Research

Working Papers

Causal Mediation Analysis in a Generalized Regression Model (Job Market Paper)

We consider a unifying framework to test for direct and indirect treatment effects in nonlinear models. Specifically, we extend a generalized linear-index model to incorporate endogenous treatments and endogenous mediators. We propose a kernel-weighted Kendall's tau statistics, which is a nonparametric rank correlation estimator, to test the significance of the direct and indirect effects of endogenous treatments on the outcome variable mediated by endogenous mediators. The proposed semiparametric model allows for treatments and mediators to be discrete, continuous, or neither of these two (e.g., censored or truncated). For the indirect effect, we construct two distinct kernel-weighted Kendall's tau statistics that capture the effect of (i) the treatment on the mediator, and (ii) the mediator on the outcome. Unfortunately, standard joint hypothesis tests using these statistics are severely under-sized, a problem that has been noted for linear causal mediation models. To address the problem, we apply a new testing method (van Garderen and van Giersbergen [2020]) that has correct size. As an empirical illustration, we assess the effect of education level on social functioning mediated by individual income, using the British Household Panel Survey data. 

Presentations: Texas Econometrics Camp 2023, EcoSta 2023, Midwest Econometrics Group Conference 2023

Identification of Individual Direct and Indirect Effects in Nonseparable Models [draft coming soon]

In this paper, I establish nonparametric identification of individual direct and indirect effects in nonseparable models. I consider a case where the endogenous treatment is binary and endogenous mediator variable is continuous. While the instrumental variable for the treatment is restricted to be binary, the instrumental variable for the mediator may be discrete or continuous. The fundamental problem of causal mediation analysis is that only outcome and mediator values for the same treatment status can be observed in the data. To handle the identification issue, I introduce the following two steps: (i) I show that we can extrapolate counterfactual mediator variable values using the result of Vuong and Xu [2017] if the instrumental variable for the treatment is binary, and (ii) I show that we can identify the value of the outcome variable for given mediator values in terms of the quantile function under assumptions of Chesher [2007] for the outcome variable. The second step requires the counterfactual distributions of outcome variable. Therefore, I establish the identification of the counterfactual distributions by using the observed distributions under specific support conditions for the outcome variable.

Social Interaction Model with Heterogenous Responsiveness [draft coming soon] [Slides]

In this paper, I extend a large network-based game model with incomplete information, specifically by incorporating heterogeneous individual payoff responsiveness. I introduce heterogeneity in responsiveness to explain why individuals in the same group, with similar backgrounds, may make different choices. I assume that each player’s private information follows a logistic distribution with a random scale parameter. This parameter can be interpreted as the responsiveness of each player, a trait that is formed after observing close friends. For instance, a person with higher responsiveness, which represents “positive experience”, is more likely to adopt an action than people with lower responsiveness. I further assume that this parameter follows a certain distribution (e.g., log-normal distribution) and its mean is a deterministic function of baseline characteristics. The deterministic part of the expected payoff is multiplied by the scale parameter. Therefore, the probability of each player choosing a specific action becomes a monotone function of payoff responsiveness. I show that the Bayesian equilibrium of this game corresponds to the heterogenous quantal response equilibrium of Rogers et al. [2009] and prove the existence and uniqueness of equilibrium.

Selected Work in Progress

Regression Discontinuity Design with an Indirect Effect (second-year paper)

Does National Police Accreditation Undermine Local Control? with Gordon Abner (Forthcoming in Policing: An International Journal )