Full course description
Overview
The aim of this bootcamp is to help participants become better consumers and producers of empirical work. The primary content of the bootcamp deals with problems in causal inference, the methods that address them, and other issues in empirical research. The pedagogical approach relies heavily on simple econometric theory and Monte Carlo methods, emphasizing the key distinction between the data generating process and the estimated empirical model. The goal is not to provide a fully encompassing empirical methods course—which would be impossible in such a short period of time. Rather, the bootcamp is designed to introduce the central concepts, develop an intuitive grasp of key issues and ideas, and provide practical tools and the most up-to-date resources for participants.
Bootcamp Structure
The bootcamp includes carefully designed assignments, discussions, and lecture to bring out core ideas. The bootcamp is functionally divided into four parts:
- Pre-Bootcamp Preparations. To maximize the value of the bootcamp, the pre-bootcamp preparations are designed to ensure that participants have a reasonable understanding of Directed Acyclical Graphs and Monte Carlo Simulations. This will include some short readings and a simulation to run.
- What goes wrong in causal inference. In this section we examine the most common reasons for biased or inconsistent empirical estimates. These include bad controls, omitted variables, measurement error, simultaneity, and sample selection. Alongside simple economic theory, our goal is emphasize the distinction between the true data generating process and the empirical models that we choose to estimate. At the end of this section, participants should have a solid and practical understanding of what causes bias in empirical work.
- The causal inference toolkit. In this section we examine the most common methods for dealing with the problems introduced in the prior sections. This includes double machine learning, instrumental variables, matching, regression discontinuity, difference-in-differences. For each of these methods, we examine the basic econometric theory alongside Monte Carlo methods to understand when these methods will and will not resolve various problems to causal inference. Inevitably, this section of the course will not cover as much depth as one might like—for those interested in further reading, we will provide references describing the state of the art on these methods.
- Other issues in applied empirical work. Throughout the course we will integrate discussion of several of other important issues in empirical work. These include handling of standard errors, scientific apophenia, graphical techniques, interpretation of coefficients and interactions in non-linear models, and other modern techniques related to partial identification. The main goal of this part of the bootcamp is to ensure that participants are at least familiar with these issues and tools, so-as to improve their ability to consume and produce empirical work.
Prerequisites
Familiarity with Stata or R will be important for this class, though all code will be provided to you for class assignments.
Resources
The following books will be references for methods throughout the class. Both are available for free online.
- The Effect: An introduction to research design and causality, Nick Huntington-Klein
- Causal Inference: The Mixtape, Scott Cunningham
When will the course be held?
This course is scheduled for five days in June:
- Monday, June 1 from 9:30 AM to 2:30 PM
- Tuesday, June 2 from 9:30 AM to 2:30 PM
- Wednesday, June 3 from 9:30 AM to 2:30 PM
- Thursday, June 4 from 9:30 AM to 2:30 PM
- Friday, June 5 from 9:30 AM to 2:30 PM
The last day to register for this bootcamp is May 27.
Tentative Schedule
- Day 1: Monte Carlo Simulations, Potential Outcomes, and DAGs. Discuss omitted variable bias and data generating processes.
- Day 2: Bad controls, simultaneity, measurement error, and sample selection.
- Day 3: Functional Form, Double Machine Learning, Matching, Weighting
- Day 4: Instrumental variables and Regression Discontinuity.
- Day 5: Panel Models and Difference-in-Differences
Who should take this course?
This course is intended for PhD students seeking to improve their fundamental understanding causal inference.
Who are the instructors?
Brent Goldfarb is a subject matter expert on entrepreneurship, strategy, and empirical methods. He is currently the Dean’s Professor of Entrepreneurship and the Academic Director of the Dingman Center in the Robert H. Smith School of Business at the University of Maryland, College Park.
Evan Starr is a subject matter expert on economics, strategic management, and empirical methods. He is currently a Professor in the Robert H. Smith School of Business at the University of Maryland, College Park.
David Waguespack is Associate Professor of Management & Organization at the Robert H. Smith School of Business at the University of Maryland. Dr. Waguespack received his PhD in Political Science, focusing on environmental politics and science and technology policy.
Return to Smith Website | Contact us: rhsmith-execed@umd.edu