An appraisal of the development pole concept in regional ...

16
Environment and Planning A, 1974, volume 6, pages 291 -306 An appraisal of the development pole concept in regional analysis D Todd Department of Geography, London School of Economics and Political Science, London, England Received 15 March 1974 Abstract. A considerable amount of recent literature on development pole concepts contains criticism that there is no adequate unifying 'theory' to underpin the concepts. This paper attempts to trace the conceptual evolution of development pole theory in order to isolate some of the causes of theoretical confusion. It concludes with a plea for integration of both the sectoral and spatial aspects of polarization. 1 Introduction Development pole theory has received a great deal of criticism of late, especially in relation to its slack theoretical constructs and equivocal interpretative posture. Indeed some students are of the opinion that ambiguities in the interpretation of pole theory preclude any identification of basic 'theory' at all (Moseley, 1973a). Nevertheless the essential premise of polarization in space and its accompanying process responses is too valuable to be lightly discarded. Accordingly this paper will attempt to trace recent contributions to development pole analysis and explore the implications of some of the results. A primary requirement of any evaluation of development pole theory is the elimination of misconceptions arising from variations in terminology. Confusion over the implications of the terminology has been endemic throughout the literature, and the terminology is badly in need of some standardization (Hansen, 1967), A simple expedient adopted here is to designate all nonspatial polarization concepts as development poles, while reserving the term development centre for explicitly spatial interpretations. The underlying processes of concentration and the concomitant spin-offs are fundamental to both designations. Unbalanced regional devel opment To place development pole theory in perspective requires an elaboration of its origin in the nineteen fifties. The concept arose out of dissatisfaction with equilibrium or balanced notions of regional development. The seeming inability of balanced regional growth theory to provide operational regional-development programme formulations resulted in support being diverted to notions of unbalanced growth as possibly being relevant to the regional problem. Hirschman (1958, p. 184) in his critique of balanced growth theory noted that, in addition to disparities in sectoral growth, there is also an inequality of growth geographically. The proposition that development appears to concentrate punctiformly in space is intuitively obvious. Indeed this supposition is now almost universally accepted as the basic tenet of planned development strategy making. The full endorsement of unbalanced development planning was given at the United Nations conference on industrial location and development at Minsk, 1968 (UNIDO, 1969). Essentially the unbalanced notion assumes a sequence of states of disequilibria that reflect varying combinations of investment conditions between overhead capital (OC) and direct productive activities (DPA), see Hansen (1965). It makes provision for a lag effect and an element of choice. Thus, for instance, if regional decision makers initially opted for providing mainly OC, the sequence of 'development via excess

Transcript of An appraisal of the development pole concept in regional ...

Environment and Planning A, 1974, volume 6, pages 291 -306

An appraisal of the development pole concept in regional analysis

D Todd Department of Geography, London School of Economics and Political Science, London, England Received 15 March 1974

Abstract. A considerable amount of recent literature on development pole concepts contains criticism that there is no adequate unifying 'theory' to underpin the concepts. This paper attempts to trace the conceptual evolution of development pole theory in order to isolate some of the causes of theoretical confusion. It concludes with a plea for integration of both the sectoral and spatial aspects of polarization.

1 Introduction Development pole theory has received a great deal of criticism of late, especially in relation to its slack theoretical constructs and equivocal interpretative posture. Indeed some students are of the opinion that ambiguities in the interpretation of pole theory preclude any identification of basic 'theory' at all (Moseley, 1973a). Nevertheless the essential premise of polarization in space and its accompanying process responses is too valuable to be lightly discarded. Accordingly this paper will attempt to trace recent contributions to development pole analysis and explore the implications of some of the results.

A primary requirement of any evaluation of development pole theory is the elimination of misconceptions arising from variations in terminology. Confusion over the implications of the terminology has been endemic throughout the literature, and the terminology is badly in need of some standardization (Hansen, 1967), A simple expedient adopted here is to designate all nonspatial polarization concepts as development poles, while reserving the term development centre for explicitly spatial interpretations. The underlying processes of concentration and the concomitant spin-offs are fundamental to both designations.

Unbalanced regional development To place development pole theory in perspective requires an elaboration of its origin in the nineteen fifties. The concept arose out of dissatisfaction with equilibrium or balanced notions of regional development. The seeming inability of balanced regional growth theory to provide operational regional-development programme formulations resulted in support being diverted to notions of unbalanced growth as possibly being relevant to the regional problem. Hirschman (1958, p. 184) in his critique of balanced growth theory noted that, in addition to disparities in sectoral growth, there is also an inequality of growth geographically. The proposition that development appears to concentrate punctiformly in space is intuitively obvious. Indeed this supposition is now almost universally accepted as the basic tenet of planned development strategy making. The full endorsement of unbalanced development planning was given at the United Nations conference on industrial location and development at Minsk, 1968 (UNIDO, 1969).

Essentially the unbalanced notion assumes a sequence of states of disequilibria that reflect varying combinations of investment conditions between overhead capital (OC) and direct productive activities (DPA), see Hansen (1965). It makes provision for a lag effect and an element of choice. Thus, for instance, if regional decision makers initially opted for providing mainly OC, the sequence of 'development via excess

292 DTodd

capacity' would ensue, in which DP A use of, and response to, the overhead facilities would lag behind the optimal balanced combination. Alternatively, 'development via shortage of OC' could be implemented whereby OC facilities would be provided subsequent to effective demand from already existing productive activities. This is in direct contrast to the balanced growth notion, which maintains an expansion path with constant proportions of OC and DPA throughout the time sequences. Clearly unbalanced growth works via a system of shocks, and fails to reach equilibrium except by an implicit assumption of a future optimum developed state.

The theory of disequilibrated growth is strictly concerned with sectoral expansion, but beoause the system assumes unequal rates of growth, the idea of polarized growth is readily assimilated into it. Polarized growth can be conceived in both a sectoral or a spatial series, or a combination of the two. Furthermore, as a process, polarized growth also possesses a temporal dimension, thereby culminating in a spatial-sectoral-temporal proposition. In general the spatial process includes notions of spatial juxtaposition economies (or spatial external economies), rapid and efficient diffusion of innovations, and a limited friction of distance factor which reduces the problem of gathering inputs. The sectoral process is concerned with deriving optimum sizes of plants within the 'pole' as a result of positive advantages emanating from external economies, and internalized economies resulting from enlarged markets. The final component of polarized growth, the temporal process, describes the successive stages in which these relationships interact.

Concepts of polarization can be applied at both general and partial levels. In other words it is conceivable to devise a general disequilibrium theory of development poles applicable to national and global perspectives, while simultaneously producing a partial disequilibrium notion of polarization particularly relevant to problem solving at the regional level. This partial view, or 'development centre' idea, readily fits Nichols (1969b, p. 193) interpretation:

"... an urban centre of economic activity which can achieve self-sustaining growth to the point that growth is diffused outward into the pole region and eventually beyond into the less developed region of the nation".

However, immediately a problem of scale comparabilities becomes apparent. Owing to the levels of scale aggregations, the two notions of development poles will of necessity adopt differing slants. The general theory will be concerned with a global or supranational systems interpretation of process, in which comparative indices of regional development will be a prerequisite. Partial theory is oriented towards planning decision making at the regional level, and is thus only indirectly concerned with the communality of interregional scales. Accordingly, because of the difficult conceptual problem of handling different dimensions of scale, most of the emphasis in the theoretical evolution of development poles has been with the partial applications in mind.

Sectoral development poles The initiator of 'growth pole' methodology was Francois Perroux, whose work was undertaken in the late nineteen forties (for a summary see Perroux, 1955). Perroux's concern was predominantly with sectoral interrelations in which a concentration of economic activity stimulated further centripetal movements to the pole. The movements of a field of forces to a centre or node (pole de croissance) were conceived in abstract terms and only indirectly as a phenomenon with spatial attributes. Once the pole was established, a dynamic process of concentration or 'polarization' could be partly offset by centrifugal influences disseminating some activities from the core. It is important to note that the polarization process theorized by Perroux is

An appraisal of the development pole concept in regional analysis 293

conceptually different to Hirschman's geographical polarization process, through which two locational extremes induce concentration from the space between them.

To ensure dynamic conditions, the Perrouxian system required one industry or firm within the pole to adopt the mantle of dominance or leadership. Accompanying the 'dominance' factor are the ancillary requirements of large size in the industry concerned and a high degree of interindustry relationships. The dynamic disequilibrium and dominance conditions owed much to work produced by Schumpeter (1949) on the decisive role of innovations promulgated by large-scale industrial enterprises. In Schumpeter's terms, the scale of industry and the rate of innovation generation were highly correlated and provided the theoretical underpinning for the dominant industry role. The extent of induced development in the associated economic activities followed from both the scale and the degree of interdependence.

The process worked through a system of interindustry linkages. A dominant or key industry would regulate the expansion of industries supplying inputs to it (a backward linkage effect), whilst inducing demand industries to grow according to the quantity and kind of intermediate products supplied to them by the key industry (a forward linkage effect). Empirically the degree of the dominance, or 'propulsive' effect, of the key industry could be gauged from the triangularization of an input-output table. Dynamism is maintained through successive rounds of expansion of the dominant, or 'motrice', industry disseminating multiplier effects to all the linked industries. After the motrice has stimulated demand past a minimum viable threshold, new technologically linked industries arise in response to the enlarged market operating within the pole's ambit.

Some criticism has been levelled at Perrouxian theory because of its cavalier treatment of the geographical dimension, while Lasuen (1969( makes the point that Leontief-style input-output tables—the cornerstones of Perrouxian theory—are static and therefore are irreconcilable with a concept purporting to be dynamic. Additionally, the dynamic aspect of the theory is deficient because no indication is given of the origins of the propulsive industries in the first place. The notion of dominance has not been without its critics either. Blaug (1964) claims that it is too elusive to be quantified, while Hansen (1967) indicates that it is not an essential requirement for regional development. At best, advocates of dominance can claim it as merely one conditional form of development out of several alternatives, at worst it is an equivocal shadow concept.

Lasuen suggests that the overemphasis that Perroux, and his proteges of the 'French school', placed on the multipliers was derived from an input-output view of polarization and it has meant a neglect of the innovative aspect, and especially the centrifugal effects, of diffusion of development away from the pole. Nevertheless the French school has attempted to devise optimum complexes of interrelated industries capable of providing a development package for regional planning, see for example Paelinck's (1972) viable minimal investment industrial complex. This reflects their continued interest in grappling with the economic forces involved in regional development, as well as their interest in construing spatial notions of development centres in their attempts to decentralize activities from metropolitan Paris.

Overconcentration on input-output analysis by Perrouxian economists deprived pole theory of further development into its other subconcepts: external economies and location theory. Although Perroux recognized the relevance of Scitovsky's (1954) pecuniary external economies (whereby a firm's output is dependent upon the output of other firms, as well as its own, for reducing its costs), he failed to stress the spatial implications of such externalities. Spatial juxtaposition economies have been recognized by location theorists since the days of Weber and have been provided with a definitive content by Hoover (1948). In essence they imply a saving in cost

294 DTodd

between successive stages of production as a result of the adjacent location of plants. Thus they complement nonspatial ideas of internalized scale economies resulting from specialization and optimal operating practices ('principle of multiples'), and pecuniary external economies, both of which are implicit in Perrouxian theory.

Two major forms of spatial juxtaposition economies have been distinguished: localization economies, or those savings resulting from the proximate locations of a number of firms with affinity linkages; and urbanization economies, that is the savings in overhead costs resulting from common usage of basic facilities by a multiplicity of distinct firms. Thus these economies appear amenable to investigation from the point of view of location theory.

Reformulation of the pole framework to make provision for technical links and the recognition of spatial externalities has given rise to the idea of industrial complexes. Hermansen (1972) defines an industrial complex as "... an ensemble of technologically and economically interconnected industrial units usually located on a given territory". In response to the obvious need for a more rigorous approach to regional problem solving, Isard et al. (1959) attempted to make operational the industrial complex concept.

The subsequent industrial complex analysis (ICA) was used in a normative project designed to procure the optimum mix of industries for the island of Puerto Rico. Isard maintained that the advantages of erecting an industrial complex revolved around the economies in having investment concentrated at one site, the operating savings resulting from agglomeration economies, and an environment conducive to personal contact linkages and diffusion of technical know-how. This latter aspect fits readily into the Perrouxian notion of the need for an innovative capability within the pole in order to ensure self-sustaining growth. The analysts went about devising the optimum mix of industries by building an activity matrix representing actual flows of physical resources between various plants. The matrix was designed to indicate the possibilities of substitution between the different processes and industries in order to alleviate supply 'bottlenecks'. From the specified quantities in the interactivity matrix, corresponding costs indices could be obtained by the application of standard location-theory methodology. Furthermore certain constraints could be introduced to handle nonlinearities in production coefficients and spatial externalities.

Unfortunately, although localization economies can be approximated from comparative costs data, Isard found no way of quantifying urbanization economies. Also the actual spatial links between plants are merely implied in the overall benefits of complex agglomeration and not differentiated in actuality. Therefore, while Isardian ICA does consider spatial externalities and the transfer cost differential part of location theory, it treats them in isolation and at best regards the spatial element implicitly rather than explicitly. Besides this, the analysis is based on static comparative indices, which have proved difficult to adjust to a dynamic framework (that is there is little provision for feedback of the multiplier-accelerator effects upon the interactivity system itself). However, ICA has been greeted favourably (Darwent, 1969) and applied to planning real poles in south Italy (Newcombe, 1969). Further use of the method has been frustrated by the lack of data necessary for the adequate application of the analysis—the paucity of complex-engineering and costs data providing a real barrier to all but those commanding expensive project teams and having access to copious government and corporation statistics.

Spatial development pole concerns Darwent (1969) made a distinction between growth poles of the kind discussed above and 'growth centres' existing within a geographical dimension. The first kind were involved with economic processes and only indirectly concerned with space, whereas

An appraisal of the development pole concept in regional analysis 295

growth centres attempt to relate the effects of spatial interaction to the socioeconomic processes involved in development. We have noted that ICA, although it stresses the need for given locations in which to operate rather than abstract economic space, does not involve itself in concepts of spatial interaction. Boudeville (1957) moved the conceptualizing one stage further by adapting a regional-impact study (based on a steel-industry pole in Brazil) to analyze the spatial distribution of inducement effects imposed by the key industry on the other industries (in this case the responses of the fabricating industry to expansion in primary metal making).

Theorizing from the lessons learned in this empirical study, Boudeville (1966) has succeeded in placing the polarizing idea into a framework of spatial typologies. Briefly this entails three different kinds of geographical spaces, or regions, each related to a scale of economic pursuit. A homogeneous space (region) is essentially descriptive, it possesses intuitively common attributes. The microeconomic component of such a region would be the establishment of markets for a firm, whilst a macroeconomic consideration might be the study of regional disposable labour. A polarized region is demarcated by centripetal flows to a centre, with microimplications, such as the relations between a parent plant and its branch associates, and general implications in the definition of market areas of different poles. The third and final regional type is a programming space, defined according to the policy intentions of its planners. Activity analysis for the firm would be a typical microeconomic interest, whereas overall development of the river-basin kind (exemplified by the Tennessee Valley Authority) would fit the macroeconomic slot.

It is apparent that Boudeville's regionalization ideas are directly comparable with the regional typologies developed by geographers, and thus provide a theoretical bridge between processes of economic growth and geographic space. Obviously the polarized region has greatest affinity to Perroux's pole, but the others are applicable too; homogeneous regions as the backcloth for establishing effective demand potential for a plant's products (where the plant is to form part of a pole), and programming regions as the empirical space in which the policy instruments needed to make a pole operational, would be employed.

The similarity between polarized space and the regional concern of the geographer has effected a transformation of development pole theory into a spatial framework including central place concepts. Indeed polarized regions and central places have much in common, especially the idea of a hierarchy of nodes capable of being demarcated by geographical boundary-searching techniques. From such a methodological linkage Hermansen (1972, p.29) defines a polarized region as "... a heterogeneous continuous area localized in geographical space, whose different parts are interdependent through mutual complementary and interplay relations around a regional centre of gravity".

In defining a geographical base for poles through central place concepts, pole theory is exposed to further influences from location theory. A deficiency of ICA (incidentally affecting pole theory) was that it considered only certain aspects of location theory. Isardian complex analysis largely absorbed Weberian location principles with their predominant emphasis on supply relationships and relative neglect of demand components. Hence, by default, ICA tended to consider the demand side of development largely in nonspatial terms, with a concomitant reduction in the spatial relevance of the methodology. On the other hand, central place theory, in contrast, has shown a marked preference for the demand side of analysis at the expense of supply considerations. Therefore it is logical to assume that a tie-up of central place notions with the detailed constructs of pole analysis should theoretically produce remunerative results. If the integration could be successfully enacted, pole theory would contain a locational analytic component

296 DTodd

capable of processing such facets as transfer cost differentials, functions expressing the spatial distribution of demand, and elements of agglomeration economies.

Unfortunately central place theories possess intrinsic characteristics that conflict with the requirement specified above. At a simplistic level they are designed around two different modes: the Christaller type service-centre functional hierarchy, and the Losch-style 'economic landscape' composed of a quasi-random mesh of industrially oriented centres. Up to the present time reconciliation of the two types has not been satisfactorily achieved because the Christaller model is aggregative, and hence deals with successive hierarchical orders of distributions of products, whereas the Losch model is nonaggregative and is essentially concerned with spatial separation. Therefore, although a Losch-style model could be grafted onto a pole framework and be designed to assess the geographical centrifugal effects of the complex, much of the associated induced effects on servicing industry would be neglected owing to the omission of a Christaller-style analysis.

Furthermore, empirical justification of central place theory is often difficult to establish. Deviations from hypothetical patterns are usually the rule rather than the exception, and the causes of these deviations can be very pertinent to pole theory, for example the asymmetric location both of physical resources and transport media. Other defects of central place studies are too well-known to be repeated here, but it is worth emphasizing that central place theory is a static equilibrium concept and thus directly at odds with the motivating theory of development poles.

Dynamic spatial pole theory Spatial ramifications of dynamic disequilibria were initially investigated by Myrdal (1957) and Hirschman (1958) working independently. As a by-product of their concern with polarizing tendencies within a region, they became interested in the degree of growth disseminated from the centre to the regional peripheries. Myrdal conceived of regional disequilibrium in terms of a systems approach. A process of 'cumulative causation' enhanced concentration at the centre as a result of positive feedback and the dynamic requirements of self-inducing growth. Factor movements are attracted to this core, with detrimental effects on the remainder of the regional economy. Hirschman arrives at a similar conclusion, in which economic growth is 'polarized' at limited areas in space, but differs in his rejection of a permanent concentrating mechanism. Both claim that some spill-over effects influence the residual regional area, and that this may be in the form of either positive or negative contributions. It is the interest in the geographical distribution of polarizing externalities which differentiates these concepts from Perrouxian theories.

The factor movements which constitute the variables in these dynamic processes can be transformed into geographical processes, for example the migration of a labour force, the intensity of capital flows over distances, or the quantities of material the inputs subsequently demanded. As such they readily fit into the geographer's realm of regional differentiation techniques and models of central place structures. Hirschman-style models deserve a certain amount of acclaim for providing a relation to geographical processes, although his interpretation of affairs cannot be accepted without reservation. The polarizing process established by Hirschman and Myrdal is really reserved for a general development scheme, that is, one that explains differentiation of regions over time within a supraregional framework. Therefore their theories have corollories, such as societal changes, administrative responses, and political feedback, which are geared to a national viewpoint. When cumulative causation is disaggregated to the individual regional level, the process mechanism, with all its attributes, loses some reality because national-level variables are either absent or blunted.

An appraisal of the development pole concept in regional analysis 297

If we now assume away the scale problem, dynamic disequilibrium growth concepts of this kind are extremely valuable for isolating the mechanisms that provoke polarization and ensure its continuance. In so doing, such concepts provide pole theory with a motivation force (unexplained by the French school) of a social and political nature as well as the traditional economic mechanism. Furthermore the notion of 'dominance', that keystone of Perrouxian theory, is not a prerequisite of growth to Myrdal or Hirschman.

After the exposure of development pole theory to a spatial perspective, geographers and planners have responded with several contributions. At the regional level, for example, Hansen (1972) establishes a regional development policy for the United States, based on a three-tier hierarchy of growth centres: regional, primary, and secondary. The criteria for establishment are based on the degree of servicing provided by the centres (extracted from central place theory), and also on the geographical extent of the labour-commuting areas. In an earlier paper Hansen (1970) argues the case for the designation of his growth centres with reference to the varying sizes of cities. Thus pole theorists are finding themselves entwined in the debate about optimizing city-population sizes and the urbanization process in general. The optimum city-size question is still far from being resolved and currently rests on intuitive judgements of the relative trade-offs between social costs and private benefits (see for example Thompson, 1965; Alonso, 1972). On the other hand, concern with urbanization processes has prompted pole theorists to consider the significance of innovation generation and the role of spatial diffusion. Innovations are vital to the self-sustaining growth mechanism, and indeed this fact was recognized at an early stage by Schumpeter and Perroux. Thomas (1972b, p.70) feels innovations are particular relevant to development pole ideas and cites three reasons for this;

"(1) by improving the competitive position of industries in which they are adopted, (2) by making it possible through the development of new products to establish new industries in the growth pole, and (3) by making it possible to establish industries that for economic reasons could not have been previously located in the pole".

The first condition is necessary to ensure the continued viability of the pole, whereas the others lead to the expansion of the pole into a fully integrated complex. In other words a process is assumed in which the initial key plants of a pole are steadily supplemented by a complex based on plant interdependence and spatial externalities.

Alternatively, Berry (1972) uses the diffusion process to illustrate the centrifugal effects emanating from a pole into its hinterland. Through a model of 'hierarchical diffusion' Berry intimates that diffusion of an innovation takes place by means of a hierarchical filtration process. In this manner an innovation is initially adopted in the contiguous zone around a major centre and then sequentially accepted by other centres in relation to a descending order of size (allowing for some distance decay constraint). He regards the filtering of diffusion as accounting for the spread of positive feedback away from the pole, and also as a medium for maintaining the dynamism needed for self-sustaining development.

The analysis of polarization in spatial terms, concentrated on determining process relations and, more specifically, isolating the economic mechanism that regulates concentration at a centre or distributes spread effects to the region adjoining the pole. Little attempt was made to describe the spatial form of the polarization process. Odland et al (1973), however, have defined a model which formulates the characteristics of spatial polarization. If Z represents polarization intensity at some point with coordinates x and y, defined over an area S, then a matrix Z(x,y, t) of locational coordinates can be produced for differing time periods. Z is differentiable at least twice over S and continually with respect to t (where t is a subcomponent of

298 DTodd

time interval T), thus providing a spatiotemporal surface. Convolutions in the surface elevation focus upon locations of high polarization intensity. In turn, the polarization intensity is dependent upon the sign of the derivative. Growth occurs when

dZ -Q- > 0 where v j G S , t G T.

To allow for negative growth, commensurate with the decline or obsolescence of a pole, a 'less than' sign can be substituted into the above relationship. Distance decay is implicitly introduced when S is disaggregated to s, representing the euclidean distance from the pole coordinate location to a set of locations defining the regional extent. If we presume that the intensity of development in S grows over time, the change relationship becomes

d2Z T ^ - < 0 for all tu ..., tn\ Su ..., Sn.

If this relationship is substantiated, polarization will occur at a node (that is the intensity of dZ/dt will decline with increasing distance from the pole), if not, the rate of development will intensify with greater distance away from the node. The model can be augmented to assess polarization occurring at several locations simultaneously. Empirical work using employment data for US cities tentatively suggests growth is tending to concentrate upon them through time in a positive relationship with increasing distance from other centres. As a complement to this work, attempts have been made to distinguish the form of the 'spread' effects resulting from the polarization process. Robinson and Salih (1971) and Moseley (1973b) have used trend-surface analysis to identify growth surfaces around poles. Furthermore various hypotheses referring to the configuration and intensity of these surfaces have been tested and verified. The relationship between interaction and growth of poles may be a principal explanatory variable in the functioning of 'development axes'. The idea of a corridor or axis of development was coined by Pottier (1963) to describe seeming linearities in the spatial expression of growth. Essentially this claims that any growth spin-offs from poles will take place along the transport arteries linking the various poles. The concept borrows, from international trade theory, the notion of enhanced interaction between two simultaneously growing centres, and assumes that expansion in trade flows will result in scale economies acting upon the transport network. Such economies are inducements for attracting any activities originating in response to the poles' demands for hinterland supplies. This process would work cumulatively, initial growth stimulating further growth and so on. On the whole, the concept appears promising because of its apparent empirical verification and also because it fits neatly into the 'transportation principle' as designed by central place theorists.

Pottier's introduction of a transport or network consideration into polarization theory has aroused further interest in the application of network concepts to this field. Campbell (1972) for instance has shown how graph theory can be used to pinpoint plant interdependencies within a pole. Digraphs (graphs with edges having an irreflexive condition, thus excluding loops or parallel edges) are constructed, displaying the network of spatial linkages between plants. From these, measures of centrality and connectivity can be gauged which describe the closeness of the links and which, in so doing, provide an indication of the transfer costs, contact relations, and external economies. The Campbell model is useful for stating the impacts of a pole on its environs by means of the stimulus that the pole industries will transmit through their linkages to peripheral industries. The only drawback is the static nature of the model, but the possibility of simulating a dynamised network should not be ruled out.

An appraisal of the development pole concept in regional analysis 299

Towards an assimilation of process analysis in development pole theory In the foregoing section we have noted the extent to which Perrouxian pole concepts have been integrated with some of the more general tools of spatial analysis. Unfortunately, in reconciling the two perspectives, much of the analytical rigour of impact techniques was lost (for example in transfer costs or multiplier studies employed in ICA). Therefore the problem becomes one of devising a development pole methodology which contains a spatial framework but which also includes the kind of process considerations focussed upon in complex analysis.

Isardian ICA is not without those pitfalls implicit in an engineering process approach, such as large demands for a planned, government investment project to fulfil the stringent normative requirements. The precise web of industrial interdependencies are mostly devoid of any degrees of freedom, and are therefore somewhat inappropriate to that private development planning which presupposes a greater allowance for behavioural response and individual initiative. It thus transpires that pole planning must adapt engineering process analysis to incorporate a wider range of alternative strategies from which the decision maker may select.

Possible remedial action for such purposes could be brought about through the use of spatial linkage analysis. Linkages per se have long been utilized to represent transitional stages between economic interdependence of the interindustry kind. However, spatial linkages can be construed as a specialized variant, and defined as a four-group typology (Wood, 1969):

(1) Process spatial link: the movement of commodities between different plants as stages in the manufacturing process (including subcontracting).

(2) Servicing spatial link: the supply of equipment and maintenance services undertaken by firms external to the production plant.

(3) Marketing spatial link: firm interdependence for the purpose of product distribution (including wholesaling and transportation).

(4) Financial spatial link: ties with financial and business advisory services.

In order to grasp the full significance of the typology, any links should be construed as interplant relationships obeying the properties of distance decay functions. Transformation of an industrial organization system into a spatial linkage network is exemplified by figure 1. This suggests the scope of a single plant's interrelations and corresponding geographical links. Complexity is compounded when the scale is increased to industry level, for not only are input and output points multiplied, but intraindustry dependence must be taken into account (that is relations between like plants of the same industry).

Input linkages

sources of materials

semiprocessed products with origins

Transforming linkages

service, marketing, and financial linkages Xu X2, ...,Xn

internalized process links, for example conveyor relations

externalized process links (subcontracting) Ei, E2, ..., En

Output linkages

final demand outlets FuF2,...,Fn

intermediate outlets for further processing A,/2, ...,/„

Figure 1. Spatial linkage network for an industrial organization system. Mu ...,Mn; Si, indicate point locations.

, Sn; etc.

300 DTodd

It is evident that the nonengineering process linkages specified in the typology have a great affinity with the mechanisms of external economies. Indeed, derivation of an index of interplant dependence by means of spatial linkages can be used as a surrogate for benefits emanating from positive externalities between industrial plants. Thomas (1972a, pp.93-94) remarks on the paucity of work relating to spatial linkages, even though he regards this as a critically important area for study. For example, spatial linkages can be used to sharpen-up the investigation of impact studies concerned with the extent of multiplier leakages out of the regional economy. Thus if a region's industries are linked to supply plants and demand outlets outside the region, it may be a better idea to invest in industries, in other regions, who use inputs from the development region, and thereby have a greater multiplier impact than direct industrial investment, which may produce only a small multiplier effect, in the development area.

It follows that spatial linkages can fit the requirement for a generalized technical process system within a geographical framework. Tosco (1971) and the New Brunswick Multiplex Corporation (undated) have drawn up process-analytical schemes for industrial complexes in the Mezzogiorno and St John respectively. These schemes provide a series of alternative courses of action from which planners or entrepreneurs can select logical continuations of the initial process. An example of a simplified alternating-process system is given in figure 2.

Methods of this kind contain several advantages when assimilated into spatial development pole theory. As noted earlier it removes much of the rigidity inherent in ICA and gives some room for manoeuvre on the part of the decision maker. For example an entrepreneur can opt for a copper plant and then select from a range of follow up activities, his selection depending upon the investment resources and expertise available. Furthermore the system is self-contained at all stages, this implies that the whole range of plant types is not needed for viability, because each product can be earmarked entirely for final demand (for example the output of a foundry does not require absorption into a bearings plant as intermediate demand, it may be supplied direct to export demand). In addition the system can be conceived of in a dynamic sense. To this end, the 'principle of multiples' can be applied. Thus initially, one iron and steel complex may, at full efficiency, support one foundry and two bearings plants. However, demand stimulation for bearings feeds back through the system inducing the erection of a second blast furnace. But for a second furnace to be fully viable requires that its output be absorbed by its own foundry and two bearings factories. Therefore the complex now contains an enlarged

ancillary servicing facilities

raw materials. iron steel ~ aluminium copper

servicing industries: tooling jobbing machining

t I stockists

1

maintenance industries

stockist

finishing industries: foundries

"forges presses sintering

component industries: bearings bolts springs wire axles brakes

final demand: ' consumer export

Figure 2. An example of a simplified alternating-process system.

An appraisal of the development pole concept in regional analysis 301

steel mill, two foundries, and four bearing units. At this stage the threshold may have been reached for the viable functioning of a press and sintering plant, and thus further plants are 'born' into the system.

• Again, in figure 2, it is possible to assign each of the plants to a specific location. Expansion of the system can be conceived in terms of a continuation of the network of spatial linkages between the coordinate located plants. Effects of spatial juxtaposition economies will tend to contain any subsequent plant 'births'. Thus development poles will result from a clustering of interdependent plants related through close physical linkages.

As an adjunct it is worth commenting on the behavioural aspect of pole theory. A flexible system of growth with allowances for alternative courses of action tends to direct attention to the role of the decision maker. This has immediate implications in terms of corporate organization theory and information communication theory. Various kinds of linkages have been construed to deal with such an interaction, mention of Tornqvist's contact linkages or Moore's (1972) 'business and legal linkages' will suffice here. However, the linkages there are aimed at describing industrial growth and are unrelated to pole theory. Ultimately the planner or entrepreneur implicated in goal formulations for development poles is faced with balancing the following criteria:

"... integration within the pole versus other locations depends on the relative costs of the alternative locations, the degree to which the processes can be separated without incurring higher costs, location of demand, the locational pattern of all plants within the firm in the case of the multi-plant firm (alternative poles), and the location of supply" (Moore, 1972, p.265). All of these five factors are fundamental to normative location theory; the first is

comparable to agglomeration economies; the second is comparable to internalized scale economies; location of demand has affinities with market potential; multiplant considerations overlap with spatial allocation models (for example 'warehousing problem'); and finally, supply origins are basic to Weber's locational triangle. Fortunately, because of the mix of private and public decision makers in pole strategies, we can assume that optimizing attitudes will predominate (that is, those closest to the normative path). Therefore the dynamic concern, whilst appreciative of behavioural deviations from the optimal, will not presuppose any marked tendency away from a planned, efficiently constituted pole(1).

A note on general theories of pole development We have already remarked upon the attempts to attach a central place component to development pole constructs. In partial theory this grafting has not come to much, largely because of the difficulties in obtaining an empirical foundation for criteria of central places comparable to the common denominators which are indicative of pole processes. In other words, at the regional level at least, those processes representative of central place development appear at odds with the spatial growth factors essential for pole functionings.

Nevertheless, it is conceivable that at an aggregate level the two kinds of processes may be compatible. At a primary level of integration, evolution of urban places within the spatial economy may be comparable to evolution of development poles within polarized space. Classically evolution of central places can be generalized into four stages, each representing the four major sectoral divisions of activities.

^This is not to argue, of course, that poles do not result from a multiplicity of random decisions by many entrepreneurs or city fathers. Pred's (1966) work on the historical evolution of US cities stresses the initial random stimulants to growth, which resulted in economic activity polarizing in some cities at the expense of others. However, the text above refers to planned development centres and not to evolving cities in general.

302 DTodd

The initial stage will depict a centre servicing a minimum threshold hinterland which is oriented almost entirely to maintaining the agricultural sector. The servicing area will expand with the excursion of the centre into manufacturing specialization, usually as the result of spin-offs from primary produce processing. The movement up the central place hierarchy evolves cumulatively, with secondary processing inducing growth of specialized service industries. Stage three is attained when the tertiary sector becomes the major urban employer. The final stage is reached as a result of distinctive interaction between high-level centres, and is reflected in a strong emphasis on financial and administrative (quaternary sector) activities.

Evidently the development pole process follows through a similar sequence of evolution—arising around a key industry and expanding by a series of rounds of additional industries which result from successive aggrandisement of market threshold sizes. As a consequence of such apparent affinities Nichols (1969a) estimated that the designation of a pole on a functioning central place will not sufficiently upgrade the local economy. This argument is based on the proposition that the central place hierarchy is largely a one-way filtration structure, in which impulses rarely spread downwards to lower-order centres, but are transmitted across the hierarchy within higher-orders of centres. Thus if a planned pole is established in an intermediate centre of the regional capital kind, hardly any growth impulse would trickle down to regional subcentres, and most of the impact effects would benefit similar intermediate centres outside the subject region. Converted to an interindustry relationship the implication is that the forward linkages are lost to the region, and backward linkages are only minor components of higher-order centres. Nichols (1969b, p.199) concludes:

"The conclusion that might be drawn from this for growth pole policies is that, although it is probably advisable to concentrate investment in that town in a region which has the strongest linkages, there are also advantages to be gained from injecting capital into lower order centres or even the agricultural base, because increases in income in these places will generate strong income multipliers in higher order centres but not the other way round. The injection of supporting capital into lower order centres will probably be even more important in regions of declining heavy industry where out-dated infrastructure militates against a quick response to the influence of a growth pole elsewhere in the region". It follows from this that a central-place inspired growth-pole theory is no solution

to a partial theory of development poles designed to obviate the regional problem. The outcome of a pole process rooted in central place constructs is the eventual dissemination of most development stimulations upwards through the system to the highest-order centres. In order to eradicate this lag in the system, an optimum arrangement would be to steer all investment into the highest-order centres in the first place. Unfortunately dire repercussions would result in which the regions most in need of development effects would receive the smallest growth impulses downwards through the hierarchical filtration structure.

However, there are two major flaws in this theory. The first is the assumption that forward linkages will always predominate over backward linkages, a situation obviously at odds with many intermediate industries and finishing processes. Concomitantly, it assumes that forward linkages will be represented by industries located in the places of final demand, that is the highest-order centres (hence the impulse transmissions up the hierarchy). But the antithesis of this situation is, of course, the industrial complex with industries, benefitting from backward and forward linkages, that are in proximity to each other and able to take advantage of spatial externalities(2).

^ Nichols' assumption is that finishing industries will be 'dominants' located at their markets. However, this proposition is not mandatory, especially in the instance of mutual benefits arising from the interdependence of an industrial complex.

An appraisal of the development pole concept in regional analysis 303

Second, the empirical role of central place hierarchies has not been fully clarified. Berry (1961) questions the thesis that an integrated system of cities within an economic landscape is necessarily conducive to optimal economic growth. Systems of cities of the 'primate' type (one major centre and many lower-order centres) were compared with those of the 'rank-size' type (monotonically declining numbers of cities with increasing order), the result was a lack of any relationship between the type of central place system and the indices of economic development. Therefore the assumption that most of the economic development impulses gravitate to the highest-order centres may be spurious.

The welter of confusion in relation to the actual functioning of central place processes tends to undermine discussions on the relevance of size requirements for growth poles which are expected to coexist within a system of cities. If development stimuli can be motivated up and down the structural hierarchy at will, then the development-pole centre requirement is condensed to a provision of a centre capable of establishing strong linkages, regardless of size or position in the hierarchy. The policy advocated by Hansen (1970) for intermediate centres to act as development poles is not without merit. These centres will continue to stimulate the development of backward linkages at a subregional level, while building up the market thresholds that provoke forward linkages to cluster upon them. Extremes of pole planning at the highest and lowest orders of the urban hierarchy will be avoided, and the resultant centre may be able to stimulate development within an adjacent depressed region (unlike the cross-hierarchy linkage national centre), as well as providing a suitable base for agglomeration economies (unlike subregional centres).

An additional disadvantage of using central place theories for pole formulations is their inability to focus on sociopolitical regulators of growth. Although diffusion of social parameters throughout central place hierarchies is a viable proposition, little attempt has been made to relate this to a general theory of development motivation. The exception to this case is Friedmann's (1966) general theory of polarized development. The principal central place is composed of the 'core' region with all the economic, social, and political attributes of development, whereas the 'peripheral' region suffers from the backwash effects of the core and is generally devoid of modernization tendencies. Core regions exist in a spatial hierarchy in relation to one another and to their peripheral dependencies. Structural transformation of the peripheral regions into new cores occurs as a result of a process of expansion and challenge, on the part of the peripheral decision makers, of the elites within the cores. The outcome is a decentralization of development to the peripheries as a concession to appease the newly arising peripheral power bases. "The actual working out of core-periphery relations, however, is made exceedingly complex by the hierarchical superposition of spatial systems ranging from sub-national province to a (theoretical) world system" (Friedmann, 1972, p. 100).

The complexity so encountered is accentuated in the special case of highly developed nations in which core-periphery dualism of the sociopolitical kind (but not economic disparities) is of minor significance. In such cases the cultural motivating forces behind development and, pari passu, alleviation of regional problems are singularly absent. It would therefore appear that a central place underpinning of pole theory has no fully validated general theory implications. That the ideas of integration have promise is not under dispute, but their present level of advancement leave much to be desired.

Summary: a programme for regional development poles Any attempt to create a comprehensive theoretical base for applying growth poles to regional problem solving would seem to be doomed to failure from the beginning.

304 DTodd

Very little integrated theory can be extracted from the manifold variations of the development pole theme. However, much of the conceptual confusion can be eliminated given recognition of two distinct theories of polarization rather than one. The general theory of polarization is conceived as an aggregate model of differentiation over the entire economy. On the other hand, a partial polarization theory is intended to provide the framework within which specific regional problems can be solved. The latter is therefore an instrument for policy making and as such can only be expected to remain consistent at the unique, disaggregated level.

There are several factors highlighted by various theorists which can be built into a less ambitious polarization programme adjusted for specific regional development. Some of the factors relevant to the formulation of such a programme are suggested below. (1) Appreciation of existing regional structure. This is a prerequisite for acquiring an understanding of patterns of regional specialization, comparative advantage, and marketing. Also it is fundamental for comprehending human resource advantages or deficiencies. (2) Projection of some form of economic growth with a preference for expansion from a base consisting of existing industries which fulfill regional comparative advantage criteria. (3) The choice of additional growth industries should be determined according to their ability to stimulate development both in terms of extra induced economic growth and in terms of human-resources development. (4) The use of spatial linkage analysis as a proxy tool for assessing agglomeration economies pertinent to a potential development pole and which are, indirectly, responsible for stimulating spin-off effects in the local community. A formalized central place framework is not regarded as a prerequisite for generating spin-offs. (5) The provision of dynamic process options. New regional investment will provoke continuing capitalization to provide a process model of self-sustaining development. Although actual predictions may be impossible, some idea of possible development paths can be estimated. A behavioural constraint implied by a series of optional sequential paths may be useful.

Previous works on polarization processes have emphasized a global or general approach to development and in consequence have neglected the partial application of pole theory or the adjustment of pole models to the individual regional problem. General polarization theories labour under the premise of assuming away empirical conditions in hypothesising processes of development 'rounds' in the future. Unfortunately the other great tradition in regional theorizing—evolutionary models-has not filled the gap in conceptual thinking. Thus, tangentially, evolutionary models of the export-base type have attempted to explain development of the existing system without allowing for the prediction of the outcome of the system's continued development. As has been noted, partial pole models can be structured to provide insights into development paths, but they neglect the historical legacy of the subject region. Only by taking into consideration regional evolution can an objective model of future growth potential be reasonably put forward. Such a model would then contain remedies for alleviating inconsistencies in the economy of the region, as well as providing the mechanism for further development using standard polarization concepts.

Acknowledgements. I wish to thank Robert Estall and Nigel Spence of the London School of Economics and Political Science for helpful comments on this paper.

An appraisal of the development pole concept in regional analysis 305

References Alonso, W., 1972, "The question of city size and national policy", in Recent Developments in

Regional Science, Ed. R. Funck (Pion, London), pp.111-118. Berry, B. J. L., 1961, "City size distribution and economic development", Economic Development

and Cultural Change, 9, 573-587. Berry, B. J. L., 1972, "Hierarchical diffusion: the basis of development filtration and spread in a

system of growth centres", in Growth Centres in Regional Economic Development, Ed. N. M. Hansen (Free Press, New York), pp.108-138.

Blaug, M., 1964, "A case of the emperor's clothes: Perroux's theories of economic domination", Kyklos, 17,551-564.

Boudeville, J. R., 1957, "Contribution a l'etude des poles de croissance Bresiliens", Cahiers de Vlnstitut de Science Economique Appliquee, Serie F, no. 10, 1-71.

Boudeville, J. R., 1966, Problems of Regional Economic Planning (Edinburgh University Press, Edinburgh).

Campbell, J., 1972, "Growth pole theory, digraph analysis and inter-industry relationships", Tijdschrift voor Economische en So dale Geographic, March-April, 79-87.

Darwent, D. F., 1969, "Growth poles and growth centres in regional planning—a review", Environment and Planning, 1, 5-32.

Friedmann, J., 1966, Regional Development Policy: A Case Study of Venezuela (MIT Press, Cambridge, Mass.).

Friedmann, J., 1972, " A general theory of polarized development", in Growth Centres in Regional Economic Development, Ed. N. M. Hansen (Free Press, New York), pp.82-107.

Hansen, N. M., 1965, "Unbalanced growth and regional development", Western Economic Journal, 4(1), 3-14.

Hansen, N. M., 1967, "Development pole theory in a regional context", Kyklos, 20, 709-725. Hansen, N. M., 1970, Rural Poverty and the Urban Crisis (Indiana University Press, Indiana). Hansen, N. M. (Ed.), 1972, "Growth centre policy in the United States", in Growth Centres in

Regional Economic Development (Free Press, New York), pp.266-281. Hermansen, T., 1972, "Development poles and development centres in national and regional

development", in Growth Poles and Growth Centres in Regional Planning, Ed. A Kuklinski (United Nations Research Institute for Social Development, Geneva), pp.1-68.

Hirschman, A. O., 1958, The Strategy of Economic Development (Yale University Press, New Haven, Conn.).

Hoover, E. M., 1948, The Location of Economic Activity (McGraw-Hill, New York). Isard, W., Schooler, E. W., Vietorisz, T., 1959, Industrial Complex Analysis and Regional

Development (MIT Press, Cambridge, Mass.). Lasuen, J. R., 1969, "On growth poles", Urban Studies, 6 (2), 137-161. Moore, C. W., 1972, "Industrial linkage development paths in growth poles: a research

methodology", Environment and Planning, 4, 253-271. Moseley, M. J., 1973a, "Growth centres—a shibboleth?", Area, 5, 143-150. Moseley, M. J., 1973b, "The impact of growth centres in rural regions", Regional Studies, 7,

57-94. Myrdal, G., 1957, Economic Theory and Underdeveloped Regions (Methuen, London). New Brunswick Multiplex Corporation, undated, A New Centre for Metal Working Industries

(New Brunswick Multiplex Corporation, New Brunswick). Newcombe, V. Z., 1969, "Creating an industrial development pole in southern Italy", Journal of

the Town Planning Institute, 55 (4), 157-161. Nichols, V., 1969a, Growth poles: an Investigation of their Potential as a Tool for Regional

Economic Development, DP.30, Regional Science Research Institute, Philadelphia. Nichols, V., 1969b, "Growth poles: an evaluation of their propulsive effect", Environment and

Planning, 1, 193-208. Odland, J., Casetti, G., King, L. J., 1973, "Testing hypotheses of polarized growth within a central

place hierarchy", Economic Geography, January, 74-79. Paelinck, J., 1972, "Programming a viable minimal investment industrial complex for a growth

centre", in Growth Centres in Regional Economic Development, Ed. N. M. Hansen (Free Press, New York), pp.139-159.

Perroux, F., 1955, "Note sur la notion de 'pole de croissance'", Economique Appliquee, 7, 307-320. Translation in Regional Economics, 1970, Eds. D. L. McKee, R. D. Dean, W. H. Leahy (Free Press, New York), pp.93-103.

Pottier, P., 1963, "Axes de communication et developpement economique", Revue Economique, 1, 58-132.

306 DTodd

Pred, A. R., 1966, The Spatial Dynamics of US Urban-Industrial Growth, 1800-1914 (MIT Press, Cambridge, Mass.).

Robinson, G. S., Salih, K. B., 1971, "The spread of development around Kuala Lumpur", Regional Studies, 5,303-314.

Schumpeter, J. A., 1949, The Theory of Economic Development (Harvard University Press, Cambridge, Mass.).

Scitovsky, T., 1954, "Two concepts of external economies", Journal of Political Economics, 62, 143-151.

Thomas, M. D., 1972a, "The regional problem, structural change and growth pole theory", in Growth Poles and Growth Centres in Regional Planning, Ed. A. Kuklinski (United Nations Research Institute for Social Development, Geneva), pp.69-102.

Thomas, M. D., 1972b, "Growth pole theory: an examination of some of its basic concepts", in Growth Centres in Regional Economic Development, Ed. N. M. Hansen (Free Press, New York), pp.50-81.

Thompson, W., 1965, A Preface to Urban Economics (The Johns Hopkins Press, Baltimore). Tosco, E. S., 1971, "Relevance of intermediate industries and industrial services for correct

location planning", in Industrial Location and Regional Development (United Nations Industrial Development Organisation, New York), pp. 160-202.

UNIDO, 1969, Report of the Interregional Seminar on Industrial Location and Regional Development, Minsk 1968 (United Nations Industrial Development Organisation, New York).

Wood, P. A., 1969, "Industrial location and linkage", Area, 1, 32-39.

p © 1974 a Pion publication printed in Great Britain