Conference Agenda
Overview and details of the sessions of this conference.
Please select a date to show only sessions at that day. Please select a single session for detailed view (with abstracts and downloads if available).
Activate "Show Presentations" and enter your name in the search field in order to find your function (s), like presenter, discussant, chair.
Some information on the session logistics:
If not stated otherwise, the discussant is the following speaker, with the first speaker being the discussant of the last paper. The last speaker of each session is the session chair. (Exception: invited sessions)
Presenters should speak for no more than 20 minutes, and discussants should limit their remarks to no more than 5 minutes. The remaining time should be reserved for audience questions and the presenter’s responses. We suggest following these guidelines also in the (less common) 3-paper sessions in a 2-hour slot, to allow participants to move between sessions. Discussants are encouraged to avoid summarizing the paper. By focusing on a few questions and comments, the discussants can help start a broader discussion with the audience.
Only registered participants can attend this conference. Further information available on the congress website https://www.iseg.ulisboa.pt/en/event/iipf/ .
Venue address: ISEG - Lisbon School of Economics & Management, R. Francesinhas 21, 1200-675 Lisboa, Portugal
Please note that all times are shown in the time zone of the conference. The current conference time is: 17th Sept 2026, 11:39:33am WEST
|
Daily Overview |
| Session | |
|
E12: Third-Party Reporting, Audit Targeting, and Non-Filing Location: Room 113 (Francesinhas 1) | |
    | |
Optimal Audit Targeting with Machine Learning: Evidence from Pakistan 1: Tulane University, United States of America; 2: Federal Board of Revenue, Pakistan This paper develops empirically implementable algorithms for optimal audit targeting with machine learning. We derive a sufficient statistic-based targeting algorithm that depends on three individualized causal effects: the immediate revenue recovered from an audit, the causal effect of an audit on long-run tax revenue, and the marginal administrative cost of an audit. We show that these effects can be estimated with a variety of machine learners including causal forests, LASSO, gradient boosted trees, and neural networks using the universe of Pakistani income tax returns, exploiting years in which audits were assigned completely at random. We implement our targeting algorithms in out-of-bag years, comparing them to the real-world policy when audits were targeted. We show that the real-world audit program in Pakistan lost almost 173,000 Rs ($1, 700) in net revenue per-audit, while our optimal policy generates 285,000 Rs ($2, 800) in expected net revenue per-audit.
| |
