Build credible causal answers in high-dimensional, AI-era data.
This course develops a modern toolkit for applied econometric work: transparent identification, principled inference, and computationally robust methods that remain reliable at scale. You will combine economic reasoning with machine learning and reproducible workflows.
Explore course map
Time
Thursday · 9:30am–12:15pm
Venue
ELB 304
Assessment
Midterm 50% · Final 50%
Prerequisite
ECON5120
Course Summary
What you will learn
- Linear and nonlinear estimation with likelihood and moment-based methods.
- Asymptotic reasoning and when it fails in nonstandard settings.
- Regularization + ML for prediction and causal inference without overfitting.
- Computational workflows using optimization, simulation, and bootstrap.
Course focus
Econ5150 is the sequel to ECON5120. It emphasizes credible causal questions, clear identification assumptions, and estimation methods that remain robust in high-dimensional and computationally intensive environments. The course integrates classic econometric theory with modern ML-enabled inference.
Core Topics
Linear Models
- Causal inference and linear projections
- OLS large-sample theory
- Shrinkage, ridge, and bias-variance tradeoffs
- Overfitting and out-of-sample prediction
Nonlinear Models
- Maximum likelihood and limited dependent variables
- Extreme estimators and GMM
- Quantile regression and nonparametrics
- Numerical optimization + ML for causal inference
Special Topics
- Time series
- Factor models
- Panel data methods
Current Reading Map
- Week 1: potential outcomes, ATE/CATE, propensity score, continuous treatments.
- Week 2: regression-based causal inference, FWL theorem, OVB, ridgeless regression.
- Week 3: finite-sample OLS properties, prediction vs. inference, modes of convergence.
- Week 4: LLN/CLT, asymptotics for OLS, ridge and bias-variance tradeoff.
- Week 5: lasso, KKT, restricted eigenvalues, Neyman orthogonality.
Textbooks
Applied Causal Inference Powered by ML and AI · CHKSS (2025)
Hansen · Econometrics (2021)
Goodfellow/Bengio/Courville · Deep Learning (Recommended)
Hastie/Tibshirani/Friedman · Elements of Statistical Learning (Recommended)