False Paradoxes and ‘Circular Logics’

The research problem attached to the agglomeration phenomenon may seem to be trivial. We have a phenomenon that can be observed in reality. Firms in the same or related industries often – although far from always – tend to co locate in space, and this should reasonably indicate that there should be some advantages connected to such a location pattern: agglomerations do exist and this may legitimately make us assume that such a spatial structure is in some sense efficient or rational. Still, the agglomeration phenomenon is often misconceived and the concept of agglomeration economies remains elusive.

A common misconception of the problem has to do with the relation between globalization and the spatial location of economic activity. All too often in contemporary economic geography, reference is being made to an alleged paradox: in an increasingly globalized world economy, localization seems to matter more, not less. The key to this alleged paradox can be found already in Adam Smith's famous dictum in his Inquiry into the Wealth of Nations in 1776, however, stating that the division of labor is determined by the extent of the market. The extension of the market explicitly refers to the spatial extension of the producers' market area. Smith's statement has a strong bearing on recent work on the logic of industrial agglomeration. If globalization means the emergence of truly integrated global markets, it follows that it is increasingly possible to access these global markets from, in principle, any point of location. It should be easy to see the increased global economic integration opens up for more local/regional/national specialization. Thus, the structural effect of such globalization is that one or a few leading global production centers should in principle be able to supply the entire world market in each industry, where previously there were many of regional, national, or local production centers. In other words, increasing agglomeration of similar and related activity should be seen as a structural effect of globalization rather than some strange paradox.

The circular logic referred to in the quotation by Fujita et al. incurs more serious problems, however. In the terminology of economists, the agglomeration phenomenon is intimately related to increasing returns. (Economic) geographers often prefer the concept of external economies of scale. Either way, we have a situation where initial events trigger subsequent action in a process of cumulative causation – or a circular logic, as Fujita et al. phrased it in the quote above. The existence of one or several firms in a certain place will lead to the attraction of more firms of the same kind, and the increased volume of activity will in turn exert an even greater attraction on further firms. Such processes tend to be self reinforcing, at least until some countering diseconomies kick in. There are several potential problems with this line of reasoning. One is that, unless it can be specified in some detail what the mechanisms of agglomeration economies are, there is the risk of ending up in circular reasoning. Paul Krugman cites the remark of a sarcastic physicist who had just heard a presentation on increasing returns: ''So you're telling us that agglomerations form because of agglomeration economies.'' Models of agglomeration have always run such a risk: the existence of agglomerations is explained by agglomeration economies. These agglomeration economies cannot really be specified or observed but their existence is proved by the existence of agglomerations! A related problem is that a proper explanation of the agglomeration phenomenon should also have to specify the limits of agglomeration economies. If we specify a number of reasons why it is advantageous for firms to locate close to other similar or related firms or in big cities – which indeed many of them do – we also have to be able to say something about why not all firms locate in such settings, or why in some case dense agglomerations seem to petrify and lose their dynamism.

The next section summarizes the factors held to make up agglomeration (or localization) economies. The final section then discusses in schematic terms how we can understand the development of agglomerations over time, by adopting an evolutionary approach.