The Empirics of New Economic Geography∗Stephen J. ReddingLSE, Yale School of Management and CEPR†May 12, 2009AbstractAlthough a rich and extensive body of theoretical research on new economic geography has emerged,empirical research remains comparatively less well developed. This paper reviews the existing empiricalliterature on the predictions of new economic geography models for the distribution of income andproduction across space. The discussion highlights connections with other research in regional andurban economics, identification issues, potential alternative explanations and possible areas for furtherresearch.Keywords: New economic geography, market access, industrial location, multiple equilibriaJEL: F12, F14, O10∗This paper was produced as part of the Globalization Programme of the ESRC-funded Centre for Economic Performance atthe London School of Economics. Financial support under the European Union Research Training grant MRTN-CT-2006-035873is gratefully acknowledged. I am grateful to the editor (Gilles Duranton), my discussant (Kristian Behrens) and participants atthe Journal of Regional Science 50th Anniversary Symposium at the Federal Reserve Bank of New York for helpful commentsand suggestions. I am also grateful to a number of co-authors and colleagues for insight, discussion and comments, includingin particular Tony Venables, and also Guy Michaels, Henry Overman, Esteban Rossi-Hansberg, Peter Schott, Daniel Sturm andNikolaus Wolf. I bear sole responsibility for the opinions expressed and any errors.†Department of Economics, London School of Economics, Houghton Street, London, WC2A 2AE, United Kingdom. Tel: +44 20 7955 7483, Fax: + 44 20 7955 7595, Email: [email protected]. Web: http : //econ.lse.ac.uk/staff/sredding/.11 IntroductionOver the last two decades, the uneven distribution of economic activity across space has received re-newed attention with the emergence of the “new economic geography” literature following Krugman(1991a), which was a key part of the citation for Paul Krugman’s 2008 Nobel Prize. Whereas traditionalneoclassical explanations for the distribution of economic activity across space emphasize “first-naturegeography” (the physical geography of climate, topology and resource endowments), this new body ofresearch stresses instead the role of “second nature geography” (the location of economic agents relativeto one another in space).The core building blocks of new economic geography models are product differentiation modeledthrough a love of variety assumption, increasing returns to scale and transport costs, which together cre-ate pecuniary externalities in agents’ location choices. When combined with either factor mobility orintermediate inputs, these three building blocks give rise to forces of cumulative causation and agglom-eration.1As workers coalesce in a location, the resulting shift in expenditure increases the incentive forfirms to concentrate production in that location (the “home market effect”). Similarly, as firms concen-trate production in a location, the resulting reduction in the consumer price index increases the incentivefor workers to coalesce in that location (the “price index effect”).In new economic geography models, the tension between these agglomeration forces and dispersionforces in the form of immobile factors of production or non-traded amenities in inelastic supply deter-mines the spatial distribution of economic activity.2For parameter values for which the agglomerationforces outweigh the dispersion forces, an uneven or core-periphery pattern of economic developmentcan emerge. As new economic geography models typically abstract from first-nature geography by as-suming that locations are symmetric, which region becomes a core or periphery is not determined withinthe model. As a result, a central implication of these models is that for a range of parameter valuesthe distribution of economic activity is not uniquely determined by locational fundamentals but insteadexhibits multiple equilibria.Although a rich and extensive body of theoretical research on new economic geography has nowemerged, empirical research remains comparatively less well developed. This paper briefly reviews theexisting empirical literature on new economic geography, focusing on areas in which progress has beenmade and highlighting possible avenues for future inquiry. The discussion is necessarily selective and no1In Krugman (1991a,b), geographically mobile manufacturing workers are the source of agglomeration, while in Krugmanand Venables (1995) and Venables (1996) tradeable intermediate inputs play this role. While initial research modeled the loveof variety using a Constant Elasticity of Substitution (CES) functional form, recent research has also considered the quasi-linearfunctional form following Ottaviano, Tabuchi and Thisse (2002).2In Krugman (1991a,b) geographically immobile agricultural workers provide the dispersion force, while Helpman (1998)emphasizes housing’s role as a non-traded amenity.2attempt is made to be comprehensive.3The remainder of the paper is structured as follows. In Section2, we discuss the relationship between factor prices and market access. In Section 3, we examine thedeterminants of the location of production. Section 4 concludes.2 Market Access and WagesAs discussed above, a key implication of love of variety, increasing returns to scale and transport costs isthat firms have an incentive to concentrate production close to large markets. In equilibrium, this resultsin either greater production and/or higher equilibrium factor prices in locations close to large markets.To derive the implications of new economic geography models for factor prices, consider the Krug-man and Venables (1995) model, in which labor is immobile across locations and a monopolistically-competitive manufacturing sector uses tradeable intermediate inputs. A first condition for producer equi-librium is profit maximization, which with constant elasticity of substitution (CES) preferences impliesthat prices are a constant mark-up over marginal cost. A second condition for producer equilibrium iszero equilibrium profits, which together with profit maximization and a homothetic cost function impliesthat equilibrium output of each variety is equal to a constant (¯x). Combining this result with equilibriumdemand for manufacturing varieties, and summing demand across all locations, the equilibrium price ofeach