Wikipedia Articles for Hull, Moscona, and Hornbeck
Some of my favorite economists at their best
Writing Wikipedia pages has been a really fun project, and I expect I will continue doing this for a while. I would like to strongly encourage other people to take this up, however. Write about the people whose work you know and love! I would like to learn new things in economics from Wikipedia.
These are all people whose work I think is incredible. It consistently blows me away.
Rick Hornbeck:
Hornbeck specializes in applying modern econometric methods to historical datasets, commonly using changes in land rents to infer effects. His 2010 paper, “Barbed Wire”, showed how the introduction of barbed wire to the American Plains greatly reduced the cost of enforcing property claims, leading to increased agricultural productivity. Before, farmers had to fence their land in with timber fences, or else have their crops damaged by wandering cattle herds. After, farmers could settle in places which had no nearby timber, and they could switch to crops which were more vulnerable to intrusions by livestock, like corn. Farmers gained about $103 million in value between 1880 and 1890, or 0.9% of GDP.[1][2]
His work has substantially improved our understanding of the role of railroads in American economic growth, contradicting the Fogel hypothesis that they were of relatively small importance.[3] First, with Dave Donaldson, he estimated the value railroads gave to agriculture. A railroad opening in a county affects that county’s, and all other counties, market access, which is given by calculating the least cost path of travel for freight before and after the railroad opened. This allows them to take into account how railroads being built in one place changes travel costs everywhere else. Without the railroad, the value of agricultural land in 1890 in the United States would be 63.5% lower.[4] Then, with Martin Rotemberg, he utilizes firm-level data from the Census of Manufactures to revise Fogel’s assumption of perfect competition. Manufacturing was characterized by profound misallocation away from the optimal use of resources, which access to railroads reduced. They attribute 25% of American economic growth to the railroad.[5]
Hornbeck has several papers which use a particular historical calamity as a source of exogenous variation, in order to uncover deep facts about the world. In “When the Levee Breaks”, with Suresh Naidu, he leverages the 1927 Great Mississippi Flood, and the subsequent emigration of tens of thousands of Black Americans, to uncover the importance of labor costs in whether technologies are adopted. Places which saw more emigration modernized as a response to higher labor costs, in what is indirect evidence for the high-wage hypothesis of British industrialization.[6]
In “Creative Destruction: Barriers to Urban Growth and the Great Boston Fire of 1872” (with Daniel Keniston), he estimates the losses from inefficient urban design. Cities can often be weighed down by past decisions — a street big enough for a quiet town might be jammed with traffic in the bustling metropolis. The city burning down allows for rebuilding more efficiently. Strikingly, plots which burned became more valuable after the fire.[7] In addition, Hornbeck has studied the long-run effects of the Dust Bowl.[8][9]
Hornbeck has also contributed to quantifying spillovers from economic activity onto others, of which his paper on the Boston Fire of 1872 is but one example. With Michael Greenstone and Enrico Moretti, he measured the effect of plants opening in a county on other plants in the same business by comparing the productivity of plants in the county a firm decided to open their factory in, compared to the runner-up county. Incumbent firms became 12% more productive after five years, compared to counties which faced lessened competition.[10] This ties into a long literature linking increased competition to increased productivity.[11][12][13]
Peter Hull:
Hull is best known for his work in econometrics. In particular, he is known for his work with Kirill Borusyak and Xavier Jaravel on shift-share research designs, or Bartik instruments.[3] Shift-share instruments use a national trend in some category, interacted with the local share in that category, to make a causal argument.
To give a concrete example, take the famous paper of Autor, Dorn and Hanson (2013) on the U.S. opening to trade with China.[4] Each region has a different level of exposure to the goods which China would export to the United States. When trade opens up with China, a region with labor-intensive goods faces a larger shock to employment than a region with capital-intensive goods, or which specializes in services. Of course, what China specializes in might be endogenous to what the US produces, so they create an instrument of exports to 8 other countries. They can then use this to make a big causal argument that import competition caused a quarter of the aggregate decline in US manufacturing employment during that time period.
The Borusyak-Hull-Jaravel approach allows for shares to be endogenous so long as shocks are exogenous, and shows that exogenous shares and endogenous shocks are equivalent. These shocks can be large in number, and pooled together. In addition, they give an estimator which addresses the criticisms of Adão et al (2019), who point out that areas with similar shares likely came to those shares for similar reasons, and are thus not truly independent observations.[5] They themselves raise concerns about studies with few or insufficiently dispersed shocks, which may in practice have too small samples. Their approach has become the default for researchers, and they were called upon to popularize their work in the Journal of Economic Perspectives.[6]
Hull has used these tools on numerous papers measuring racial discrimination, educational interventions, and healthcare. In “Mortality Effects and Choice Across Private Health Insurance Plans” (with Abaluck, Bravo, and Starc) he studies how much different healthcare plans can reduce mortality, and how able consumers are to assess which plans are better. One cannot simply regress observed mortality on cost to find the most effective plans, as the observed rate of mortality could be driven by unobserved selection and sorting. To overcome this, they employ an instrumental variable strategy, where the exit of a particularly bad or good plan causes people to shift onto better plans. They estimate that shifting seniors away from the worst 5% of Medicare Advantage plans would save 10,000 lives a year.[7][8]
Hull, with Alsan, Barnett, and Yang, measured the effectiveness of Flint, Michigan's “IGNITE” program in reducing recidivism. Ordinarily, since the program was done by all inmates — and even if it weren’t, spillovers between groups might bias the estimate down — we would not be able to test the effectiveness. They used as instrumental variable quasi-random court delays. Some people get held in jail for longer than others for arbitrary reasons, so they spend a longer time exposed to the program. In contrast to the conventional wisdom that “nothing works” in rehabilitation, they found that recidivism was reduced by 25%.[9][10]
Hull has also used instrumental variables to tackle racial discrimination in bail discrimination. One cannot simply look at a disparity and infer bias. It could reflect a true difference in the underlying rate of pretrial misconduct, or behaviors which are observed by the judge but not by the economist studying it. One cannot control for criminal status as an indicator of dangerousness, because those past criminal convictions could also be due to bias. If we knew the true rate of pre-trial misconduct then the random assignment of judges is sufficient to wipe away those unobserved behaviors, but we cannot know the true rate of misconduct simply by looking at the sub-sample of prisoners who were granted bail.
Hull and Arnold and Dobbie are able to exploit the randomized assignment of judges to different cases to infer the rate of misconduct. A judge who gave bail to all prisoners would reveal the true distribution of risk; while this judge does not exist, they can extrapolate from judges of different leniency. They attribute two thirds of the disparity in pre-trial release between Black and white defendants to racial discrimination.[11]
Jacob Moscona:
Moscona studies how technological change can ameliorate the consequences of climate change, and how social structure affects outcomes in sub-Saharan Africa. His first paper, “Segmentary Lineage Organization and Conflict in Sub-Saharan Africa”, with Nathan Nunn and Jim Robinson, tests whether societies with a greater sense of duty to distant relatives have longer-lasting conflicts. They find that it does, and that conflicts between a few individuals are more likely to escalate into large group conflicts when people are obliged to come to the aid of their distant cousins.[2] Moscona and Seck investigate how different social structures — whether people feel obliged to their peers of the same age, or to their extended family members — affect who gets transfers. Transfers to adults improve the welfare of children in kinship-based societies much more than in age-based societies.[3][4] Their results challenge the external validity in assessments of cash transfers, such as in Egger et al (2022),[5] and shows how welfare effects can be culturally contingent.
Moscona, largely with Karthik Sastry, has also studied technological change and climate change. In “Does Directed Innovation Mitigate Climate Change?”, they use changes in climate conditions, interacted with how this changes the optimal mixture of crops, to find the elasticity of innovation to market conditions. They estimate that technological change could ameliorate 13% of future climate change damages.[6][7][8] That technology is developed to meet the needs of the developed world means that the developing world receives technology which is mismatched to their ecological conditions. In “Inappropriate Technology”, Moscona and Sastry attribute 15-20% of the differences in agricultural productivity between countries to this.[9]