Economic Geography[223]

The sectoral specialisation of countries, and of regions and cities within coun­tries, is the result of firms' location decisions. A firm's location decision gives rise to an economic problem when two things are true.

However, while fundamental, these two ingredients are incompatible with the perfectly competitive paradigm that still dominated much of mainstream economics (and TTT) in the late 1970s and early 1980s. Starrett (1978) high­lighted this theoretical impasse in his “spatial impossibility theorem”: if space is homogenous, there does not exist any competitive equilibrium with ship­ments between distant locations. A policy-relevant implication is that any analysis trying to explain how economic interactions per se shape the economic landscape has to abandon the assumption of perfect markets and the associated efficiency property of market equilibrium.

There are many ways out of this impasse: while there is only one way to be perfect, there are many ways to be imperfect. Most obviously, a first solution is to acknowledge that space is not homogenous. Places differ in terms of their relative abundance of natural resources, proximity to natural means of com­munications and climatic conditions. This is the way out investigated by TTT. However, it looks like an inadequate explanation of the dramatic differ­ences in economic development that one observes even between areas that are not very different in terms of those exogenous properties.

Localised pecuniary externalities are at the core of NEG. Eventually, their comparative advantage lies in the possibility of relating their emergence to a set of well-defined microeconomic parameters. This has proven to be quite difficult in models based on the concept of technological externalities as these still remain mostly “black boxes” (see Duranton and Puga 2004), although some progress has been made by, among others, Storper and Venables (2004). Differently, building on NTT, starting in the late 1980s NEG managed to show how pecuniary externalities arise in sectors characterised by relevant trade costs (due to transportation as well as to administrative and cultural bar­riers), increasing returns to scale, and monopolistic competition. In those sec­tors, when a new firm begins producing in a certain location, it increases local demand for upstream activities (“market expansion effect”) and local supply for downstream ones (“market crowding effect”). It generates a pecuniary externality insofar as the firm bases its entry decision on its own profit and this, due to imperfect competition, does not perfectly reflect all the changes in the payoffs of upstream and downstream activities.

Agglomeration takes place when the final impact of the market expansion effect dominates the impact of the market crowding effect (see Krugman 1991; Krugman and Venables 1995; Venables 1996). Consider, for instance, the situation depicted by Venables (ibid.), which predates Krugman and Venables (1995) in terms of working paper versions. There are three vertically linked activities: intermediate production, final production and consump­tion. For simplicity, assume that final production uses only intermediate inputs, intermediate production employs only labour and workers are the only source of final demand. If, for any reason, a new firm starts producing intermediates, it will increase labour demand and intermediate supply. Due to excess demand and supply, respectively, wages will go up while intermediate prices will fall. This is bad news for the other intermediate producers (market crowding effect). However, it is good news for final suppliers, who experience falling production costs and higher demand by richer workers. As new final producers are induced to enter the market, the expansion of final production will feed back into stronger intermediate demand so that intermediate suppli­ers will also benefit (market expansion effect). When the latter effect domi­nates the former, both final and intermediate firms will end up agglomerating in the same place.

This mechanism was not new when Tony constructed his model. For exam­ple, it had been described by both Marshall (1890) and Ohlin (1933). The crucial contribution of Tony's model and thus of NEG was that the mecha­nism in question was translated into a general equilibrium model with solid microeconomic foundations. As a result, the evolution of the spatial land­scape was related to microeconomic parameters: agglomeration is more likely to take place in sectors where increasing returns are intense, market power is strong, customers and suppliers are easily mobile and trade costs are low. The reason is that more intense returns to scale and stronger market power weaken the market crowding effect, while more mobile customers and suppliers amplify the market expansion effect.

The impact of trade liberalisation is arguably the central insight of NEG models. At first sight, it did not sound that new as it was reminiscent of Kaldor (1970), who predicted the loss of the industrial base of a less devel­oped region facing trade liberalisation with respect to a more developed one. However, NEG models provided a more detailed understanding of how the economic landscape evolves as trade impediments are gradually eliminated. NEG models showed that lower trade impediments affect the balance between market expansion and market crowding effects in a very nonlinear way. In particular, they may cause self-reinforcing uneven development between growth centres and stagnant peripheries as economic activities agglomerate in the former and shun the latter. Only the congestion of immobile resources and non-tradable inputs can stop the emergence of self-sustaining “core­periphery” patterns. Moreover, processes of circular causation like the one described by Venables (1996) can sustain multiple spatial equilibria and thus make the spatial economy settle in a suboptimal equilibrium due to path dependency, self-fulfilling expectations and lock-in effects, factors which remind one of Venables (1984).

These insights were “iconoclastic” when they made it into the public policy discourse, especially in Europe. The late 1980s and early 1990s were crucial years for the project of European integration of “One Market, One Money” (Commission of the European Communities 1990). The project was based on a promise of inclusive development that would benefit all regions of the Old Continent, in particular the least prosperous ones. The promise rested on the expectation that lower barriers to the international movement of goods, labour and capital in an integrating Europe would eventually lead to regional convergence in prices, factor returns and thus living standards between and within countries.

This expectation mainly derived from neoclassical growth arguments founded on the assumptions of constant returns to scale and perfectly com­petitive markets. In some of their more sophisticated versions, these argu­ments acknowledged the long-term horizon in which convergence would take place but considered, if not convergence, at least regionally balanced growth as the inevitable final outcome of economic integration. This can be seen as the economic essence of the “European dream” (see Ottaviano 2019b). On the one hand, as economic growth is typically driven by what happens in a limited number of dynamic geographical areas that lead the way, any leap forward along the development path requires the geographical concentration of economic activities and thus an ineluctable division of areas in more and less developed countries. The emergence of an economic “core” and an eco­nomic “periphery” are two sides of the same coin. On the other hand, the implied polarisation of the economic landscape is not an issue as long as the success of more dynamic areas automatically ends up also benefiting less dynamic ones. If the core grows, the periphery will eventually grow too. As time goes by, the wealth gap between the two remains constant and may even shrink as long as it is easier to follow than to lead the way. In this respect, European integration promotes inclusive development. It not only allows for the concentration of economic activities in the most dynamic regions that are the engines of growth, but also facilitates the diffusion of such growth to the least dynamic areas thanks to their belonging to a common economic space.

NEG forcefully made the point that the foregoing did not need to be the only ineluctable outcome (see Baldwin and Venables 1995; Krugman and

Venables 1996). A quarter of a century later, the promise of inclusive develop­ment does not seem to have been kept, and the strengthening of regional imbalances is increasingly becoming a threat not only to social cohesion but also to the political stability of the European Union, as possibly heralded by Brexit. Almost 30 years since the introduction of the Single Market, the icon­oclastic insights of NEG can still help understand the causes of this lack of geographical convergence. Tapping the coherent body of results systematised by Fujita et al. (1999), Fujita and Thisse (2002) and Baldwin et al. (2003) would be an obvious starting point.

4