Advanced Techniques in Applied Economics
Spring 2026 · UPF Graphical models, causal discovery, latent variables, interference, and positive dependence.
Advanced Techniques in Applied Economics — Spring 2026
This course studies modern structured statistical methods for applied economics, with a focus on graphical models, causal discovery, latent-variable models, interference, and positive dependence.
Announcements
- May 1: A reminder that projects are due on May 4th, at 4am.
- Apr 28: The schedule for the presentation day is posted below.
- Apr 12: Please send to me your chosen topics for the final projects ASAP this week.
- Mar 21: The last two lectures are scheduled on May 4, 2:30-6:30pm.
- Mar 20: Slides for Lectures 1 and 2 are available.
- Mar 20: Suggested project topics are released.
Instructor
Piotr Zwiernik
Email: piotr.zwiernik@upf.edu
Office hours: 20.202 by appointment
Suggested reading
This is not a causal inference course but this area of data analysis relies heavily on similar ideas. I will try to make my lecture as self-contained as possible but some material that may be useful:
- R. Alexander, Telling stories with data.
- J. Peters, D. Janzing, & B. Schölkopf. Elements of causal inference: foundations and learning algorithms.
- V. Chernozhukov, C. Hansen, N. Kallus, M. Spindler, V. Syrgkanis, Applied Causal Inference Powered by ML and AI.
- S. Højsgaard, D. Edwards, S. Lauritzen. Graphical Models with R.
- The lecture slides and selected papers mentioned during the course
Lectures and timeline (tentative)
| Lecture | Topic | Slides |
|---|---|---|
| 1 | Conditional independence as structure | Lecture1 |
| 2 | Gaussian and non-paranormal graphical models | Lecture2 |
| 3 | DAGs, Markov equivalence, and interventions | Lecture3 |
| 4-5 | Causal discovery, linear structural equation models, and non-Gaussian identification | Lecture4 |
| 6 | Latent variable models — from trees to neural nets | Lecture5 |
| 7 | Unobserved confounding and adjustments | Lecture6 |
| 8 | Positive dependence and total positivity | Lecture7 |
| 9-10 | Presentations |
Projects
See the proposed project topics.
How to give short technical talks?
Presentation Schedule
Time: 14:35–18:30
Format: approximately 15 min presentation + 5 min discussion
Breaks: one 15 min break and one 10 min break
Note: Times are orientative; we will try to keep roughly to the schedule.
Schedule
14:35–14:55
Juan Carlos Cisneros, Erik Solé
Robustness of graphical model selection for economic network inference
14:55–15:15
Simone Alberto Distefano, Carlos Rubiano
Causal Discovery under Non-Gaussianity: LiNGAM and Mean Independence Extensions
15:15–15:35
Orhun Özel, Ece Taşan Özel
Network Effects of Oil Price Shocks on Inflation using a DAG-SVAR Framework
15:35–15:55
Michael Fehl, Francesco Zucca
Kalman filtering and extensions
15:55–16:10
Break
16:10–16:30
Kai Faulkner
Gaussian mixtures and hidden segmentation
16:30–16:50
Siyang Zhu
The Impact of Media Censorship: 1984 or Brave New World?
16:50–17:10
Anqi Liu, Megan Yeo
Network spillovers in applied economics
17:10–17:20
Break
17:20–17:40
Shuting Zhang, Lorenzo Marzano
Causal inference under interference
17:40–18:00
Pietro Fraccaroli, Gimelgo Xirinda
Does the economic impact of hurricanes in Florida propagate through channels beyond geographic proximity?
18:00–18:20
Zhekai Pang
Robust Score Matching
18:20–18:30
Closing / buffer