VoxEU Column Development Productivity and Innovation

Is the new economic geography passé?

Paul Krugman suggests that his Nobel-prize-winning “core-periphery” model was perhaps more relevant a century ago than today. This appears to be true in terms of overall manufacturing concentrations in Europe and North America, which are unravelling. Large-scale agglomeration forces, however, are alive and well in the developing world, as are localised sectoral clustering phenomena in industrialised countries.

In his Nobel Prize lecture on December 8, Paul Krugman argued that his core-periphery model of economic geography is in some sense becoming obsolete. It is not true that “the latest wrinkle in theory must be the latest wrinkle in the way the world works”, he pointed out, and he went on to speculate that “the world is becoming less new geography and more classical”. In the accompanying Nobel interview, he characterised his geography models as “retrofitting”.1 Has the Nobel Prize therefore been awarded for an elegant model of a vanishing phenomenon? Are “second nature” agglomeration economies – forces that make firms want to locate close to each other for different reasons than “first nature” features of geography like good weather or access to natural resources – no longer important?

In his Stockholm lecture, Krugman used the example of the US car industry, which for close to a century had been tightly concentrated in the country’s northeastern manufacturing belt but has been showing clear signs of spreading southwards since the 1990s. He also pointed to Kim’s (1998) study of long-run regional specialisation trends in the US, which reveals that the pinnacle of manufacturing concentration was reached in the first half of the 20th century, with a strong dispersion tendency in evidence since around 1950. While no similarly long-run continent-wide data series exist for Europe, the available evidence tells a similar story: since the mid-1970s, the spatial concentration of manufacturing employment across EU regions has fallen continuously and significantly (Brülhart and Traeger, 2005). Moreover, French long-run regional data show the same pattern as US statistics – manufacturing concentration increased markedly between 1860 and 1930 only to decrease again between 1930 and 2000 to below its 1860 level (Combes et al., 2008 and on Vox).

Long-established regional manufacturing concentrations in Europe and North America provided the original empirical inspiration behind the new economic geography. These concentrations are evidently unravelling. And yet we contend that spatial agglomeration processes such as those modelled by Krugman and others remain eminently relevant in our age.

Agglomeration spurs growth in developing countries

Agglomeration forces loom large for the 85% of the world’s population who do not live in high-income economies. As documented in the latest World Development Report (World Bank, 2008), spatial inequalities and urbanisation are rising rapidly across the developing world.

These increasingly uneven economic landscapes, while commonly associated with social upheaval, urban slums, and rural decline, may well be a blessing in disguise. Dynamic variants of new economic geography models imply that spatially concentrated economies will grow faster (Baldwin and Martin, 2004). Recent empirical research suggests that developing economies may indeed stand to gain considerably from further spatial concentration. Brülhart and Sbergami (2008) find that, across a range of statistical specifications, poor economies grow faster if their internal economic geography is initially more concentrated  or becomes more concentrated over time. This is illustrated in the graph below. It plots the estimated growth effects of urbanisation in a large cross section of countries against those countries’ (log of) per-capita income in 1960. It appears that urbanisation provides an impetus to economic growth for the majority of countries. Taking the plotted estimates at face value, urbanisation is good for growth in countries with a GDP per capita up to some $10,000 (in PPP terms), which corresponds roughly to the current development level of Brazil or Bulgaria. In the richest economies, however, agglomeration (measured in a range of ways) appears if anything to inhibit growth, although those effects are not statistically significant.

The growth effect of urbanisation as a function of GDP per capita

Source: Brülhart and Sbergami (2008), Table 1

Like all cross-country regression estimates, the findings of Brülhart and Sbergami (2008) need to be interpreted with a certain dose of caution. Methodological obstacles to measurement and econometric identification are daunting. That said, the estimates do suggest with rather striking regularity that urbanisation promotes growth in developing countries. Importantly, policies inhibiting spatial economic concentration are most damaging, in terms of foregone growth, in the poorest countriest.

Clusters are in

While broad manufacturing agglomerations in developed countries seem to be coming apart, and spatially concentrated rich economies seem to grow no faster than more dispersed economies, agglomeration forces continue to shape rich-world economies “below the radar” of aggregate data. Duranton and Overman (2005) zoom in to the level of four-digit manufacturing sectors (of which there are some 230) and of British postcodes (which identify the location of individual buildings). They find that more than half of these industries are significantly more spatially concentrated than what a random location process would imply – clustering is pervasive. Most of this clustering occurs at a small spatial scale, below 50 kilometres. And the most strongly clustered sectors are localised for no apparent “first nature” reason. The list of most concentrated sectors is topped by publishing of records and books, various textile sectors, and the manufacture of cutlery. Clearly, “second nature” agglomeration forces are at play here. Moreover, these clusters do not seem to be unravelling. Using similarly detailed data, Simpson (2007) reports that the degree of clustering of UK industries in fact increased slightly between 1997 and 2005.

Small-scale sectoral clusters are not only a pervasive and increasingly well-documented economic phenomenon, they have also become something of an obsession of business consultants and policy advisers in the last two decades – sometimes with vague reference to new economic geography models. In this respect, rigorous theoretical and empirical work sends a cautionary message. As shown forcefully by Duranton (2008), theory is ambiguous on whether the market produces insufficient clustering, cluster policy is particularly prone to capture by special interests, and empirical work suggests that any productivity gains from cluster promotion are very small at best. On the last issue, recent research on French micro data by Martin et al. (2008), summarised on Vox, confirms that there are large productivity gains to be reaped from clustering, but that these gains are largely internalised by firms already, such that there is little to be gained by prodding them into clustering further.

While the new economic geography and related theories of “second nature” agglomeration processes may no longer capture the essential forces that shape the geography of broad manufacturing categories across industrialised world regions, they remain of acute relevance to the spatial transformation processes under way in much of the developing world, as well as to high-resolution clustering phenomena in rich economies. Thanks to the wave of research sparked by Paul Krugman in the 1990s, economic geography has become hip again among economists, after a long period of relative neglect.2 The evidence of ongoing agglomeration processes worldwide should suffice to stop this from being a passing fad.


Baldwin, Richard E. and Philippe Martin (2004) Agglomeration and Regional Growth. In: V.J. Henderson and J.F. Thisse (eds), Handbook of Regional and Urban Economics, vol. 4, Elsevier North-Holland.

Brülhart, Marius and Rolf Traeger (2005) An Account of Regional Concentration Patterns in Europe. Regional Science and Urban Economics, 35(6): 597-624.

Brülhart, Marius and Federica Sbergami (2008) Agglomeration and Growth: Cross-Country Evidence. CEPR Discussion Paper #6941 (forthcoming in the Journal of Urban Economics, 2009).

Combes, Pierre-Philippe, Miren Lafourcade, Jacques-François Thisse and Jean-Claude Toutain (2008) The Rise and Fall of Spatial Inequalities in France: A Long-Run Perspective. CEPR Discussion Paper #7017.

Duranton, Gilles (2008) California Dreamin’: The Feeble Case for Cluster Policies. Working Paper, University of Toronto.

Duranton, Gilles and Henry Overman (2005) Testing for Localisation Using Micro-Geographic Data. Review of Economic Studies, 72(4): 1077-1106.

Fujita, Masahisa and Jacques-François Thisse (2008) New Economic Geography: An Appraisal on the Occasion of Paul Krugman’s 2008 Nobel Prize in Economics. CEPR Discussion Paper #7063.

Kim, Sukkoo (1998) Economic Integration and Convergence: US Regions, 1840-1987. Journal of Economic History, 58(3): 659-683.

Krugman, Paul (1991) Increasing Returns and Economic Geography. Journal of Political Economy, 99(3): 483-499.

Simpson, Helen (2007) An Analysis of Industrial Clustering in Great Britain. Report for the Department of Trade and Industry, London.

World Bank (2008) World Development Report 2009: Reshaping Economic Geography, Washington DC.


1 For Krugman’s Nobel lecture slides as well as videos of both the lecture and the interview, see http://nobelprize.org/nobel_prizes/economics/laureates/2008/. The seminal new economic geography paper is Krugman (1991).

2 See Fujita and Thisse (2008) for a concise summary of the intellectual antecedents to and subsequent developments of the new economic geography.

3,045 Reads