American Economic Review, 108 (11): 3339-76
Awarded FREIT-EIIT Best Graduate Paper Prize, 2014
Media coverage: Trade Talks, AEA Research Highlights, Vox, CentrePiece, chrisblattman.com, Bloomberg, pseudoerasmus.com, Defacto (in Hungarian)
This paper uses a natural experiment to estimate the causal effect of temporary trade protection on long-term economic development. I find that regions in the French Empire which became better protected from trade with the British for exogenous reasons during the Napoleonic Wars (1803-15) increased capacity in mechanized cotton spinning to a larger extent than regions which remained more exposed to trade. In the long-run, regions with exogenously higher spinning capacity had higher activity in mechanized cotton spinning. They also had higher value-added per capita in industry up to the second half of the 19th century, but not later.
Away from Home and Back: Coordinating (Remote) Workers in 1800 and 2020
with Mara P. Squicciarini and Nico Voigtländer
This paper examines the future of remote work by drawing parallels between two contexts: The move from home to factory-based production during the Industrial Revolution and the shift to work from home today. Both are characterized by a similar trade-off: the potential productivity advantage of the new working arrangement made possible by technology (mechanization or ICT), versus organizational barriers such as coordinating workers. Using contemporary data, we show that organizational barriers seem to be present today. Without further technological or organizational innovations, remote work may not be here to stay just yet.
Technology Adoption and Productivity Growth: Evidence from Industrialization in France
with Mara P. Squicciarini and Nico Voigtländer
We construct a novel plant-level dataset to examine the process of technology adoption during a period of rapid technological change: The diffusion of mechanized cotton spinning during the Industrial Revolution in France. We document new stylized facts that can help explain why major technological breakthroughs tend to be adopted slowly and – even after being adopted – take time to be reflected in aggregate productivity statistics. Before mechanization, cotton spinning was performed in households, while production in plants only emerged with the new technology around 1800. This allows us to isolate the plant productivity distribution of new technology adopters in mechanized cotton spinning. We find that this distribution was initially highly dispersed. Over the subsequent decades, mechanized spinning experienced dramatic productivity growth that was almost entirely driven by a disappearance of plants in the lower tail. In contrast, innovations in other sectors (with gradual technological progress) shifted the whole productivity distribution. We document rich historical and empirical evidence suggesting that the pattern in cotton spinning was driven by the need to re-organize production under the new technology. This process of ‘trial and error’ led to widely dispersed initial productivity draws, low initial average productivity, and – in the subsequent decades - to high productivity growth as new entrants adopted improved methods of operating the mechanized technology.
All aboard: The effects of port development
with César Ducruet, Dávid Krisztián Nagy and Claudia Steinwender
This paper examines the effects of port development on the economy. By using scarce local land intensively, ports put pressure on local land prices and crowd out other forms of economic activity. We use the introduction of containerized shipping – a technology that substantially increased land requirements at the port – to estimate the effects of port development. We find an important role for the crowding-out effect both at the local and at the aggregate level. First, we show that the causal effect of the shipping boom caused by containerization on local population is zero – port development increases city population by making a location more attractive for firms and consumers, but this well-known market access effect is fully offset by the crowding-out mechanism. Second, to measure the aggregate implications, we add endogenous port development to a standard quantitative model of cross-city trade. Through the lens of this model, we estimate that containerization increased aggregate world welfare by 3.95%. However, relative to the positive welfare effects of a trade-cost reduction in standard models, our model implies a sizeable welfare cost associated with the increased land-usage of ports, partly offset by welfare gains from endogenous specialization based on comparative advantage across port- and non-port activities. In terms of the distributional effects, we find that initially poorer countries gained more from containerization as they had a comparative advantage in port development.
previously circulated as "Drivers of Fragmented Production Chains: Evidence from the 19th century"
This paper uncovers a novel mechanism through which information frictions matter for trade in differentiated goods; the product specification mechanism. We estimate the effect of a reduction in communication time on imports of three product categories in 19th century cotton textile trade; yarn, plain cloth, and finished cloth. In order to identify causal effects, we use exogenous variation in the ruggedness of the submarine seafloor to predict in which year countries get connected to the global telegraph network. The telegraph dramatically reduced the time it took to exchange information expressed in words, but did not affect the exchange of physical objects such as product samples. Using evidence from cotton traders' communication, we show that the examined three products differed in their codifiability, that is, in the extent to which merchants specified product attributes in words. Empirically, we find that communication time reductions had the largest effect on imports of the most codifiable product; yarn, and the smallest effect on the non-codifiable product, finished cotton cloth. Our results suggest that the effect of ICT on trade and fragmentation of production depends on the technology-specific codifiability of product specifications.