterns and the solution of practical problems. Smith (1966, pp. 107-109) attempted to portray location decisions within the margin as involving a payoff between financial profits and psychic income. But this requires either the knowledge to identify the optimum from which psychic income draws the entrepreneur(Pred, 1967, pp. 87-88) or some complex calculus of alternative sources of satisfaction under conditions of uncertainty. Thus, the concept of total utility maximization seems as impotent in location analysis as it has proved toke in the theory of consumer behavior, where it originated.

The first major departure from traditional economic thinking in geography was that of Pred (1967, 1969). He challenged the logical consistency of the assumptions associated with "economic man," the motives ascribed to him, his level of knowledge, and the mental acumen that he is supposed to possess. Instead, man is seen as possessing both limited knowledge and limited power to use it:

Every locational decision is viewed as occurring under conditions of varying information and ability, ranging, at least theoretically, from null to perfect knowledge of all alternatives, and as being governed by the varying abilities (as well as objectives) of the decision-maker(s) [Pred,1967, p. 24].

Any entrepreneur thus has a place in what Pred terms the behavioral matrix(Figure 5.4). A position toward the bottom right of the matrix indicates a good level of knowledge as well as good ability to use it, and there would bead high degree of probability of a good choice of location, perhaps one near the economic optimum. As knowledge and ability decrease, toward the top left of the matrix, the probability of a good locational choice is reduced.The emphasis is on probability, because good knowledge and ability are nota guarantee of a good choice of location (though they make it very likely),just as there is an outside chance that a firm with little knowledge of the alternatives and a poor management could be lucky enough to make a good decision.

The behavioral matrix was put forward as a device to assist in understanding real-world deviations from location patterns produced by deterministic models based on the assumption of economic man. Variations in the information available to individual decision-makers and in their ability to use it helps to explain the simultaneous presence of random and non random elements in spatial distributions. The argument is that the regular elements are the result of locational decisions in general but perhaps not perfect accord with theory of the deterministic kind (that is, near-optimal decisions made from a position toward the bottom right of the matrix). The random elements reflect choices made from a position of more limited information or ability, which would tend to depart significantly from the economic optimum.

Pred applied the behavioral matrix to the interpretation of patterns of agricultural land use and the distribution of central places as well as to industrial location, but it is the latter which is the present concern. His approach is illustrated in Figure 5.4, which relates specific locations in an imaginary situation to the position of firms in the matrix. There are three area sin which some industrial activity is possible. Each area is bounded by spatial margin s to profitable operation, and contains an optimum location designated 0. The location of thirteen imaginary firms is indicated by dots, each of which is connected by a line to the place in the behavioral matrix that best summarizes the firm's hypothetical information and ability-to-use characteristics. Those toward the bottom right of the matrix have in general chosen locations near one of the optima, while of the four firms with very limited information and ability, three are in extra marginal (unprofitable)locations and the other is within a margin but well away from the optimum.The choice of satisfactory locations by some firms with information or ability-to-use handicaps emphasizes the failure of the matrix position to predict exactly how good a choice will be made. The general message is that the apparently chaotic qualities of the spatial distribution of most manufacturing production at any one date is ascribable to the fact that the

real-world is populated by a broad spectrum of bounded rational, satisficing locational actors and not by undifferentiated profit maximizers[Pred, 1967, pp. 91-93].
As Pred (1967, p. 121) concedes, the behavioral-matrix interpretation of location decisions is a useful way of conceptualizing the effect of imperfections in the ability of the entrepreneur ,and the information available to him, but it has obvious weaknesses. These include the facts that the two axes of information and ability are unlikely to be independent (the more able businessman is likely to obtain more information and that there are enormous operational difficulties involved in practical application (Claus and Claus, 1971). As a model that could be used in the explanation of a specific location pattern its value is clearly very restricted ted, because even if numerical scales could be attached to the matrix for thc~' positioning of a set of real-world firms, their actual locations could still not be predicted.

Pred's work came at the end of a decade in which Simon's ideas had gained increasing currency, in the general context of the development of a behavioral theory of the firm. This stimulated further research into location decision-making behavior~r. A number of diverse yet interconnected strands of inquiry can be identified in the subsequent investigations of geographer sand of those economists enticed from their longstanding liaison with homoeconomicus by the VJiles of a more capricious creature.

One of these strands has been concerned with the goals of industrial organizations and their decision-makers. Cyert and March (1963) drew attention to the possibilities Of multiple goals, when an organization is made up of a number of individUals each with his or her own department engaged in a different function. -thus, there can be separate and conflicting goals with respect to production, inventory, sales, and market shares, as well as relating to profits. Other goals include growth and security, emphasized by Galbraith (1967) as being of the new managerial technostructure. Hamilton (1974) has described a mass of evidence on the locational implications of alternative goals, which he summarizes as follows (Hamilton, 1974, p.14)

Some business pOlicies and goals are likely to have greater spatial repercussions for the Spatial form and growth of the firm and, hence, for industrial distributions generally.... the goals of firms can be diverse and the spatial implication_nS of policies for achieving those goals can be equally diverse.... Most firms undoubtedly seek some level of profitability and some do strive, for_r maximum profits. However, many firms also seek to achieve other goals: growth of the firm, larger control of the market for particular pro~ducts, diversification of interests, entrepreneurial satisfaction or simply self-preservation.... Some of these goals may be incompatible, but Others clearly overlap and may in practice converge. One management may decide that corporate growth, for instance, may best be achieved by diversification: the emergent pattern of plant locations within the corporation_n is likely to differ significantly from a corporation whose management is bent on growth via larger control of the market fora fairly narrow range of products.

As Hamilton points out, for many firms the choice of location is likely to be a by-product of a particular policy designed to achieve some non spatial goal.

Another research strand has been concerned with the level of informationavailable to those making a location decision. Tornqvist (1970) has stressed the importance of contact systems, or how firms are placed in relation to communication flows whereby information is disseminated and received.Closely related to this is research on the diffusion of innovation. Some have examined the process of search, whereby the extent of the search and the order in which it is conducted affect the solution arrived at: it is unlikely that the actual search for a location will review all possible alternatives, and those near at hand are likely to be better known than those further away.The experience of learning is also important: learning should increase with the length of the decision process and the number of alternatives examined,and also with how often a decision such as that of location has been made before. The position of an organization on the information-available axis of pred's behavioral matrix is thus subject to a number of influences, some of which are quite complex in their operation. Flows of information from the external environment may be accepted or rejected ("filtered out") by the individual's perception process ("coding mechanism"), which can depend on such conditions as geographical location, cultural group affiliation, socioeconomic status, personality, age and past experience, education, and personal aspirations (Lloyd and Dicken, 1977, p. 318)

While some research has concentrated on particular aspects of locational choice, such as goals, aspirations, ability, and information, there have also been attempts to understand broader behavioral strategies. These may incorporate the uncertainty under which real-world decisions are made, as well as the characteristics of the decision-takers themselves. As in behavioral studies generally, there is a distinction between approaches based on some a priori theory or model and those that favor generalization from empirical experience.

An example of the application of an existing theoretical framework to the study of locational behavior is provided by game theory. Introduced by von Neumann and Morgenstern (1944), adopted with enthusiasm in economics and reg~ional science, and a subject of the occasional flirtation from geographers, game theory addresses problems of decision-making under conditions of uncertainty that depend on the actions of other participants. Such situations are reduced to the simple structure of a game, in which the players seek the best solution in the light of their own objectives, having regard for the likely response of rivals or competitors. Attempts to integrate game theory into conventional spatial economic analysis have been made,notably by Isard et al. (1969) and Webber (1972).

Isard has attempted to apply game theory to certain problems in classical location theory. An example is provided by a reinterpretation of Weber's agglomeration analysis (Isard and Smith, 1967). They assume an island with a deposit of iron ore, and that three countries (1, 2, 3) are interested in building a refinery to process this ore for export. There are three ports (P"P2, P3), and a cost analysis shows that R" R2, and R3 are the best location sat which the three countries can each run an independent refinery. The locations chosen are where the ore deposit comes closest to the three ports (Fig- l ure 5.5). Following Weber, an area within which agglomeration would be possible is found in the center of the island by the intersection of the relevant isodapanes. The area thus defined (shaded in Figure 5.5) offers a lower unit delivered price of refined ore at the ports, providing all three countries to locate their facility there, because here economies of agglomeration are sufficient to overcome increased transport costs. The area within which agglomeration is possible is termed the joint action space. Within this space ;the best location for each country would be at the point closest to its original least-cost location and its port (A " A 2, and A 3), but in order to take advantage of economies of agglomeration all must be in the same place.

Isard and Smith develop a Weberian Locational Game, in which the conflict between the interests of the three nations are resolved under alternative assumptions regarding the way the participants will behave. Successive concessions on the part of the three participants tend to lead them (not unexpectedly) from their preferred locations at A " A 2, and A 3 toward a point of compromise in the middle of the joint action space.

Game theory was also applied to the problem of locational choice where firms are competing for an areally distributed market. The basic frame of reference is the Hotelling case, following Stevens (1961b). Isard and Smith claim that game theory helps to clarify many of the structural properties of locational interdependence problems, indicating certain common elements.In addition, cooperative procedures for the resolution of locational conflict are suggested, which may have practical applications.

However, there are some major objections to the adoption of game theory as a general framework for the analysis of location decision-making under conditions of uncertainty. Chisholm (1975, p. 141) emphasizes that"while game theory is clearly a useful way in which a firm can formulate its own approach to the task of finding an optimal strategy . . . it manifestly cannot supply a framework for identifying a stable and optimal spatial allocation for a whole system of competing firms." Game theory is applicable only to a few highly unrealistic situations; the greater the complexity, the more difficult it is to solve the mathematics required to generate a solution.The "payoff matrix" whereby players attach values to particular strategies or outcomes seems to depend on assumptions no less exacting than those involved in conventional optimizing behavior; as Simon (1959, p. 266) remarked over twenty years ago, game theory "requires of economic man even more fantastic reasoning powers than does classical economic theory. "Thus, little is now lost by assigning game theory to the same oblivion earned by certain other esoteric devices that have temporarily distracted the search for theory in behavioral studies.

The more empirical approach appears to offer a greater prospect of legitimate generalization concerning the actual decision-making process. An early contribution was made by Stafford (1969, 1972), who recognized a sequential structure in which the entrepreneur considers a variety of strategies, from in-site expansion to alternative feasible sites, before a location is finally chosen. He sees demand conditions defining the larger area of search, with specific sites considered on the basis of cost. Other research along similar lines includes that of Townroe (1969, 1975) and Lloyd and dicken (1977, pp. 325-339). The latter have constructed a detailed description of how a small manufacturing organization makes decisions, from initial entry into the industry, through locational search, to evaluation andchoice.

One example of the results of this essentially inductive, empirical approach must suffice. Rees (1972, 1974) has examined the investment location decisions of large British and American firms, based on interviews with executives. He claims that this has confirmed the validity of a conceptual framework of locational search, learning, and choice evaluation, illustrate din Figure 5.6. Using the terminology of systems theory favored by some of the practitioners of the behavioral approach, Rees (1974, p. 190) describes this as "a simple stimulus-response mechanism." Rees's framework brings together a number of the notions already introduced in this section. The first phase of the decision-making process is the recognition that a problem of growth exists with respect to the satisfaction of demand: in the prevailing jargon, a "stress threshold" has been reached. The alternative responses to in-situ expansion may be relocation, acquisition of another firm, or opening a new branch plant. A new plant involves a three-stage search procedure,the outcome of which leads to a decision and, finally, to the allocation of resources. It also generates "feedback" into learning behavior and into the decision-making environment, which now includes the branch plant. As such a framework is found to accord (or disagree) with further observation,it can be confirmed or modified as a steadily more convincing empirical generalization.

One helpful feature of Rees's model is the distinction between the short term and long-term response of the organization. A major weakness of the traditional theory of the firm is that it tended to ignore the time-frame within which profit-maximization was to be sought. The behavioral approach is much more realistic in its recognition that the environment within which the industrial organization operates is in a constant state of flux, and that how firms adapt to change will vary according to whether they seek to satisfy their goals (profit maximization or otherwise) over five years, ten years, or whatever the case may be. Dynamic aspects of industrial location are the subject of recent empirical research (Collins and Walker, 1975) but little byway of generalization has yet emerged.

While the behavioral approach could fairly be described as the mainstream of industrial location analysis during much of the 1970s, some disquiet has been expressed as to its actual achievements. Townroe (1975, p.39) concedes that development since the appearance of Cyert and March's book [A Behavioral Theory of the Firm] in 1963 has not matched expectations. Teleradiography approach is essentially short run and the firm is regarded as rather a passive agent, reacting to stimuli from an external environment,a view which is hardly in accord with, for example, Galbraith.

The degree of control that a large corporation can exercise over its operating environment is a major feature of the "new industrial state" described by Galbraith (1967). Townroe goes on to say:

It is clear that the strength of the behavioural view lies in the insights provided rather than as a basis for prediction, or for evaluation, either of the locational behaviour of a firm or group of firms or of policy instruments.

Doubts have also been expressed as to whether the mass of empirical inquiries that are being conducted will actually yield generalizations:

No progress can be made in "behavioural" studies until the assumption that the spatial behaviour of industrial firms displays consistencies is confirmed.... reality is complex, and amenable only to very loose general statements. Without some normative, simplified framework (relating either to corporate or social goals) "behavioural " theory will merely lead us back to description [Wood, 1975, p. 46].

A fitting conclusion is provided by Keeble (1977a, p. 306), who says of the behavioral analysis of location decision-making that it "seems now clearly to be subject to the law of diminishing returns." However, we dissent from his rider that "work in this particular field will be useful only where either new techniques are used or new questions posed": the indiscriminate adoption of new techniques (factor analysis, probability models, and the like)has been a major contributor to the tendency of the behavioral approach to degenerate into a sterile form of speculative empiricism, while the basic question of how it is that industrial organizations select particular locations still remains.


One aspect of the recent concern with the actual decision-making process has been deliberately omitted from the discussion of the behavioral approach above: the impact of industrial organization. This topic is important enough to require separate treatment, and some of the findings may absolve it from the criticism that the behavioral approach in general has promised more than it has been able to deliver. Again, it is impossible to provide a comprehensive review of the rapidly growing body of literature in this field;all that can be attempted here is to reveal the major features of the more instructive approaches that have been adopted.

The focus on industrial organization, like the behavioral approach in general, is a response to the inadequacy of traditional theory as it confronts the reality of the contemporary industrial world. Two distinct but interconnected aspects are recognized: (1) the increasing scale and complexity of the organizational unit, as large multiplant and even multinational firms have replaced the small, single-plant concern assumed in much location theory;(2) the changing nature of the environment within which the industrial organization operates, with new conditions and patterns of interdependencies unrecognized in traditional theory. Work in this field derives special significance from the fact that, in addition to the conceptual relevance, there is a particular social relevance to studying the spatial behavior of large industrial organizations (McNee, 1974, p. 50). Indeed, it is in understanding the operation of the giant capitalist corporations that increasingly dominate the world economy that industrial location analysis makes one of its most important connections with the contemporary concern with questions of uneven development and human welfare.

The possibility of a relationship between scale and location decision-making has been recognized in economics for some time. Tiebout (1957) claimed that the "personal" element plays the largest role in locational choice in the small firm and the new entry into an industry. He argued that the of officers of the Ford Motor Company, for example, would not be expected to allow personal considerations to enter into the decision, which would be based in all probability on revenue and cost considerations. In contrast, the clever small businessman, enjoying relatively large returns to capital invested, maybe able to get away with a bad location. The implication is that such a businessman is intuitive, the major corporation more rational. However, such simplistic distinctions are challenged by modern organization theory, which sees firms as coalitions of individuals each of whom may have his or her own goals. The larger the firm, the greater the potential for internal functional specialization, goal diversity, and possible conflict. Single-minded dedication to pursuit of profit-maximizing locations, or any other objective, may thus be more a feature of the (small) firm with a single mind in control.

Geographical interest in the implications of large-scale industrial organization first crystallized around the theme of the geography of enterpriseæ(McNee, 1960). As McNee (1958, p. 322) explained, most of the mass production industries tend to be highly concentrated among a few corporation giants. Thus general principles of location and spatial interaction in the automobile industry of the United States could be derivations the study of the aggregate automotive industry. But such principles might have greater validity if they were supplemented by comparative study of location and spatial interaction within each of the great automobile institutions General Motors, Ford and Chrysler.

Others followed McNee's lead: Krumme (1969) examined how firms adapt to change, while Steed (1971a, 1971b) used case studies to focus on spatial aspects of corporate growth, decision-making, and adjustment to the environment.

Contemporary explanations of growth and change in large firms tend toke associated with what Thomas (1979, p. 6) describes as a "behavioristic organizational/managerial framework." Wood (1978, p. 147) abstracts three key notions from organization theory:

that organization structure is heavily influenced, if not determined by environmental uncertainty; that this uncertainty is based primarily in the roles and actions of other organizations; and that strategies of firms are therefore derived from their varying relative dependence upon other organizations.

The emphasis here and elsewhere on organizational structure as "a response to environmental uncertainty, rather than as exerting independent control of the environment" (Wood, 1978, p. 149), might appear to conflict with the perspective of Galbraith (1967), who sees the modern corporation as increasingly able to control conditions that are taken as given in the conventional theory of the firm. These views are reconciled by seeing the large corporation as attempting to reduce uncertainty, for example by research and development activities that inform about the environment and facilitate its manipulation. Shifting the context from perfect competition to oligopoly,where a few major concerns dominate an industry, necessarily involves recognition of the power to control that goes with size and monopolistic tendencies. Such control can extend as far as the political system, as exemplified in the influence that certain multinational firms have been able to bring to bear in their "host" countries in Central and South America. Part of the contemporary perspective on the multinationals is that uncertainty is anathema to the firm and it will therefore tend to internalize the condition of uncertainty whenever it can. The assumption is that the condition is at least alleviated if the firm can establish control or some appreciable measure of influence over the uncertainty in the environment external to the firm. In this way the firm attempts to convert uncertainty into risk as such firms are more used to or prefer taking calculated risks than coping with states of uncertainty [Thomas, 1979, p. 10].

Government activity can hardly be internalized by the private corporation,but influence on government helps to eliminate what is in many parts of the world an important element of uncertainty in the firm's operating environments.

Thomas (1979, p. 3) suggests four different kinds of environments within which firms operate: firm specific, industry specific, geographic-area specific, and social-political-economic system specific. Geographers have tended to stress the area-specific or spatial aspects of the environment.Wood (1978, pp. 147-148) points out that for larger multiplant firms the national and international environment may impose broad structural requirements and determine the "corporate context" of particular plants; local and regional environments can vary and thus affect the performance of individual plants and hence that of the organization as a whole. Norcliffe(1975) demotes the traditional factors of labor and transportation costs from a dominant role in the determination of location patterns and puts in their place infrastructure availability, internal and external economies, and contact and linkage patterns. These are all essentially characteristics of place, through which the operating environment will be differentiated.

The growth and organizational structure of the large multiplant firm is closely associated with the integration (and control) of specific features of the environment. Uncertainty as to supply of inputs or markets for outputs may be reduced by acquisitions or mergers. The backward and forward linkages common to any industrial organization thus become internalized through the process of vertical integration. Competitors pose a threat an dare themselves a source of uncertainty, so they may also be subject to acquisition: the process of horizontal integration, or growth in the same line of production. The vulnerability implicit in concentrating on a single product may be reduced by diversification, which can involve new plants or acquisitions. Linkages will also exist with suppliers of machinery, repair and maintenance operations, and sources of information on matters ranging from financial advice to technical developments. The modern industrial organization is becoming increasingly complex in its structure and operates within an intricate system of interdependencies. This perspective is an important elaboration of the familiar notion of localized external economies, recognized since Weber's treatment of agglomeration.

A major focus of recent research has been the manner in which large multiplant firms are organized in space. Lloyd and Dicken (1977, p. 368), citing chandler and Redlich (1961), emphasize the importance of geographical space itself to organizational development: geographical dispersion was The first step in forming the modern industrial enterprise, because it made necessary the functional differentiation between "headquarters" and "field."Arising out of this fundamental spatial-organizational distinction are the conflicting tendencies of concentration and dispersal, which can be observed at different geographical scales throughout the world economy.Hamilton (1978, p. 7) summarized this situation as follows:

The rise of large corporations has embodied two contrary long-term spatial tendencies. First, the increased concentration of capital, decision making and administrative control leading to a diminution in the numbers of firms and hence also of decision control centres in operation in each industry. Concentration occurs at headquarters of decisions and services for the entire corporation.... Greater concentration may result, from rationalization and increased scale economies through organizational and technical change. Second, the increased dispersion or d~centralization of production, services, marketing and, to a lesser degree~decision-making. This contrary tendency results from corporate expansion and its concomitant creation or acquisition of branches or subsidiaries and franchised outlets and from functional diversification or product differentiation.

This increasing spatial concentration of control in dispersed systems of production has important implications for the general process of economic development and differentiation of social well-being, as has been shown bHymer (1975) and Smith (1977a, pp. 123-126; 1979, pp. 322-325). The spatial arrangement of the world capitalist economy is seen increasingly to mirror the hierarchy of control in the modern corporation. Three distinct level of control are recognized, with their associated facilities progressively more dispersed. At the top is the senior management making strategic decisions in the major national or world business centers. At the bottom, and most dispersed, is the day-to-day operation of the productive enterprises. Between them is the control and coordination of the lowest level, administering decisions made at the top.

Increasing scale is likely to broaden the spatial extent of a firm's operations. Taylor (1975) divides operational space into three successively more~restricted spaces: action space defined by material linkages, informativeness defined by information flows, and decision space within which location of new plant might be considered. As firms grow, their action space~will expand with their linkage pattern, information will flow in from a wide~area, and possibilities for new facilities will open up in places hitherto not considered. The operational space is thus extended from the local, through regional and national, to the international level. The locational evolution of an industrial organization is thus portrayed as a spatial expansion with growing scale and structural complexity, mediated by the behavioral process of receiving information and making decisions.

A simple portrayal of the geographical pattern of corporate growth ha~been provided by Hakanson (1979), incorporating various aspects of the work of the organizational school. Figure 5.7 shows the corporation's action space, divided into a "core area" where it was first founded, the remainder of the "home country," and an outer circle representing the rest of the world_itself subdivided into different foreign countries. During its early life the firm is closely tied to the immediate surrounding environment,with incremental in situ growth. Stage one is thus that of the single-plant

firm of traditional theory. The second stage involves penetration of the home market via sales offices, the expansion of central management, andrew production capacity away from the mother plant. Stage 3 sees the first incursions into foreign markets, via a network of sales agents; at home, production capacity may be expanded outside the original core area. Next, sales offices replace some of the overseas agents. Finally, production plants appear in the foreign markets, as acquisitions or branches. Stage 5 thus represents, in simplified form, the fully fledged multinational corporation. Diversification can take place at any stage, but it becomes the dominant growth strategy in the ultimate stage of formation of the multiproduct, multinational conglomerate.

The fortunes of individual industrial plants will obviously be related to their position and status within the wider corporate structure. Expansion,contraction, or closure of local production capacity is thus dependent on what happens elsewhere_on the geography of control. Dicken and Lloyd(1978) have suggested a typology of firms according to control characteristics, connected with the process of change in a local area's industrial struc

ture. Figure 5.8 distinguishes between local single-plant firms, multiplant firms with local headquarters, and firms with headquarters elsewhere, domestically or foreign owned. Exits (closures), entries (new plants), and the levels of employment of survivors may all be associated with organizational status. Such a conceptual framework can be used to guide empirical research, as Dicken and Lloyd show in the cases of Manchester and Merseyside. Closely related to this is research on the experience of acquisition activity in British industry (Leigh and North, 1978). The crucial issue is whether a local (regional or national) economy with a preponderance of branches of externally owned corporations is more or less vulnerable to the impact of those economic forces that make for the contraction or expansion of industrial activity.

This section has provided a selective review of some recent development sin the study of the locational implications of modern industrial organization. What is the significance of this perspective for the kind of theory reviewed in the previous chapter? One major line of criticism has been the inability of a Weberian analysis to accommodate external (or internal) economies associated with inter industry (or intra industry) linkages. Despite weber's attempt to build agglomeration into his theoretical framework,and the tendency for industrial activity to concentrate about certain point sin Losch's economic landscapes and Isard's graphic synthesis, externalitieshave certainly not been dealt with as adequately as transportation and other readily identifiable cost items. Wood (1969, pp. 32-33) has stated that "traditional location theory regards as awkward exceptions the complex agglomerations that are supposed to be based upon the principles of external economies of scale and close functional linkage," and that there is little toke gained from taking Weber as a theoretical starting point.

However, it can be argued that the difficulties raised by agglomeration and linkage are essentially problems of measurement (Smith, 1970a). If this could be overcome it might be possible to deal with these variables operationally within existing models derived from classical location theory. Agglomeration economies express themselves through reduced operating costs,and their spatial variations could be treated in the same way as any other cost item. Industrial linkages are also cost-reducing factors and could be incorporated into the Weber model by having each input source (supplying firm) or output destination (purchasing firm) as a corner in the locational polygon. From a practical point of view the solution of a problem in which there are many input sources is not as difficult as might be imagined (seePart Three). While recognizing shortcomings in Weber's framework, Smith accused Wood of throwing the baby out with the bath water. Bater and walker (1970) agreed that the breadth of the Weberian approach had not been fully acknowledged.

Another point of criticism is that transportation and labor costs are muchness important to the modern, multiplant corporation than they were when weber wrote. Norcliffe (1975, p. 23) supports this argument as follows:"For the majority of lighter industries transportation costs do not vary greatly from location to location within the industrial heartland, although in peripheral regions remote from important markets, costs do begin to rise more steeply." This basin-shaped cost topography is precisely what is demonstrated by models based on Weber's location triangle and isodapanes, basis demonstrated in Part Two. Skepticism may also be expressed concerning norcliffe's "demotion" of the labor factor: the modern multinational corporation seems far from indifferent to the attractions of sources of cheap labor on a world scale (Lloyd and Dicken, 1977, p. 383), as well as within nations.

The initial inclination to see Weberian or "neo-Weberian" theory as obsolete has now given way to a recognition that it does have some relevance to contemporary industrial organization. Hamilton (1974, p. 9) has observed that "the traditionally-framed location question_the choice of a place forgiven production_may always have been more relevant to the decisions of the multi-functional firms than to decisions by uni-functional industrial firms," for whom the choice of production in situ would be more common.Elsewhere, he concedes that the general methodology of Weberian and neo- Weberian schools, with its concern for cost relationships, is still relevant, even though technological changes and rising concern for environmental quality and social welfare have altered the parameters and provided that it is cast within the behavioural framework of organizations, their goals and operation within spatial societal systems [Hamilton, 1978, p. 13].

Hakanson (1979, p. 136), in his model of corporate expansion, states that"in the case of internal growth, the location of vertically integrated production plants is governed by the Weberian minimum-transport principle."

The most vigorous rehabilitation of traditional location theory in the context of the large business enterprise has come from Dicken (1977a), himself a notable contributor to the development of the organizational perspective.His position is summarized as follows:

I believe that it is possible to argue that least cost location theory still has some relevance in helping us to understand the spatial organization of activities_not at the scale normally assumed, but rather at a greatly enlarged organizational and geographical scale. In other words, I would argue that such theory is relevant in precisely those circumstances of corporate organizational structure which are held to have rendered it impotent.

The spatial coincidence of many administrative functions at the top level in the control hierarchy seems to support the agglomerative aspect of Weberian theory. With respect to manufacturing, multinational firms practice an international division of labor in which individual plants specialize in those activities for which their comparative advantage is greatest. New plants may be opened and old ones expanded, contracted, or closed, with a flexibility of locational adjustment much closer to that assumed in traditional theory than has the small single-plant firm. Especially important is the increasing American investment in the Third World in search of cheap labor:

In Weberian terminology it would appear that these labour locations fall within the critical isodapane for specific types of production. Transport costs are more than offset not only by low wage rates but also by the fact that labour productivity is as high as, if not higher than, that in the united States [Dicken, 1977a, p. 141].

As new approaches are developed, their adherents often find it hard to resist the urge to reject totally the conventional wisdom. Those associated with the traditional perspectives may respond in an unduly defensive manner, reluctant to recognize the need for reform. As the debate is engaged,and as alternative theoretical frameworks confront reality, the new ideas a replaced in perspective and such old ideas as have stood the test or found anew relevance may be resurrected. In industrial location theory, we conclude, with Dicken (1977a), that perhaps it is time for Weber to be lifted back into the bath.


The last of the theoretical developments to be discussed is very much a product of the 1970s. As with earlier changes in perspective, it arises from the inadequacy of existing theory. But the move toward what we shall refer to as a "structural" approach differs from the others in important respects.First, it has arisen as much from the apparent inability of existing theory to provide a guide for economic development policy as from the failure to explain reality. Second, it involves a more general and fundamental critique,challenging the contemporary behavioral decision-making perspectives as well as traditional location theory. Third, the alternative proposed involves a radical shift in philosophy, the emphasis being on the need to understand industrial location within a framework of political economy.

The most important single influence on the radical critique and the development of a structural approach has been the resurgence of interest in marxian analysis. Resistance within the capitalist world to the supposedly

ideological basis of Marxism has had a major bearing on the belated introduction of a political-economy perspective into location analysis in Europe~and North America. As Hamilton (1978, p. 3) recognizes, among the most compelling economic theories are those emanating from the powerful socialist thinking embodied in the writings of Marx and lenin. Because of their strong political overtones, predicting and charting the ultimate collapse of capitalism, their economic ideas have been largely ignored by western location theorists. The latter erect, in effect,their own ideological and political barriers against the introduction of

Marxist-Leninist thinking.

The development of location theory, like any academic endeavor, cannot be isolated from its broader societal context. A major point of the radical critique is that the type of economic theory generally accepted in the West,and forming the foundation of existing location theory, is itself ideological.It serves to promote a view of economic life in which the pursuit of personal self-interest somehow maximizes collective welfare, without analyzing the forces molding individual behavior. Above all, it implicitly assumes a capitalist mode of production as the natural order, without doing much to elucidate how capitalism actually works. Conventional economic theory is concerned with appearances, rather than with the actuality of human conduct in the struggle to make a living. Its portrayal of a myriad of individual ~producers and consumers reacting to market forces overlooks the reality of ;a growing concentration of economic power. Thus, location theorists still conveyed the impression of an economy dominated by small-scale single plant firms, a century after Marx recognized the forces creating steadily larger industrial enterprises, and fifty years after Lenin's analysis of imperialism under the emerging monopoly capitalism.

The assault on conventional location theory begins with Weber. Hamilton(1978, p. 5) asks why Weber did not incorporate at least the significance of capitalist industrial organization in his location theory. The answer is not,of course, ignorance, nor even in Weber's case some ideological predisposition toward an uncritical acceptance of capitalism. As Gregory (1979)points out, Weber had been trained in political economy, and he did place explicit restrictions on his abstract formulation of location theory. Quoting weber (1929, pp. 12-13) himself:

We shall see that the kind of industrial location which we have today i snot entirely explained by the "pure" rules of location, and therefore is not purely "economic. " It results to a large extent rather from very definite central aspects of modern capitalism and is a function of modern capitalism which might disappear with it. It results, we may say in hinting at the main point, from degrading labour to a commodity bought today and sold tomorrow, and from the ensuing laws determining the labour market and from the local ''agglomeration of workers" created thereby.

This problem will have to be solved by the "realistic" theory (Weber, 1929,pp. 225-226):

For at this point it becomes necessary to consider particular economic systems. I do not wish to assert that the creation and development of labour locations can be explained by economic reasons; but if it can be so explained the reasons will be related to the position of which the particular economic system gives to labour.... The further realistic theory must therefore consider how labour is handled in the particular economic system which is studied.

The significance of what Weber said is that this broader context has been completely neglected by his followers. The simplicity of Weber's "pure"rules of location, with its indexes, location triangle, and physical analogy,reduced industrial location to a purely mechanical exercise in applied mathematics. That the very existence of cheap labor implies social relations has been forgotten. In the subsequent development of location theory, the inheritance of classical, nineteenth-century political economy evident in weber's book has been lost; the neoclassical conception of an economy has been accepted, virtually without examination. Even where imperfect competition was recognized, within the locational-interdependence school, the focus was on the role of distance in the creation of monopolistic market control rather than on how monopoly actually arises under capitalism anton its implications for the spatial organization of industrial activity. The contemporary organizational decision-making school, while recognizing the importance of multiplant, multinational firms, has its own special limitations, failing to distinguish basic forces of the development of capitalism in the firm's operating environment and reducing the firm itself to "adoptive stimulus-response systems" or some similar idealized abstraction. Reality is replaced by universal a historic concepts, which then become the reality of the theoretician.

The most detailed critique of conventional location theory has come from holland (1976a), whose particular concern is with the failure of theory to account for uneven economic development. He raises numerous technical points, but also assails the entire neoclassical model in economic (and location) theory for its supposedly self-balancing properties. His most severe castigations are aimed at Isard and the regional science school:

It is scientific not in the sense of a physical science, nor even of a social science which attempts to organize and abstract from empirical evidence,

but in the sense of intuition and hunch combined with unrealistic and inapplicable analytical techniques [Holland, 1976a, p. 26]. If anything, its influence has been perverse, obstructing relevant theory, technique and policy. [1976a, p. 18].

Impressive by virtue of its mathematical sophistication and rigor, the ultimate achievement of Isard and his associates_their so-called general theory_obscures far more than it reveals of the nature of real (capitalist) society.

Despite the severity of Holland's critique, his actual contribution goes little further than that of the organizational decision-making school, in his recognition of the dominant role of "big-league" firms in shaping the world economy. Theories of regional imbalance are shown to have greater empirical relevance than spatial versions of neoclassical equilibrium theory, but a broader perspective is required. This, in the words of Dunford (1977, p.513), is

to consider the constituent regions to be part of a single social formation dominated by the capitalist mode of production, and to interpret the characteristics of regions as products of the historical laws governing the spatial organization of industrial capitalism.

In short, what is required is the perspective of Marxism.

The most thoughtful and persuasive arguments in this direction have been provided by Massey (1979) in the latest version of a critique of industrial location theory that first appeared in 1973. Massey underlines the weakness of existing theory and points the way toward an alternative using Marxian economics.

The basic point of Massey's critique is the preoccupation in classical location theory with an abstract firm without effective structural relationships to the rest of the economy. At first sight this is a familiar argument, elements of which have already been observed in the work of the behavioral and organizational schools. However, Massey (1979, p. 59) sees the new approaches as being similarly concerned with an idealistic model of the individual firm:

An attempt is made to distill common factors from heterogeneous real phenomena, and then use these to develop a formal model representing the "essential firm, " abstracted from the particularities of history_and of geography.

Thus, the simple conception of the single-plant firm which accorded with the reality of nineteenth-century industry has been replaced by a model of the firm as a complex organizational structure, so as to come to terms with the reality of the mid-twentieth century. The objection is not to the evident truth of the recognition that such a change in industrial organization has taken place but to the way in which behavior is conceptualized. On the one hand there is the observed behavior of individual firms, on the other the theory or model of the firm in general. The problem is that the abstract model fails to explain elements of individual behavior, attributing them rather unhelpfully to "additional factors." There is a fundamental epistemological or scientific-procedural objection to approaches that, in effect,split behavior into that which conforms to the (given) ideal-type firm and that which is left unexplained as a deviation from the ideal. What Massey seeks is a framework for explanation that incorporates the variety evident in real-world locational behavior: "Instead of striving towards an abstract model of behaviour from which historical and individual variation has been removed, the attempt should be always to analyze behaviour in its real historical context" (Massey, 1979, p. 61).

What she offers is the structural approach. The aim is "to reformulate the method of explanation in industrial-location theory_to interpret behaviour as the product of the overall structure of the system in which the firm is set, and of its place within that structure" (Massey, 1979, p. 70).This raises the crucial question of the connection between spatial and non spatial economic phenomena. Massey continues:

Industrial-location theory cannot itself account for spatial behaviour, or for historical changes in such behaviour, because the ultimate causes lie beyond the realm of spatial analysis. It is as an aspect of a process geared to production and [capital1 accumulation that location factors and spatial differentiation are taken into account in company decisions, and from which differential spatial impacts result. And it is in this a-spatial_rather than in the simply spatial_sphere that the primary causal links exist between the wider economy and the individual company.

Again, there are echoes of some other contemporary approaches in this.However much they may be predisposed toward a general model of the firm, those who adopt the organizational decision-making perspective do claim to seek an understanding of the firm's behavior in relation to an external environment. The difference is to be found in how this environment is conceptualized and examined, with the firm and its location problem as an integral feature of the wider whole.

The best way of explaining the structural approach further is by summarizing the procedure adopted in the single major research project undertake n date (Massey, 1977, 1980; Massey and Meegan, 1979). The empirical focus is the spatial implications of change in the electrical engineering and electronics industry in Britain in the second half of the 1960s. Rather than looking immediately at the behavior of individual firms, the first task is to isolate the "determining structure." This is to be found in specific conditions prevailing in Britain's capitalist economy, namely a fall in profitability and lack of international competitiveness. Different firms in different sectors responded in different ways to the national economic crisis, depending on the actual circumstances involved.

The sector making heavy electrical machinery may be taken as an example. There was severe overcapacity, in the form of fixed capital and labor,which constituted a drain on profits. There was also a need to cut costs in the interests of competitiveness. The problem of excess capacity was solve din part by plant closures. The impact of this on profitability is explained viaM~rxinn "rr~nomir r~teonriPc ThP rAf~ ~f nrAfit I^i ;r AAd~ AA~ -.

Plant closures reduce v and c, and hence increase the rate of profitp providing that surplus value is not reduced in the same proportion as capacity. The actual plants closed were those where productivity was lowest, i.e., where least s was being generated per unit of capital advanced as c or v. Cuts in costs were also made by standardization of production and by savings in labor costs_the latter achieved by new factories in cheap-labor locations(where v is reduced and therefore rises).

An important feature of Marxian analysis that distinguishes it from contemporary behavioral perspectives on location decision-making is the emphasis placed on strict profit-seeking objectives. The competitive process is seen as leading to a fall in the general rate of profit, forcing firms to find whatever means may be available to keep up profits and accumulate capital in the interests of survival. Variations from place to place in the cost of labor and other inputs is one source of variation in profitability. As Mandel(1978, p. 75) puts it, "the problem of the extension of capital into new realms of production_whether technical or geographical_is ultimately determined by a difference in the level of profit. " There is more than a hint of concurrence with the profit-maximizing objectives of that discredited ideal type economic man. But the fact is that, if capitalism is a competitive struggle (even in its contemporary monopoly stage of development), those who survive will do so by virtue of some advantage with respect to making profits. No existing technical framework is more appropriate for the identification.

The main features of Massey's structural approach are summarized in figure 5.9. The analysis begins from the top, not at the level of the individual firm. Conditions in the British economy (itself responding to forces in the international capitalist system) affect the various industrial sectors, and the impact is traced down to the individual firm and its locational strategy.Plant location is an integral part of the general problem of changes that maybe required in the production process and the organization of labor. The actual behavior of the individual firm takes on its meaning in the broader economic context that the structural approach seeks to reveal. Working up from the bottom can explain neither the individual elements nor the system as a whole.

Massey and Meegan's study of the electrical engineering industry brings anew perspective to the analysis of the location of specific industries. The same approach can be applied more broadly, relating changes in industrial location to other elements of the economic, social and political system. Thesis exemplified by an examination of regional development problems in france undertaken by Dunford (1979). He attempts to show how particular spatial forms of economic activity are associated with different stages in the development of capitalism:

It is possible to identify various phases in the development of capitalist economies associated with stages in the development of the productive system and in the organisation of the labour process and division of labour. The structure of the productive system is characterised by the dominance of particular branches of production and specific relation between industry and agriculture and this in turn is closely associated with the u<of specific kinds of spatial differentiation and specific forms of territorial development. But these phases are also characterised by particu~lar sets of class relations and class alliances and particular forms of integration of local populations with the central state reflected in different local social and political arrangements [Dunford, 1979, pp. 81-82].

In France, the decline of precapitalist modes of production in agricultural accelerated rural depopulation and the concentration of employment in metropolitan centers. But the rising cost of labor and infrastructure in these regions, combined with a restructuring of leading industrial sectors, led to,new stage of development and hence to new spatial forms. The main features of this present period are "deindustrialization" and concentration of the tertiary sector in the metropolitan regions, the implantation of modern~capitalist enterprises in rural areas, and the development of complex unit of production. Dunford (1979, p. 94) emphasizes the importance of labor a!a location factor, associated with the new forms of industrial organization in which different functions are performed at different levels within a hierarchical structure:

These levels and activities can be separated from one another and, given an adequate development of transport and telecommunications, locate din different places where their different infrastructural and labour requirements can be met, increasing the rate of surplus-value. It is these structural changes and these possibilities which enable these companies to make use of the spatial differentiation in the value of labour-power, in the levels of skill and in the degree of organisation of the labour movement.

The state is viewed as facilitating the spatial reorganization required by the new stage in the process of capital accumulation, through regional development programs and other measures designed to assist the development of new industrial complexes. The chief merit of the Marxian approach adopted by Dunford is its breadth, which permits industrial location to be analyzed as an integral part of the totality of economic, social and political process,from which uneven regional development arises.

As the structural approach to industrial location analysis is still in its infancy, it is premature to pass firm judgments. However, it can be seen to address certain fundamental defects in existing theory as a framework forth e explanation of actual location choice and change. Although developed within the context of capitalism and adopting the economic analysis devised by Marx for the specific purpose of explaining how capitalism operates, the structural approach has a broader applicability. The idea that individual elements must be understood within some broader determining~ structure is equally appropriate to any mode of production. Indeed, this is the kind of argument that advocates of the systems approach have been putting forward for some time. The crucial distinction is that the structural approach does not seek general knowledge of all types of systems, nor does it seek to impose some ideal-type model of a system on real economic structures. Its objective is that of understanding the specific historical circumstances of the human practice of making a living, one integral element of which is industrial location.

In conclusion, it should be recognised that the contribution of the structural approach, like that of the other contemporary approaches discussed in this chapter, is in the realm of positive rather than normative theory. In part Two the focus of attention returns to the question of the "best" location, or optimality, on which traditional theory still offers the most useful general guide. But when problems of uneven development arise again, (In part Four) we shall find that the structural approach reasserts itself_not only in the interpretation of the origins of regional economic development patterns but also in understanding the constraints imposed on the implementation of optimizing models and other attempts to improve our sorry reality in the practice of planning.