Research

Papers:

Sparse Models and Methods with Grouped Patterns of Heterogeneity with an Application to Consumer Demand

This paper introduces post-Lasso methods to time-varying grouped patterns of heterogeneity in linear panel data models with heterogeneous coefficients. Group membership is left unrestricted and the model is assumed to be approximately sparse. I estimate the parameters of the model using a “grouped fixed-effects” estimator that minimizes a post-Lasso least-squares criterion which can handle conditions where the number of variables is very large. I provide conditions under which the estimator is consistent as both dimensions of the panel tend to infinity and provide inference methods. I apply this method to estimate demand using US consumer data.

Nonlinear Panel Data Random Coefficient Models: an Application to Consumer Demand (with Stefan Hoderlein, BC, and Alexander Meister, Rostock)

This paper shows that the distribution of marginal effects of a very general class of structural models is nonparametrically identified allowing for arbitrary dependence between time invariant unobservable and the covariate of interest, with as little as two observations are available for the individuals. Based on this insight, we construct a semiparametric sample counterparts estimator, and apply it to US consumer data.

Binary Choice Difference-in-Difference Model with Heterogeneous Treatment Effects (with Stefan Hoderlein, BC, and Solvejg Wewel, Rostock)

This paper extends the familiar Difference-in-Differences Model to binary choice outcome variables. The random coefficient formulation of the proposed model additionally allows for heterogeneous treatment effects. The main result is the identification of the average treatment effect on the treated (ATT). We present several extensions including identification of the joint distribution of the actual and counterfactual latent outcome variable in the treatment group and the inclusion of covariates. We suggest an estimator for the ATT and evaluate the properties with Monte Carlo simulations. We apply this estimator to estimate the effects on a Sugar Sweetened Beverage tax and find that it led to the fraction of households purchasing soda to decrease by about 0.12.

Graduate Research Assistant Work

Mentor: Dr. Stefan Hoderlein

Professor, Department of Economics

Boston College

Worked on a paper focused on estimating marginal effects in nonlinear panel data with heterogeneity. Coded the project in R and ran Monte-Carlo simulations.

Did additional work on right-censored regression models and different machine learning techniques.

Undergraduate Research

• Exact Nonparametric Inference for a Binary Endogenous Regressor

[June 2015 - current]

Mentor: Dr. Brigham Frandsen

Assistant Professor, Department of Economics

Brigham Young University

Description:

• This paper describes a randomization-based estimation and inference procedure for the distribution of potential outcomes with a binary treatment and instrument. The method imposes no parametric model for the treatment effect, and remains valid for small sample sizes or a weak instrument when conventional large-sample methods break down.
• I coded this program in python using parallel computing methods and analyzing other challenges with drawing a random sample from a multivariate hyper-geometric distribution.
• This method is illustrated using data from a randomized trial of college student incentives and services. We are attempting to create a database of all possible values for our method.
• Testing Censoring Point Independence

[January 2015 - April 2015]

Mentor: Dr. Brigham Frandsen

Assistant Professor, Department of Economics

Brigham Young University

Description:

• Examined every paper in major economic journals in the last five years in which censoring point independence was assumed. Gathered the data used for each of these papers and used Stata to test censoring point independence.
• Treatment Effects With Censoring and Endogeneity

[September 2015 - December 2015]

Mentor: Dr. Brigham Frandsen

Assistant Professor, Department of Economics

Brigham Young University

Description:

• Attempted to translate the code that was written in Stata and Mata, and rewrite it in R. Thus, it could be used as a command line once the package was installed.
• Party Bias in Union Representation Elections: Testing for Manipulation in the Regression Discontinuity Design When the Running Variable is Discrete

[April 2015 - September 2015]

Mentor: Dr. Brigham Frandsen

Assistant Professor, Department of Economics

Brigham Young University

Description:

• Gathered data from NLRB offices across the country and consolidated it into one data set. I used this data set to analyze bias and look at elections when Republicans are in charge of the NLRB and when Democrats are in charge of the NLRB to examine possible bias in worker union elections.
• Below are histograms of the results of our research.