Microeconometrics and Policy Evaluation Summer School : Program content

Discover the content of the PSE Summer School’s Microeconometrics and Policy Evaluation program.

An in-depth program content

The Microeconometrics and Policy Evaluation program 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 or R, 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.

Course listing:

  • Methods of Policy Evaluation
  • Machine Learning for policy evaluation
  • Practice classes

As an option, students can volunteer to present in a small workshop environment (3 students per session, in presence of one instructor). Selected students will be given a 20-min slot to present their project (at an advanced or intermediate stage), and get feedback about their empirical strategy. Slots are limited (15 students max, first-come first-served).

Course details

by 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.

Structure

  • Introduction: setting and real world experiments
  • Comparing similar individuals
    • Regression models
    • Matching models
    • Regression discontinuity
  • Simulating unobserved outcomes
    • Instrumental variables
    • Selection models
  • Intertemporal comparisons
    • Before-After
    • Difference-in-Differences and extensions
    • Synthetic controls
  • Summarizing methods

Selected key references

  • Graduate-level texts
    • Cameron A. C. & Trivedi P.K., 2005, Microeconometrics: Methods and Applications, Cambridge University Press.
    • Greene W.H., 1993, Econometric Analysis, Prentice Hall.
  • Other higher-level texts
    • Davidson R. & MacKinnon J.G., 1993, Estimation and Inference in Econometrics, Oxford University Press.
    • Wooldridge J.M., 2001, Econometric Analysis of Cross Section and Panel Data, MIT Press.

by 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.

Structure

  • Introduction to machine learning
    • Cross-validation
    • Bias-variance tradeoff
  • LASSO and (control) variable selection
    • LASSO and related regression based estimators
    • Lack of consistent variable selection
    • Treatment effect estimation in the presence of many control variables
  • Tree-based regression methods
    • Regression trees
    • Random forests and related machine learning methods
  • Estimating treatment effect heterogeneity using tree-based methods
    • Causal trees
    • Causal forests
  • Estimating treatment effect heterogeneity using “generic” machine learning methods

Main references

  • Belloni A., Chernozhukov V. & Hansen C., 2014, “High-dimensional methods and inference on structural and treatment effects”, Journal of Economic Perspectives, 28(2), pp 29-50.
  • Davis J. & Heller S. B., 2017, “Using Causal Forests to Predict Treatment Heterogeneity: An Application to Summer Jobs”, American Economic Review, 107(5), pp 546-50.
  • James G. et al., 2013, An introduction to statistical learning, Springer.

Additional/background references

  • Mullainathan S. & Spiess J., 2017, “Machine learning: an applied econometric approach”, Journal of Economic Perspectives, 31(2), pp 87-106.
  • Varian H. R., 2014, “Big data: New tricks for econometrics”, Journal of Economic Perspectives, 28(2), pp 3-28.

Supplementary/Technical references

  • Athey S. & Imbens G., 2016, “Recursive partitioning for heterogeneous causal effects”, Proceedings of the National Academy of Sciences, 113(27), pp 7353-7360.
  • Belloni A., Chernozhukov V. & Hansen C., 2014, “Inference on treatment effects after selection among high-dimensional controls”, The Review of Economic Studies, 81(2), pp 608-650.
  • Chernozhukov V. et al., 2018, “Double/debiased machine learning for treatment and structural parameters”, The Econometrics Journal, 21(1), C1-C68.
  • Chernozhukov V. et al., 2018, Generic machine learning inference on heterogenous treatment effects in randomized experiments, Working paper.
  • Hastie T., Tibshirani R. & Friedman J., 2009, The elements of statistical learning, Springer.
  • Wager S. & Athey S., 2018, “Estimation and Inference of Heterogeneous Treatment Effects using Random Forests”, Journal of the American Statistical Association, 113(523), pp 1228-1242.

by Martin Mugnier and Liam Wren-Lewis

All students are expected to have basic knowledge of the STATA software. Students will attend 1.5 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 observational data from various sources (e.g. cross sectional databases collecting information on labor market and health-related individual outcomes, as well as firm-level data). This course is intended for intermediate students, it uses the Stata software and takes place in the computer room of PSE (equipped with computers with Stata installed).
  • The advanced group will attend the course Machine learning in R (professor Martin Mugnier). This course is intended for students who don’t need guidance to use built-in commands, and it covers (broadly) the machine learning material taught by Philipp Ketz. This course is intended for advanced students, it uses the R software and takes place in a standard classroom (participants are expected to bring their own laptop with R).