The Microeconometrics and Policy Evaluation programme presents recent developments in the microeconomic analysis of impact evaluation, with courses taught by experts in their fields. Providing a credible estimation of a causal effect has become a standard in economic analysis, both in research papers and policy reports. But it is also equally important to integrate the estimated effects into economic models, in order to improve the design of policies. The programme therefore proposes a comprehensive approach of policy analysis.
The course “Methods of policy evaluation” introduces the main methods currently used for program evaluation, while the course “Machine learning for policy evaluation” presents recent advances in machine learning techniques for policy analysis. Students will also attend practice classes in STATA, according to their level of proficiency (intermediate vs. advanced).
The courses are designed for graduate students wishing to gain additional skills in applied microeconometrics for policy evaluation, as well as for researchers and professionals in public and private institutions who work on impact evaluation.
Workshop: present your paper
Participants will have the opportunity to submit a paper to be presented within this program, in a small workshop environment (3 students per session, in presence of one instructor). Selected students will be given a 20-minutes slot to present their project (at an advanced or intermediate stage), and receive feedback about their empirical strategy. Slots are limited.
Methods of Policy Evaluation (David Margolis)
The econometric techniques for evaluating the impact of policies using micro-level data have evolved substantially over time, and understanding which technique is most appropriate in any particular situation is crucial to providing credible estimates of the effects of policy. This course will explore a wide variety of techniques, focusing not only on the estimators themselves and why they work, but also on the types of data they require and the assumptions that must hold for the estimates to be valid.
1. Introduction: setting and real world experiments
2. Comparing similar individuals
a. Regression models
b. Matching models
c. Regression discontinuity
3. Simulating unobserved outcomes
a. Instrumental variables
b. Selection models
4. Intertemporal comparisons
b. Difference-in-Differences and extensions
c. Synthetic controls
5. Summarizing methods
Selected key references
W.H. Greene, Econometric Analysis, Prentice Hall.
A.C. Cameron & P.K. Trivedi, “Microeconometrics: Methods and Applications”, Cambridge University Press.
Other higher-level texts
J.M. Wooldridge, “Econometric Analysis of Cross Section and Panel Data”, MIT Press.
R. Davidson & J.G. MacKinnon, “Estimation and Inference in Econometrics”, Oxford University Press.
Machine Learning for program evaluation (Philipp Ketz)
The objective of this course is to understand how machine learning methods can be useful for program evaluation and what the difficulties and challenges are. We start with a short introduction to machine learning, highlighting its usefulness for prediction. Then, we discuss two recent adaptations of machine learning in economics, where the focus lies with causal inference. They concern the consistent estimation of the average treatment effect in the presence of many control variables and the analysis of heterogeneous treatment effects in the context of randomized controlled trials (RCTs), respectively.
1. Introduction to machine learning
a. Bias-variance tradeoff
c. Overview of popular methods
2. Estimating average treatment effects in the presence of many controls
a. Problem with “naive” estimators
b. “Double” machine learning as solution
3. Estimating heterogeneous treatment effects in RCTs
a. Tree-based methods
b. Generic methods
Selected key references
Belloni, A., V. Chernozhukov, and C. Hansen, (2014), “High-dimensional methods and inference on structural and treatment effects”. Journal of Economic Perspectives 28(2), 29–50.
Davis, J. and S. B. Heller, (2017), “Using Causal Forests to Predict Treatment Heterogeneity: An Application to Summer Jobs”. American Economic Review 107 (5), 546–50.
James, G., D. Witten, T. Hastie, and R. Tibshirani, (2013), “An introduction to statistical learning. Springer”.
Mullainathan, S. and J. Spiess, (2017), “Machine learning: an applied econometric approach”. Journal of Economic Perspectives 31(2), 87–106.
Varian, H. R, (2014), “Big data: New tricks for econometrics”. Journal of Economic Perspectives 28(2), 3–28.
Athey, S. and G. Imbens, (2016), “Recursive partitioning for heterogeneous causal effects”. Proceedings of the National Academy of Sciences 113 (27), 7353–7360.
Belloni, A., V. Chernozhukov, and C. Hansen (2014b), “Inference on treatment effects after selection among high-dimensional controls”. The Review of Economic Studies 81(2), 608–650.
Chernozhukov, V., D. Chetverikov, M. Demirer, E. Duflo, C. Hansen, W. Newey, and J. Robins, (2018), “Double/debiased machine learning for treatment and structural parameters”. The Econometrics Journal 21(1), C1–C68.
Chernozhukov, V., M. Demirer, E. Duflo, and I. Fernandez-Val, (2018), “Generic machine learning inference on heterogenous treatment effects in randomized experiments”. Working paper.
Hastie, T., R. Tibshirani, and J. Friedman, (2009), “The elements of statistical learning”. Springer.
Wager, S. and S. Athey, (2018), “Estimation and Inference of Heterogeneous Treatment Effects using Random Forests”. Journal of the American Statistical Association 113 (523), 1228–1242.
Practice classes (Margherita Comola and Liam Wren-Lewis)
All students are expected to have basic knowledge of the STATA software. Students will attend 2 hours a day of practice classes, according to their level:
The intermediate group will attend the course Policy Evaluation in Stata (professor Liam Wren-Lewis) which illustrates the estimation methods presented in the theoretical class taught by David Margolis using data from seminal papers in economics that use a variety of techniques (e.g. an RCT, RDD, IV, etc).
The advanced group will attend the course Advanced Topics in Stata (professor Margherita Comola). This course is intended for students who don’t need guidance to use built-in Stata commands, and it provides an introduction to Stata programming tools. Topics discussed in class include: matrix-based Mata sintax, log-likelihood functions, coverage probabilities, matching algorithms. Participants are encouraged to bring their own laptop with Stata.
Contents – Microeconometrics and Policy Evaluation