The Causal Effect of Transport Infrastructure: Evidence from a New Historical Database

Abstract

In this paper, we analyze the effect of transport infrastructure investments in railways. As a testing ground, we use data from a new historical database that includes annual panel data on approximately 2,400 Swedish rural geographical areas during the period 1860-1917. We use a staggered event study design that is robust to treatment effect heterogeneity. Importantly, we find extremely large reduced-form effects of having access to railways. For real nonagricultural income, the cumulative treatment effect is approximately 120% after 30 years. Equally important, we also show that our reduced-form effect is likely to reflect growth rather than a reorganization of existing economic activity since we find no spillover effects between treated and untreated regions. Specifically, our results are consistent with the big push hypothesis, which argues that simultaneous/coordinated investment, such as large infrastructure investment in railways, can generate economic growth if there are strong aggregate demand externalities (e.g., Murphy et al. 1989). We used plant-level data to further corroborate this mechanism. Indeed, we find that investments in local railways dramatically, and independent of initial conditions, increase local industrial production and employment on the order of 100-300% across almost all industrial sectors.