Dave Donaldson, Clark Medalist 2017
American Economic Association Honors and Awards Committee
Dave Donaldson is an empirical trade economist, with some of his work straddling the intersection of international trade and development economics. He has not only established himself as a leader in the field, but he has also formed and become the principal practitioner of a distinctive style of research, based on important conceptual questions, careful data work and credible identification combined with state-of-the-art structural methods. Donaldson's research adds profound insight into classic and fundamental issues in international trade.
Implications of Within-Country Market Integration
Donaldson’s paper “Railroads of the Raj: Estimating the Impact of Transportation Infrastructure?” (American Economic Review, forthcoming) investigates the economic benefits from building transportation infrastructure studying the case of railways in 19th century India. This paper is widely viewed as both a methodological breakthrough and substantively important paper in the field. Donaldson assembled a new and rich data set from archival sources about the expansion of railroads in India through the 19th and early twentieth century and the volume of inter-regional trade in the same period. He then uses the data to look at the effect of access to railroads on real agricultural incomes. To check that this effect does not come from building railroads where the growth was predicted to be, he uses the fact that a number of proposed lines did not get built or did not get built when they were proposed to be built. Assuming that the proposal was based on what the contemporary experts thought were the areas of greatest demand for transportation, these un-built railroads should also have an effect if they were any good at predicting growth. He finds no such effect.
The second part of the paper builds a quantitative model where the effect of trade on real agricultural GDP is fully captured by one sufficient statistic: the share of expenditure that each Indian district allocates to goods produced in the district. When that share is low, it indicates that the relative price of imports in the district is low, and in turn, that the welfare gains from trade are large. Controlling for shocks to technology (mainly rainfall in this case), he finds that observed changes in real GDP following access to the railroad move almost one for one with the sufficient statistic predicted by the model, thereby making the case that the benefits of the railways is indeed the result of increased trade.
A second paper on the theme of intra-country trade costs is “Railroads and American Economic Growth: A ‘Market Access’ Approach” (Quarterly Journal of Economics 2016, with Hornbeck). This paper uses the expansion of railroads in the United States to revisit the question posed by Fogel’s seminal work on American railways. Donaldson and Hornbeck use county level data from the 19th century on agricultural land values, combined with detailed GIS maps that show when railroads arrived in each county. They then use the entire network to derive each county’s “market access,” equal to the trade-cost weighted distance to other markets, which are weighted in turn by their access to other markets. Then they estimate the impact of the railroads on land values by regressing changes in land values on changes in market access induced by the expansion of the railroads. These estimates suggest that in a counterfactual world without railroads, U.S. GNP would have decreased moderately more than according to Fogel’s “social savings” estimates.
“How Large Are the Gains from Economic Integration? Theory and Evidence from U.S. Agriculture, 1880-2002,” co-authored with Costinot continues on this theme, using a new approach to measure the gains from economic integration within the context of agricultural markets. Their approach is based on an assignment model in which heterogeneous land is allocated to multiple sectors/crops in multiple local markets. They implement this approach using data on crop markets in 1,500 U.S. counties from 1880 to 2002. They directly observe the modern production function of different crops for each type of fields, which they obtain from the agronomic Global Agro-Ecological Zones (GAEZ) project from the Food and Agricultural Organization (FAO). They do not observe productivity shocks going back into the past, but instead make the key identifying assumption of that paper that these do not reverse comparative advantage of fields for particular crops. Under this assumption, they show how to use data on total farm sales in the county, total output per crop, and total land allocated to each crop to infer the unique vector of crop prices and crop-and-county productivity shocks that are consistent with profit-maximization and factor-market clearing in the county. Since production functions are linear, this can be done by solving computationally tractable linear programming problems. Their exercise is much in the spirit of showing that general equilibrium theory has content; here profit-maximization and factor market clearing implies that relative prices should be equal to the slope of the production frontier evaluated at the point at which production takes place.
As a test of the model, Costinot and Donaldson check if the constructed county-level prices resemble the observed state-level prices; they observe 50% correlation for 2002 and 80% correlation for 1880. They next estimate the price wedges relative to some reference price, for example, the commodity price in a county relative to the commodity price in Chicago. Under the assumption of perfect competition, these wedges can be interpreted as measures of transportation costs. They document that these price wedges have systematically declined from 1880 to 2002. In this sense, commodity markets have experienced a great degree of integration in 2002 relative to 1880. They then consider a counterfactual exercise, where they set the 1880 price wedges to their 2002 levels. They compute the allocation of land, production, and total value of output in 1880 under the new counterfactual prices, and compare them to the actual total value of output in 1880. They find that this leads to a 90% increase in total output. This gives a measure of the productivity gains in the transportation sector, which can be thought of more broadly as “the gains from economic integration.” This is a very important paper, once again both substantively and methodologically, and moreover delivers a provoking result regarding the gains from integration/trade (they are of the same magnitude as productivity gains in agriculture).
Empirical Investigation of Comparative Advantage
Donaldson has done original work providing empirical evidence on the the theory of comparative advantage. With Costinot and Komunjer, his 2011 Review of Economic Studies paper “What Goods Do Countries Trade? A Quantitative Exploration of Ricardo’s Ideas?” develops a rich structural Ricardian model, extending Eaton and Kortum’s famous setup. They quantify the extent to which the productivity differences determine trade flows. This approach is a significant advance to theories that explain trade flows through differences in endowments (as in the Heckscher-Ohlin model).
The paper is a theoretical investigation of the Ricardian model with many countries and many goods and multiple varieties of each good. The model captures the prediction that countries should export relatively more in sectors in which they are relatively more productive by providing closed-form solutions for bilateral trade flows as a function of observed productivity. The model takes into account that all countries will not produce all varieties of every good, but rather those varieties in which they are relatively more efficient. This implies that differences in observed productivity tend to be smaller than true differences in productivity (since countries with low true productivity levels produce the relatively small number of varieties for which they received good productivity draws). Combining the empirical equation that comes from the theory with data on trade flows and producer prices (which should be proportional to productivity in the model), they first show that countries do export goods where their relative productivity is higher, as one might have expected. Using their estimate of the key structural parameter of the model (the dispersion of productivity across varieties within a sector), they then quantify the welfare impact of this Ricardian channel across sectors. They find that it only generates a small part of the gains from trade highlighting, perhaps, that comparative advantage operates mostly at the within-industry product level. This is an important paper mainly because it shows how to confront the predictions of the Ricardian model to the data without having to rely on ad-hoc measures of export performance (something that had lead many trade economists to focus on cross-country differences in factor endowments rather than technology as the main source of international specialization).
Arguably another one of Donaldson’s most promising work is the new paper, “Nonparametric Counterfactual Predictions in Neoclassical Models of International Trade,” (American Economic Review, forthcoming). In this paper, Donaldson and co-authors Adao and Costinot develop a new methodology to construct nonparametric counterfactual predictions, free of functional-form restrictions on preferences and technology, in neoclassical trade models. To do so, they first establish the equivalence between such models and reduced models in which countries directly exchange factor services. This equivalence implies that, for an arbitrary change in trade costs, counterfactual changes in the factor content of trade, factor prices, and welfare only depend on the shape of a reduced factor demand system. They then provide sufficient conditions under which estimates of this system can be recovered nonparametrically. Together, these results offer a strict generalization of the parametric approach used in gravity models.
“Evolving Comparative Advantage and the Impact of Climate Change in Agricultural Markets: Evidence from a 9 Million-Field Partition of the Earth” (Journal of Political Economy, 2016, with Costinot and Smith) uses very rich data to investigate the consequences of climate change how changes in production patterns within countries might mitigate the adverse consequences of climate change. The analysis demonstrates that there are large potential gains associated with adapting the production but little advantage to changes in the pattern of trade. The agricultural productivity specification again makes use of the GAEZ productivity data for 10 distinct crops. This dataset uses agronomic models and high-resolution data on geographic characteristics and climatic conditions to predict the yield that would be obtainable crop by crop at each of more than 9 million grid cell covering the surface of the Earth. Crucially, the GAEZ dataset is available both under contemporary growing conditions and under the climate change scenarios used by the UN’s Intergovernmental Panel on Climate Change (IPCC). Consequently the authors can conduct counterfactual exercises. The paper finds that climate change aggregates to a large negative productivity shocks for many countries around the world, that is if there were no reallocations around the world welfare would decrease substantially. However, there is enough heterogeneity in these shocks over space that after reallocating production according to comparative advantage across crops within each field, welfare changes become smaller by an order of magnitude. Furthermore, there is so much productivity heterogeneity across fields within countries that opportunities for international trade appear to have small effects on the welfare consequences of climate change.
“The More We Die, the More We Sell? A Simple Test of the Home-Market Effect?” (with Costinot, Kyle, and Williams) investigates the extent to which domestic demand creates comparative advantage. It explores the home-market effect in the context of drug exports. Building on previous work on the effect of demographic changes on innovation and product entry, the paper establishes that countries that for demographic reasons are expected to have high demand for a certain type of drug are actually more likely to be net exporters of the same drugs. They find that correlation between predicted home demand and sales abroad is positive and greater than the correlation between predicted home demand and purchases from abroad, which is strong evidence for the role of the home market in creating comparative advantage.
Another recent paper is “The Elusive Pro-Competitive Effects of Trade” with Arkolakis, Costinot, and Rodríguez-Clare. This paper is a sequel to the AER paper by Arnaud, Arkolakis and Rodríguez-Clare. An important limitation of this earlier paper is that it abstracted from any potential “pro-competitive effects” of greater integrations by hard-wiring constant monopoly markups. Indeed, one may have conjectured that the gains from trade could be much higher if higher trade intensifies competition. Motivated by this observation, the new paper considers a richer class of models that allows for demand to be non-isoelastic and thereby for trade to have an impact on monopoly markups. The class of models, while restrictive, contains many models with variable markups that have been used in the previous trade literature. The surprising result is then that, at the estimated demand parameters, the effect of trade on markups turns out to reduce, rather than increase, the welfare effects of trade. More precisely, the welfare gains are bounded from above by the same macro-level elasticity-based estimates provided in Arnaud, Arkolakis and Rodríguez-Clare’s AER paper.
The basic insight behind this seemingly paradoxical result is that a reduction in the barriers to trade has two opposing effects on monopoly markups: it reduces the markup of domestic firms (since they face tougher competition from abroad), but it increases the markups of foreign firms relative to the isoelastic case (since they now have lower costs of serving the domestic market and decreases in costs tend to be incompletely passed-through). Given their estimated demand parameters, the second effect turns out to dominate. This result qualifies some important conditions under which we may expect the “pro-competitive” effects of trade to be present, and it provides yet another useful benchmark by encompassing the kind of models that trade economists commonly use for studying the effects of trade liberalization.
Dave Donaldson is not only the most exciting economist in the area of empirical trade, but has also made several important methodological and substantive contributions.