Research

Random Assignment with Non-Random Peers: A Structural Approach to Counterfactual Treatment Assessment, Forthcoming, Review of Economics and Statistics

Efforts to leverage peer effects by changing assignment have often fallen short due to endogenous peer choice. To address this, I build a two-part model: agents form networks via continuous linking decisions; conditional on realized networks, outcomes are determined. I provide results on identification of both parts of the model. Using data from a randomized study in India, I estimate the model, assess its performance in out-of-sample prediction, and simulate outcomes under preferential assignment rules. This paper contributes new methodology for identifying effects of alternative assignments in the presence of network endogeneity, as well as identification of network formation models.

Available here.

The Effects of Public Campaign Funding: Evidence from Seattle's Democracy Voucher Program, 2022. Journal of Public Economics 211:104666 (with Thomas Noonen)

During each election cycle, the city of Seattle distributes four $25 vouchers to every registered voter, which may be donated to and redeemed by campaigns for city office. Through a difference-in-differences research design, we study the causal effect of Seattle's program on various outcomes in city council elections in the first two cycles after implementation, with comparison cities drawn from large cities in Washington and California. We find a 53% increase in total contributions and a 350% increase in number of unique donors; these effects are largely explained by large increases in small donations, defined as contributions less than $200. We find statistically insignificant evidence of decreases in private donations, although our point estimates suggest moderate-to-substantial crowd-out ratios. We further estimate a highly significant 86\% increase in number of candidates coupled with substantial decreases in incumbent electoral success. These results provide some of the first causal evidence of the effect of decentralized public campaign finance schemes, while speaking to the impacts of campaign regulation more generally on donation activity, candidate entry, and incumbency advantage.

Available here.

Identification of Network Structure in the Presence of Latent, Unobserved Factors: A New Result using Turán’s Theorem, Revise and Resubmit, Quantitative Economics (with Sida Peng)

Empirical research on network-based spillovers is often limited by lack of data on network connections. We provide new identification results for linear-in-means processes in which covariance in outcomes across agents is due to both network effects and latent, unobserved factors. Our main result makes prominent use of Turán's Theorem, as we show that restrictions on network density and maximum degree are sufficient to separately identify network structure and latent factors. When there is a single factor, network density can be as high as 1/4, while maximum degree may be as large as 5/12 of all agents. In contrast to prior methods, our results allow for identification of network structure solely from covariance in outcomes, allow for latent rich unobserved heterogeneity, do not require exogeneity assumptions or even observation of time-varying covariates, permit identification of the absence of peer effects, and are not dependent upon any specific network formation model.

Available here.

Efficiency in the Two-Way Connections Model, Revise and Resubmit, International Journal of Game Theory

Well-known theoretical results for the ``connections" model of network formation emphasize the tension between stability and efficiency. We set aside stability in order to characterize efficient networks as generally as possible. Our most general model allows for both (i) general link formation technology, whereby links are costly to form and bounded in the unit interval; and (ii) arbitrary heterogeneity in the value derived from linking to other agents. We provide necessary conditions for efficiency in the general model. We then fully characterize efficient networks in two restricted cases, both of which are generalizations of the seminal models, and we show that well-known results can be derived as special cases of both. In contrast to existing works, we provide alternative, more direct proofs, while also fully characterizing the boundary cases.

Available here.

The Effects of Increasing Community Participation on School Management Outcomes: Evidence from India (with Natalia Cantet, Clara Delavallade, Zhe Liu, and Rebecca Thornton), Under Review

Policies aimed at improving school governance via community-based management are common but understudied. We present the results of a cluster-randomized trial of a multi-faceted education intervention that involved support for local School Management Committees (SMC) in rural India. The intervention increased SMC meetings held by 17 percent and completed school improvement plans by 38 percent, gains that persist for one year after the intervention ends. We find a 14 percent increase in the number of teachers and a 25 percent increase in the likelihood of having a kitchen. We provide suggestive IV analyses to show that the increases in teachers and having a kitchen may be attributed to the increased school management committee activities.

Available here.

The Impact of Missing Links on Linear Reduced-form Network-Based Peer Effects Estimates (with Jungyoun Kim), Under Review

Applied researchers frequently estimate network-based peer effects models using observed network data that includes only a subset of the true links. We consider the effects of this mismeasurement on reduced-form linear-in-means and linear-in-sums estimates. Our results require an assumption that the expected covariance of characteristics between linked agents is the same regardless of whether the link is observed or not. Analytic results show that the linear-in-means peer effects estimate is in general attenuated, and this is a special case of ``classical" measurement error. In contrast, linear-in-sums direct and peer effect estimates may be attenuated, augmented, or consistent; the inconsistency depends upon the missingness mechanism and the relationship between the network and covariates. We demonstrate the effect of mismeasured links in both models using two datasets and through simulations. These results show that the effects of mismeasured networks on subsequent estimands is quite sensitive to the parameter that is being estimated.

Available here.

Network Partitioning and Social Exclusion under Different Selection Regimes (with Clara Delavallade and Rebecca Thornton)

While most social programs are based on some form of exclusion of sub-populations, we know little about how being excluded, and the selection process, affect social inclusion. This paper compares peer effects of an after-school program, under three different (randomly assigned) network-formation regimes: endogenously formed, popularity vote, and randomly assigned. We find substantial evidence of homophily within endogenously-formed and elected networks. When participation was randomly assigned, we find segregation of friendships due to the program. We do not find this among elected networks, mainly because they were already highly partitioned. Lastly, we find that social exclusion – not being elected in a school with popular voting – reduced education aspirations and self-confidence.

Available here.

Local or Imported? Middlemen, Market Power, and Input Sourcing in a Network Economy, with Jonas Hjort and Yue Yu

Input sourcing is central to the process of economic development. We first document that, contrary to conventional wisdom, the share of inputs originating abroad is almost as high in Uganda---a landlocked, low-income country---as in rich, open economies. The monthly transaction-level data we use, which cover the universe of formal, firm-to-firm and firm-to-foreign country trade in the Ugandan economy, also reveal that indirect imports are sourced from few firms, and via short supply chains. Together these three ``new'' facts motivate a model of input sourcing in a networked economy. Inputs, output quantities, and output prices are determined by firms maximizing profits subject to input prices and both a domestic and international sourcing network. Decreases in the cost of imports lead to lower prices for importers' output in the domestic market. We exploit exchange rate shocks and firms' prior direct and indirect import origins to test the model's predictions. When the relative cost of foreign goods decreases, direct importers import more. However, first- and higher-order indirect importers import less, instead buying more foreign content domestically. Pass-on price effects appear to ``magnify'' network structures that economize on high fixed costs of importing, widening access to foreign content but hindering substitution to direct international trade.

Efficient Sampling in Networks, with Jungyoun Kim and Tyler McCormick

Network Influence on Technology Adoption: Evidence from Microfinance Groups in Ghana, (with Emma Riley and Abu Shonchoy)