Syllabus

Class	Date	Topics to Cover	Pre-class reading	Due
1	Tue Aug 25	Course Intro/Productivity/Computational Tools	Julia reference slides
2	Thu Aug 27	What is structural modeling?	Lewbel paper	Reading Quiz
3	Tue Sep 1	Structural modeling process	Keane YouTube talk	PS 1
4	Thu Sep 3	Random Utility Models & Logit	Train, Ch. 1-2, 3.1-3.3, 3.7-3.8	Reading Quiz
5	Tue Sep 8	Coding Day - go over PS 2		PS 2
6	Thu Sep 10	GEV	Train, 4.1-4.2	Reading Quiz
7	Tue Sep 15	Coding Day - go over PS 3		PS 3
8	Thu Sep 17	Mixed Logit, Finite mixture models, EM algorithm	Train, 6.1-6.3, Ch. 14	Reading Quiz
9	Tue Sep 22	Coding Day - go over PS 4		PS 4
10	Thu Sep 24	Dynamic choice models	Rust (1987)	Reading Quiz
11	Tue Sep 29	Coding Day - go over PS 5		PS 5
12	Thu Oct 1	Estimating dynamic models without solving	Hotz & Miller (1993); Arcidiacono & Miller (2011)	Reading Quiz
13	Tue Oct 6	Coding Day - go over PS 6		PS 6
14	Thu Oct 8	Simulated Method of Moments	???	Reading Quiz
15	Tue Oct 13	Coding Day - go over PS 7		PS 7
16	Thu Oct 15	Model Fit, Counterfactuals, Model validation	???	Reading Quiz
17	Tue Oct 20	Midterm Exam
18	Thu Oct 22	Causal Modeling: DAGs and Potential Outcomes	Mixtape
19	Tue Oct 27	Overview of Reduced-form Causal Inference Techniques	Mixtape	Reading Quiz
20	Thu Oct 29	Measurement Error & Factor Models	Cunha, Heckman & Schennach?	Reading Quiz
21	Tue Nov 3	Coding Day - go over PS 8		PS 8
22	Thu Nov 5	Regression and Partial identification	Krauth (2016), Oster (2019)	PS 9
23	Tue Nov 10	ATE, LATE, MTE
24	Thu Nov 12	ATE, LATE, MTE
25	Tue Nov 17	Treatment Effect Heterogeneity
26	Thu Nov 19	Treatment Effect Heterogeneity
27	Tue Nov 24	Machine Learning for Causal Modeling
---	Thu Nov 26	No class (Thanksgiving)
28	Tue Dec 1	Matrix Completion Methods
29	Thu Dec 3	Other Machine Learning Causal Methods
30	Tue Dec 8	Presentations		Presentation
31	Thu Dec 10	Presentations		Presentation, Referee Report
---	Mon Dec 14	Final Exam (Referee Report due)		Research Proposal

Intro to Structural Modeling

• What is structural modeling?
• Economic fundamentals: preferences (incl. strategic behavior), technology, outcomes, ...
• Policy-invariant parameters
• DGPs. Why DGPs? Because we need to resolve endogeneity.
• Counterfactuals
• Fit
• Validation
• Estimation methods (MLE, GMM, etc.)
• Estimation vs. calibration
• Preferences vs. Constraints vs. Preference heterogeneity
• References: Keane, Rust, Llull syllabus:
• "structural error" vs. "misspecification error"
  • simple example: returns to schooling (choice equation, outcome equation)
• Why aren't structural models used more often? Because it's exceedingly difficult to estimate a realistic model
• Examples of structural models from all fields of applied microeconomics

Programming tools

• GitHub
• Julia
• Optimization
• CLI
• Automation
• Web scraping / APIs?
• How to use OU's supercomputer (or cloud supercomputers)?

Discrete Choice Models

• Binary logit / probit
• Interpretation as latent index model
• Multinomial logit
• McFadden's logit
• DDC

Estimation methods

• EM algorithm
• CCPs
• MM/GMM/Simulated MM

Treatment Effects

• Potential outcomes model
• DAGs
• Experiment (ATE)
• IV (LATE)
• Regression as causality (unconfoundedness)
• RD
• Diff in Diff
• ATT, TUT, ITT, ATE, CATE, ...

Causality from observational data

• quantifying selection on unobservables (e.g. Altonji, Elder, Taber JPE paper)
• partial identification

Factor Models

• Measurement error
• Adjusting for latent factors that are measured with error (using multiple measurements)
• i.e. directly include the latent unobservable variable that your error-ridden measure attempts to capture

Machine Learning

• How ML helps get a better treatment effect estimate
• Schrimpf notes and repo
• examples of how to use it in methods they currently use (e.g. Diff in Diff, even study, IV, etc.)
• how unsupervised learning can help in detection of unobservable groups (and link back to EM algorithm and finite mixture models)

Other stuff?