Working Papers

Silk Roads to Riches: Persistence Along an Ancient Trade Network (with Zofia Ahmad)
Revise and Resubmit at the Journal of Economic Growth. Version Updated: February 2021. [SSRN]
Nominated for Best Paper (Applied Economics) at the Econometric Society's DSE Winter School 2020

The Silk Roads were a decentralized network of trade routes that connected ancient cities across Eurasia. Goods, ideas, people, and technology moved along the roads for over 1,500 years. Using a detailed georeferenced map of the entire trade network, this paper finds that areas within 50 KM of the historic location of the Silk Roads have higher levels of economic activity today. The persistent effect of proximity to the ancient trade network is associated with increased access to modern transportation infrastructure and the historical diffusion of technology along the routes but cannot be explained by differences in contemporary or historical levels of population density. This analysis is complemented by individual-level data from 22 countries; we find that districts with populations closest to the Silk Roads have higher rates of inter-group marriage, suggesting a weakening of social boundaries between groups that might possess differential technological knowledge.

Combining Continuous and Categorical Responses as Dependent Variables: The Easiest Solution Works (with William N. Evans)
Revise and Resubmit at the Southern Economic Journal. Updated: October 2021. [SSRN]

To combat rising item non-response in household surveys for some continuous variables, surveys are increasingly supplementing these questions with categorical responses or “unfolding brackets.” This survey technique is particularly prevalent in health surveys. Using data from four different data sets with this type of survey question, we demonstrate that those not responding to the continuous question are fundamentally different from those that answer, and ignoring the categorical data in regression models when it is the outcome of interest induces a sample-selection bias. Making distributional assumptions about the error in the regression allows for the combination of the categorical and continuous data in a maximum likelihood estimation (MLE) model. A “quick fix” that uses the means from the continuous variable within a category to impute a value for those that respond to the categorical question allows one to estimate the model by OLS. Results from our examples show that this method yields estimates that are strikingly similar to those from the MLE model.

State Regulation and Competition in Outpatient Surgery Markets (with Elizabeth L. Munnich)
Updated: March 2021.

U.S. healthcare markets have been characterized by increasing levels of hospital consolidation, alongside the introduction of substitutes for services that were previously only available in hospitals. We consider this in the context of the outpatient surgery market, where ambulatory surgery centers have grown in prominence as an alternative to outpatient surgery in hospitals. We focus on the impact of regulation on competition in outpatient surgery markets by exploiting variation in Certificate of Need (CON) laws over time and across states, as well as along state borders. The removal of these regulations is associated with an increase in the number of ASCs per capita, counties in which ASCs operate, and share of a county's outpatient procedures performed in ASCs. At the same time, we find no evidence that this deregulation leads to hospital closure, or an increase in the total volume of outpatient procedures. These findings contribute to the literature on competition in healthcare markets, and have important implications for policies designed to constrain healthcare spending.

Education and Fertility: Evidence from a Policy Change in Kenya
IZA Discussion Paper No. 6778. (Update currently under construction)

This paper investigates the relationship between women’s education and fertility by exploiting a 1985 policy change in Kenya that lengthened primary school by one year. An instrumental variables approach measures the exogenous variation in treatment intensity across birth cohorts. The reform led to an increase in education, a delay in marriage, and reduced fertility beginning at the age of 20. The effect on fertility becomes increasingly negative through age 25. The findings suggest that postponement of marriage, reduction in the marital education gap, and increased early use of modern contraceptives contribute to reduced fertility. These results are consistent with women having greater control over their fertility decision.

Luke Chicoine • Amazon • Senior Economist • Arlington, VA • chicoine.econ@gmail.com