Causal Inference &
Semi-Parametric Inference
Course Information (Fall 2022)
Course Name: Causal Inference and Semi-Parametric Inference (MGMT 69000 - 108; CRN 29206; 3 credit hours)
Time and Place: 9:30 am - 10:45 am TR, Aug 22, 2022 - Dec 10, 2022; RAWL 2077
Instrutor: Weibin Mo
Office Hours: 11:00 am - 12:00 pm TR, Aug 22, 2022 - Dec 10, 2022; KRAN 711
Email contact is preferred beyond the lecture time and office hours. Please initiate your email title with "[MGMT 69000CISI]". Emails are read in a regular basis and will be responded to within 24 hours in general.
Course Description
Causal inference has received increasing interests from both academic and industrial research across disciplines in recent decades. Social, biomedical and behavioral sciences need to leverage the causal inference tools when studying the effects of policy interventions. Such a problem is also known as program evaluation in econometrics. Statistics and computer science communities have seen rapid development in flexible machine learning techniques to approximate arbitrary relationships underlying the big and complex data. In order to drive effective evidence-based decision making, more and more researchers are turning their focuses from mining association to revealing causation under the proper, honest casual inference framework. In tech companies, thousands of randomized controlled experiments (A/B testing) are being operated to justify business impacts of new product launch. New challenges of confounding and interference (spill-over) are motivating more advanced experiment design. Moreover, companies are taking automated actions on and receiving feedbacks from their customers via huge data streams. It generates an unprecedented amount of causal inference problems for these companies to understand and optimize for counterfactual actions (what-ifs) from data.
In this course, we will cover selective frameworks and techniques in causal inference and discuss cutting-edge research topics. Theories and methodologies are the main focuses. In most causal models, minimal parametric assumptions or identification restrictions based on the targeted causal estimands are made, with the remaining distributional assumptions for the observed and unobserved variables left as unspecified/unrestricted. To understand the properties of causal estimates, the semi-parametric inference framework is also introduced for model-based, assumption-lean and model-free analyses.
Targeted audiences for this course are graduate students with certain quantitative research background. Graduate-level courses in mathematical statistics (STAT 52800), advanced econometrics (ECON 67100, 67200) or their equivalents can be helpful prerequisites for this course.
Course Schedule
The general agenda for this course is as follows.
Causal inference concepts and problems;
(Partially) Linear models, generalized methods of moment (GMM), just-and over-identified GMM, model misspecification;
Model-based and model-free statistical inference, consistency, double/multiple robustness, efficiency, influence function (IF), Neyman orthogonality to nuisance estimation from high-dimensional methods and general nonparametric (machine learning) methods, augmentation/de-bias;
Regression-based and propensity score-based methods, inverse-propensity score weighted estimate (IPWE), matching, augmented IPWE (AIPWE);
Compliance, choice, instrumental variables (IVs), local average treatment effect (LATE), two-stage least squares (2SLS), agent choice model for action/treatment, marginal treatment effect (MTE), policy-relevant treatment effect (PRTE);
Panel/Longitudinal/(Repeated) Cross-sectional data, difference in difference (DID), one/two-way fixed and random effect models, staggered adoption (event study design);
(If time permitted) Synthetic control; regression discontinuity design; (dynamic) discrete choice models.
The detailed schedule is provided below.
Learning Outcomes
Understand key concepts and research problems in causal inference;
Access the internal validity (against model misspecification and potential confounding) and external validity (against domain shift) of causal conclusions;
Understand various causal structural models, their model-based/parametric and model-free/nonparametric identifications and estimation methods;
Leverage flexible machine learning techniques and semi-parametric inference tools to make valid statistical inference (confidence interval) for the causal effect estimates;
Justify the optimality (efficiency) and robustness (consistency) of the causal effect estimates and their inference procedures.
Course References
Textbooks
Imbens, G., and Rubin, D. (2015). Causal Inference for Statistics, Social, and Biomedical Sciences: An Introduction. Cambridge: Cambridge University Press. DOI: 10.1017/CBO9781139025751.
Hernan, M. A., and Robins, J. M. (2020). Causal Inference: What If. Boca Raton: Chapman & Hall/CRC. URL: www.hsph.harvard.edu/miguel-hernan/causal-inference-book/.
Tsiatis, A. A. (2007). Semiparametric Theory and Missing Data. New York: Springer-Verlag. DOI: 10.1007/0-387-37345-4.
Tutorials, Surveys and Resourses
Abadie, A., and Matias, D. C. (2018). "Econometric Methods for Program Evaluation." Annual Review of Economics. 10(1):465-503. DOI: 10.1146/annurev-economics-080217-053402.
Guo, R., Cheng, L., Li, J., Hahn, P. R., and Liu, H. (2021). "A Survey of Learning Causality with Data: Problems and Methods." ACM Computing Surveys. 53(4):1-37. DOI: 10.1145/3397269. Github repository: github.com/rguo12/awesome-causality-algorithms/.
Huber, M. (2021). "An Introduction to Flexible Methods for Policy Evaluation." In N. Hashimzade and M. A. Thornton (eds) Handbook of Research Methods and Applications in Empirical Microeconomics (pp. 82-111). Edward Elgar Publishing. DOI: 10.4337/9781788976480.00010.
Imbens, G. W. (2022). Causality in Econometrics: Choice vs Chance. Econometrica, 90(6), 2541-2566. DOI: 10.3982/ECTA21204.
Kennedy, E. H. (2016). "Semiparametric Theory and Empirical Processes in Causal Inference". In: He, H., Wu, P., Chen, DG. (eds) Statistical Causal Inferences and Their Applications in Public Health Research. ICSA Book Series in Statistics. Springer, Cham. DOI: 10.1007/978-3-319-41259-7_8.
Kennedy, E. H. (2022+). "Semiparametric Doubly Robust Targeted Double Machine Learning: A Review." arXiv preprint. arXiv: 2203.06469.
Wager, S. (2020). "STATS 361: Causal Inference". URL: web.stanford.edu/~swager/stats361.pdf.
Workshop on Research Design for Causal Inference: www.law.northwestern.edu/research-faculty/events/conferences/causalinference/.