High-Speed Railways and Local Growth: An Empirical Assessment Based on the French Experience
Koning Martin, Paris East University - IFSTTAR - SPLOTT, Blanquart Corinne, Paris East University - IFSTTAR - SPLOTT, Delaplace Marie, Paris East University – IFU - Lab’Urba
The paper sets out a model aimed at assessing the economic impacts of high-speed railways (HSR) in France.Because the choice of cities served by HSR is not random, we use a selection model to correct our estimates of the wider economic impacts.
While the virtuous relationship between transport infrastructures and local development remains one the most believed myths, the empirical studies looking at the effects of high-speed rail services show mitigated results. This seems problematic because the major cost of these infrastructures requires reliable assessments, notably to select and schedule the building of competing projects.
This article tries to identify the local economic impacts generated by high-speed railways in France. We provide an exploratory analysis of 492 French urban units (UUs), observed over 1982-2006. The variable of interest is the evolution of the employment growth rate, with a focus on executive jobs of the “superior metropolitan functions”.
This analysis lies on a double originality. The econometric study of the traditional “key drivers” of the employment’s evolution questions the role of UUs’ initial size, the role of specialization or functional diversity and the influence of geographical or historical amenities. In addition, we differentiate the observation across time (before and after the creation of the high-speed service) and across UUs, the ones that have received the “treatment” (one high-speed train – HST - partly or completely placed on a high-speed line - HSL) and the ones that haven’t. As the treatment is a priori not random, we combine the estimation of the “treatment effect” on local growth with the estimation of an “selection equation” that aims at explaining the choice of the UUs benefiting from the public policy.
The estimates show that empirical assessments of railway services cannot neglect endogeneity issues between infrastructures and local development. Thus the selection model shows results with an opposite sign to the ones based on the Ordinary Least Squares method. Our estimates particularly emphasize the differences between UUs served by HSL or simply by HST. The UUs served on HSL have faced a lower rate of jobs (executives) creation but the intrinsic effects of infrastructures on the annual growth rate are about +1.3% (+3.7%), once corrected from the selection bias. By contrast, the UUs served by HST partly placed on classical railway have experienced better average performances but the effect directly due to infrastructures is negative: -1.1 % and -3.0 % respectively. These results could be useful to complete the traditional socio-economical assessments of infrastructure projects.
Association for European Transport