The last years have seen a strong increase in the study of industrial location and economic geography. The deepening economic integration within so called tax and customs unions such as the European Union has fuelled new interest in the field of industrial location. The fact that industrial production has a tendency to agglomerate in central regions has been a major concern to policy makers not just within the EU.
Recent findings by researchers such as Fujita, Krugman and Venables (1999) identifying new modelling tricks to compare industrial organizations, international trade and economic growth more accurately has led to an increasing number of studies being conducted. The Three Main Schools of Industrial location Developments of the theory of industrial location can be tracked down by following the evolution of the three main schools of thought as outlined below.
For a tabular illustration, please refer to the Appendix. One of the first studies of trade and location is Ricardo’s classical theory of comparative advantage (1817), which was later on linked with the Heckscher-Ohlin (H-O) model. The H-O model predicts that the patterns of trade between countries are based on the characteristics of the countries. The H-O model says that a capital-abundant country will export the capital-intensive good while the labour-abundant country will export the labour-intensive good.
Those studies contributed to the establishment of the Neo-classical theory (NCT), which is characterised by perfect competition, homogeneous products and falling returns to scale. Krugman (1993) named the factors determining location “first nature”. He explains that economic activity is spread or concentrated over space according to the spread or concentration of the following features: The spatial distributions of natural endowments, technologies as well as factor endowments and factor intensities.
However, he argues that the dominating location pattern is inter-industry specialisation: sectors of industry tend to settle in locations with a matching comparative advantage. Assuming zero trade costs, the spatial distribution of demand only affects the patterns of trade, but not the location of production. Assuming trade costs and evenly distributed demand over space, then the locational dispersion of activity will correlate positively with the level of trade costs.
Following the geographical distribution of demand, high levels of trade costs therefore induce perfect dispersion of industries. New trade theories (NTT) were developed to explain high levels of intra-industry trade and the large proportion of world trade that takes place between similar countries. NTT excludes the exogenous, ‘first nature’ elements, but places a major emphasis on market size. Theoretical modelling has shown that the interaction of scale economies and trade costs encourages the concentration of manufacturing production in countries that have good access to large markets.
The presence of scale economies encourages firms to choose only one location, and the presence of trade costs encourages them to locate in the country that has the larger market for their goods. According to Krugman it is determined primarily by the size of the labour force in a particular country, while assuming that labour force is immobile across countries. These models introduce activity-specific features, which Krugman (1993) labelled “second nature”, such as imperfect competition, differentiated products and increasing returns. The typical outcome has two layers.
First, there is inter-industry specialisation, with sectors clustering in locations which offer best access to product markets. Second, there is intra-industry specialisation across firms, each of which produces a unique, horizontally differentiated variety of the industry’s product. Intra-industry trade will take place as long as some firms are left in the smaller market. As trade costs fall towards zero, however, all increasing-returns activity will tend to concentrate near the core market, and intra-industry trade between the core and the periphery vanishes.
In the New Economic Geography (NEG) models, location becomes entirely endogenous: ‘second nature’ is the main determinant of location. Because production factors are assumed mobile, even market size is explained within the model. The analytical starting point is normally a two- or three dimensional space with uniformly distributed labour and output of a single industry. This distribution tends to be unstable, due to the assumed ‘second nature’ characteristics of the economy, such as market-size externalities and input-output linkages.
These characteristics generally produce self-reinforcing agglomeration processes. Therefore, any shocks to the initial distribution will cause the economy to reach a new equilibrium. There are many possible and locally stable equilibria. Which one is attained depends on the starting distribution, on the nature of the shocks and on the industry characteristics. Illustrated below are the multiple equilibria which can arise in a so called Bifurcation model: A non-linear relationship between the share of manufacturing labour between two regions and trade cost.
The starting point, which is marked by two different levels of trade cost (T) in turn determines profitability of firms. Depending on the starting point, they either predict convergence or divergence. At high values of T it is most profitable to produce in both locations, leading to divergence. At low levels of T, production will tend to locate in a single location (the core). Bifurcation Core Periphery Model In view of the multiple outcomes which can be accommodated in this theoretical framework, empirical evidence is limited.
In these models, agglomeration mechanisms are confronted with increases in the prices of the immobile factors. The outcome may be high agglomeration levels resulting in the centre-periphery structure, which in turn lead to higher economic integration. The new economic geography literature extends this line of research by showing that international or inter-regional demand differences are themselves likely to be endogenous – either because of mobility of workers (Krugman, 1991) or because of mobility of firms which demand intermediate goods (Venables, 1996).
In industries, which are linked through an input-output structure, the downstream industry will form the market for upstream firms. The home-market effect1 then means that upstream firms are drawn to locations where there are relatively many. In addition to this demand (or backward) linkage there may also be a cost (or forward) linkage. Having a larger number of up-stream firms in a location benefits downstream firms who obtain their intermediate goods more cheaply – by saving transport costs, and perhaps also benefiting from a larger variety of differentiated inputs or more intense competition in the upstream industry.
The combination of these backward and forward linkages creates the possibility of a clustering of vertically related industries. Since these demand and cost linkages are stronger when the proportion of intermediate goods in production of final goods is higher, we should expect the level of geographical concentration to be higher in industries that are more intensive users of intermediate inputs in final production.