![]()
N. A. Phelps
[Paper first received, February 2003; in final form, October 2003]
Introduction
The recent interest in clusters or agglomerations of economic activity can draw on a long analytical tradition in economics and geography.1 However, although such clusters or agglomerations are enduring features of the industrial landscape and a perennial source of theoretical and empirical interest, there have been few explicit theoretical treatments of the changing geographical scale at which agglomeration is apparent.2 The industrial districts of the late 19th centuryâthe inspiration for Alfred Marshall's theory of agglomerationâwere, in the main, self-contained singular industrial towns or else quarters of large cities (Hall, 1962; Wise, 1949). Today, however, the major examples of industrial agglomeration are of an entirely different geographical scale. Take, for example, the likes of âSilicon Valleyâ near San Francisco (Saxenien, 1983), Britain's âMotor Sports Valleyâ (Henry and Pinch, 1999), the neo-Marshallian nodes represented by global cities (Amin and Thrift, 1992) that draw together suburbs and satellite towns and cities into polycentric agglomerations (Hall, 1997) or âscatteratedâ urban forms exemplified by Los Angeles (Gordon and Richardson, 1996, cited in Coffee and Shearmur, 2002). Such scatterated urban forms have European parallels in the dense networks of towns and cities of the Randstad (Batten, 1995) or the regionalised economy of the South East of England (Allen, 1992; Coe and Townsend, 1998) and have been likened to the extended metropolitan forms found in east Asia (Dick and Rimmer, 1998).
The emergence of these increasingly spatially diffuse forms of agglomeration raises questions for economic geography. First, they direct attention explicitly to questions of the spatial scale at which external economies and agglomeration processes are present. From a geographical point of view, then, explanation of the variable spatial scale of agglomeration remains a key question that, nevertheless, has rarely been addressed directly. As Gottdiener has suggested, there is a need to understand
the processes that have worked and reworked settlement space by increasing its scale, fragmenting its communities and nodal points of interaction; by decentralizing its businesses and recentralizing them in new, functionally specialized nodes (Gottdiener, 2002, p. 178).
A few recent exceptions aside, geographers still have a tendency to interpret today's spatially expansive urban agglomerative forms in terms carried over from Alfred Marshall. For economists, cities, including the sorts of sprawling urban formations noted above, have been and continue to be taken as evidence of the importance of increasing returns to scale, the presence of âmarket-size effectsâ, âurbanisation economiesâ or âpecuniaryâ external economies. Hence, it is tempting to perceive invariant principles to the clustering or agglomeration of economic activity. However, is geographers' and economists' reliance on these invariant themesâof Marshallian externalities and of âsizeââtoo restrictive to capture important changes in economies over time? Can variations on these principles shed light upon the changing form and causes of agglomeration?
Secondly, these diffuse forms of agglomeration are notable for throwing up rather anonymous âintermediateâ locations or places that are nevertheless important in economic terms. This begs the question of whether, in considering contemporary manifestations of external economies, the clustering of economic activity and the spatial scale at which these occur, more analytical effort ought not to be directed at uncovering the economic dynamics of intermediate locations. Clearly, what count as intermediate locations have varied geographically and historically to include suburbs, edge cities or edgeless cities, but they might be defined as the relatively new locations that exist between established urban centres. In comparison with the enormous weight of theoretical and empirical interest in industry clusters, we know very little about the economic basis of these seemingly banal locations.
In this exploratory paper, I address these two questions. This in turn rests on a need for sensitivity to the type of externalities, the spatial scale at which they are present and their meshing together. Some types of externalities are more mobile than others. Moreover, the salience of different externalities to particular sectors of the economy has changed over time. It should be no surprise therefore that any renewal of the theory of agglomeration ought to move beyond general modelling of invariant principles of agglomeration (Krugman, 1996) to more geographically and historically context-sensitive explanations. Some of these historical contrasts in agglomeration have been drawn in very broad schematic terms in a previous paper (Phelps and Ozawa, 2003). This paper makes a further contribution to renewing the theory of agglomeration. It first considers the virtues of variations on the classical themes of agglomeration theory in explaining today's sprawling urban forms. In particular, the paper considers whether neo-Marshallian analyses and the idea of âborrowed sizeâ are able to capture adequately the tension between the mobility of pecuniary external economies and the relative fixity of technological externalities. It then points to the economic significance of intermediate locations and explores their economic basis in terms of the meshing of different external economies. In conclusion, it is suggested that analytical enquiry might more usefully focus on the meshing of different types of internal and external economies as these produce a âbanalâ economic geography of intermediate locations.
External Economies, Agglomeration and Geographical Scale
The analysis of industrial agglomeration has proceeded from Alfred Marshall who deployed the concept of external economies.3 The existence of external economiesâthe advantages that are open to and shared among a collectivity of businessesâpromotes the geographical agglomeration of economic activity. These advantages are chiefly, especially in the geographical literature, discussed in relation to a given singular town, city or territorial scale. Specifically, Marshall identified a trinity of external economies: those connected with inputâoutput transactions, those of labour market pooling and technological externalities.
Whilst drawing on the work of Alfred Marshall, from the work of Hoover in the 1930s until comparatively recently, one additional distinction which was made in geographical literature on agglomeration was that between âurbanisationâ and âlocalisationâ economies (see, for example, Karaska, 1969). In part, this distinction could be read as an early example of geographers' sensitivities to the expanding geographical scale over which external economies operate. However, since the instances of localised agglomeration being studied coincided with or were contained within single urban areas, it is less the geography of externalities than their type that geographers were drawing attention to. Here, urbanisation economies appear to equate to what economists refer to as pecuniary external economies (although these are not confined to urban areas) and localisation economies to technological externalities. Moreover, urbanisation economies appear to combine aspects of pure external economies of scale effects and the diversity of external economies of scope effects (Parr, 2002a, 2002b).
As the discipline of economics has evolved, it is the distinction between pecuniary externalities (market size or true external economies of scale effects) and technological externalities (Brown and Jackson, 1985; Harrison, 1992; Martin and Sunley, 1996; Skitovsky, 1954) that has emerged from Marshall's original concept. It is this distinction that we wish to work with in the remainder of the paper when considering the geographical scale at which forces of agglomeration have operated. This binary distinction cuts across the trinity of Marshallian externalities. The specialised inputâoutput transactions fall fairly squarely into the category of pecuniary externalities (Phelps, 1992). The economies of labour market pooling, however, embody both pecuniary and technological effects. Although acknowledging the existence of very localised forms of agglomeration, the preference of economists has been to concentrate on pecuniary externalities not least because of the difficulties of incorporating the latter into formal equilibrium models. Finally, the technological externalitiesâthe industrial atmosphere, as it is sometimes referred toâare those that have been and continue to be localised.
Working from their different perspectives and from different sides of this pecuniaryâtechnological divide, economists and geographers have paid comparatively little attention in theoretical terms to the changing extent of industrial agglomeration apparent over time. Yet one inescapable essence apparent to those concerned with agglomerative processes is that, to quote a geographer, âagglomeration effects in large industrial cities can not be assumed to be geographically fixedâ (Scott, 1982, p. 118; original emphasis). The same observation was made somewhat earlier by the economist E. A. G. Robinson. Commenting on the dissolution of Britain's industrial districts prior to the Second World War, Robinson drew a distinction between mobile and immobile external economies. It is this logic which led the economic geographer Richard Walker to suggest that agglomeration economies are an âhistorically contingent feature whose force has gradually diminished to be replaced by the economies of internal organisational scale open to large companiesâ (Walker, 1981, p. 385), although he has in subsequent work re-emphasised the persistence of localised agglomeration as part-and-parcel of urban and regional specialisation and differentiation (Storper and Walker, 1989).
In the remainder of this section, I want to consider the question of the geographical availability of external economies. Given the likes of major historical trends (such as in the technological and sectoral basis of economies), industry specifics, national institutional specifics and even ambiguities in the concept itself, this question is not readily reduced to associating particular types of external economy with fixed geographical scales. Instead, I approach the question indirectly, drawing attention to the relative mobility of different types of external economies.
The Immobility of Localised or Technological Externalities
The distinction between pecuniary and technological external economies has rarely been apparent within geographical work on agglomeration; however, until very recently, in the guise of âlinkage studiesâ, much of the geographical research effort focused on urbanisation economies or pecuniary effects, rather than technological externalities per se. Most notably, a large body of linkage studies in the late 1960s and 1970s examined these pecuniary effects in promoting agglomeration (Hoare, 1985). These studies aimed to trace the significance of local linkages to the agglomeration of industry but instead appeared to indicate that agglomeration often existed without strong localised linkages (for example, Lever, 1972). Whilst linkages may have played an important part in Alfred Marshall's late 19th-century industrial districts, their role in the sometimes ailing agglomerations of late-Fordism was altogether more questionable (Phelps, 1992).
Somewhat curiously, then, interest in the connections between localised linkages, external economies and localised agglomeration was revived with a string of industry studies in the Los Angeles areaâan area frequently referred to as the exemplar of urban sprawl and the dissolution of the classical monocentric city spatial structure. This work represented a deepening of our understanding of the role of inputâoutput transactional external economies in promoting...
