Project Abstract/Summary
This CAREER research project will develop new econometric and statistical tools for identifying and estimating causal effects in settings where the current approaches impose unrealistic assumptions. The validity of every econometric method is grounded in the assumptions placed on the model. In some cases, these assumptions may be fundamental to the identification and estimation strategy, but ultimately unrealistic in real economic settings. Research that uses these strategies to estimate policy effects or causal impacts may be unreliable if these assumptions are not true. This project will provide alternative strategies that are valid without unrealistic assumptions. The project also will develop methods to test for the legitimacy of underlying assumptions. To ensure the usability of the proposed methods, code will be developed for popular statistical software and will be posted in publicly accessible software repositories. By examining what fundamental assumptions are used to identify causal effects, what the implication of these assumptions are, and how to determine if the assumptions are realistic, this project will contribute to the continual improvement of accurate empirical analysis, an essential input for successful economic policy making. In terms of educational activities, the investigator will support the pipeline of women into economics through the development and mentorship of undergraduate student groups and undergraduate research experiences. To facilitate that activity, the investigator will develop and analyze surveys to determine which student activities are most impactful and make available supporting material on public online platforms for other educators to use.
In the estimation of causal effects and policy evaluation, models of endogeneity are commonplace as individuals or states who choose treatment or to enact a policy often are different than those who do not in unobservable ways. The control function approach is a popular method that addresses endogeneity by controlling for the unobserved differences. However, current control function approaches are limited by unnecessary and often unrealistic restrictions in the identification strategy. This research consists of three projects aimed at eliminating these restrictions and developing a more general control function approach. First is the identification of triangular random coefficient models with continuous endogenous variables. Unlike current control function approaches in the literature, this project will develop a general control function approach that does not depend on scalar heterogeneity in the first stage. The second project will extend the first to the setting of binary endogenous variables; i.e., the policy evaluation setting. The current literature depends critically on a monotonicity assumption in the first stage to identify local average treatment effects, whereas the proposed general control function approach will not. The final project will explore the sensitivity of control function approaches to misspecification of nonlinearity and heteroskedasticity as well as provide guidance to empirical researchers on how to compose specifications that are flexible enough to avoid these issues.
This award reflects NSF’s statutory mission and has been deemed worthy of support through evaluation using the Foundation’s intellectual merit and broader impacts review criteria.
Principal Investigator
Alyssa Carlson – University of Missouri-Columbia located in COLUMBIA, MO
Co-Principal Investigators
Funders
Funding Amount
$130,886.00
Project Start Date
06/15/2024
Project End Date
05/31/2029
Will the project remain active for the next two years?
The project has more than two years remaining
Source: National Science Foundation
Please be advised that recent changes in federal funding schemes may have impacted the project’s scope and status.
Updated: April, 2025