“Transport Costs, Geography, and Regional Inequalities”

Pierre-Philippe Combes and Miren Lafourcade

IED Discussion Paper 122, April 2002

The interaction of economic growth and geography and its impact upon the inequality of incomes and opportunities across regions has fascinated and perplexed economists for a long time. No consensus has emerged theoretically or otherwise on whether a fall in transport costs tends to exacerbate or dampen preexisting regional inequalities. On the one hand, increasing returns imply the shift of more productive resources away from peripheral and towards core regions. The latter thereby benefit disproportionately from the higher diversity of goods and larger market sizes. On the other hand, increasing competition in factor markets may reverse this trend, implying eventual convergence. However, a simple time series analysis of employment data across regions (in France for example) seems to lay low such a simple story. Empirical work in this area has, in turn, been plagued by corresponding difficulties. These include data constraints on transaction costs that have prevented a satisfactory exploration of network effects in interregional trade, and ad hoc assumptions regarding the nature and magnitude of transport costs. This perhaps explains the disparate results flowing from this literature.

In this paper, Combes and Lafourcade argue that the two central problems

that need to be overcome are data precision and the correspondence between the theoretical model and its empirical specification. Since transport costs, intermediate inputs, and real geography are all critical in the distribution of activities in France, they utilize a new data set on transport costs between the entire 341 French Employment Areas (EA) to derive a detailed measure of transport costs between the EA’s. In order to address the second issue, they develop a tractable and fully-specified economic geography model. This model enables the derivation of a key relationship between sectoral local employment levels and transport costs, regional incomes and the number of firms. A structural estimation demonstrates that for most sectors, the theoretical model is not rejected and that the estimated coefficients are consistent with plausible values for structural parameters.

Combes and Lafourcade use the estimated model to provide simulations on local sectoral employment and production conditions. A strong core-periphery profit structure seems to imply the existence of large concentration incentives in the French economy. Moreover, whereas strong agglomeration effects seem to strengthen specialization at the local level in many French regions with declining transport costs, it appears that at the national level, dispersive forces are becoming prevalent. These observations coincide fairly closely to the time series estimates of relative employment growth in Paris, Lyon, and Marseille between 1978 and 1993.