Spotlight Research Faculty
Martin Rotemberg
Railroads and the Rise of American Manufacturing
In the 19th century, American manufacturing developed alongside a spreading railroad network that integrated large domestic markets with vast land and commodity resources. My research with Rick Hornbeck studies the impacts of railroads on manufacturing growth, in which we show how the canonical social savings approach1 can miss substantial impacts of railroads on economic growth.
Social savings calculations are an appealing method of welfare analysis that, in this context, implies the economic gains from the railroads are bounded above by the total increase in shipping costs from transporting the same quantity of goods if the railroads were removed.2 This social savings approach assumes that the marginal social value of all other economic activity is approximately equal to its marginal cost.3 In the presence of distortions that drive a wedge between the social benefit and private costs of a firm expanding production, such as markups or credit constraints, then the rationale behind social savings calculations breaks down: there are further welfare effects from the expansion of other sectors in response to railroads decreasing transportation costs. Strikingly, while Fogel emphasizes that assuming an inelastic demand for transportation provides an upper bound estimate on the railroads’ impacts, the opposite becomes true in the presence of these market distortions: a greater elasticity of demand for transportation magnifies the economic impacts of the railroads by yielding greater changes in activities whose value marginal product exceeds marginal cost across other sectors in the economy.
We examine the impacts of railroads on manufacturing productivity growth, as internal domestic markets became increasingly integrated into the United States and the manufacturing sector grew dramatically in the latter half of the 19th century. We calculate how changes in the national railroad network affected county “market access,” and estimate the impacts of changes in county market access on changes in manufacturing productivity.4
Estimating the impacts of market access has several advantages. First, there is a great deal of within-state variation in changes in county market access even after controlling flexibly for changes in local railroad density, which helps to address the potential endogeneity of local railroad construction. Much of this variation comes from the natural geography of the United States, for instance, railroads increased market access by more in places that didn’t already have good access to navigable waterways. Second, county market access is derived from a general equilibrium trade model that captures how counties are affected both directly and indirectly by changes in county-to-county transportation costs, and the estimated relative impacts of market access can then be used to explore aggregate economic impacts from counterfactual transportation networks (e.g., removing the railroads or replacing the railroads with potential extensions to the canal network). We draw on the productivity literature to examine impacts on manufacturing productivity from increased market integration and changes in resource allocation.
Our results so far show that increased market access generates substantial increases in county manufacturing productivity. The overall increase in productivity is driven by gains in reallocative efficiency, as increased market access leads to greater input usage in areas with high-value marginal product relative to marginal cost. Non-parametric graphs show the estimated impacts of market access on productivity and output are log-linear (as predicted by the theoretical derivation of market access). As shown in the figure below, the linearity property is preserved both if we use aggregate county-level data or the disaggregated county/industry data that we digitized.
Aggregate economic impacts reflect not only the reallocation of a fixed amount of inputs within the United States economy but also gains from increasing aggregate inputs in the United States economy (e.g., through migration). In ongoing work, we are using our estimated relative impacts of market access on manufacturing, identified off of observed relative changes in market access across counties, to estimate counterfactual aggregate changes in county-level market access under various scenarios, such as replacing the railroad network with an extended canal network. We find that the railroads were responsible for a large fraction of manufacturing growth over the late 19th century. We are currently digitizing the complete Census of Manufacturers for 1850-80 in order to build on our results.
1 Fogel, Robert W. 1964. Railroads and American Economic Growth: Essays in Econometric History. Baltimore: Johns Hopkins University Press.
2 The railroads represent a productivity improvement in the transportation sector, and the general intuition is that the economic gains from that productivity improvement are captured by the demand curve for transportation.
3 There was an important debate between Paul David and Robert Fogel on specific examples of when marginal costs could be less than marginal social values, such as the extent of increasing returns to scale.
4 This work relates to research on the economic impacts of railroad construction by authors such as Jeremy Atack, Fred Bateman, Michael Haines, and Robert Margo.