Seminars & Events
Title: TBA
Abstract: TBA
An applied econometrics hackathon where the responsible and smart use of AI meets academic research rigor.
Teams from across geographical Europe compete first in a remote 48-hour preliminary round. Then, ten finalist teams travel to Milan for an in-person final at Bocconi University, where they complete a new research challenge and get to present their findings to Bocconi faculty, international guests, sponsors from the economic consulting sector and fellow students.
Title: Quantitative and qualitative evidence from the use of AI tools in Bocconi exams
Abstract: TBA
Title: TBA
Abstract: TBA
Title: Can Technology Overcome Small Markets? Machine Translation and Access to Knowledge across Languages
Abstract: TBA
Title: Logical similarities in parliamentary speeches over EU treaty ratifications
Abstract: How did members of national parliaments of EU member states debate EU treaty ratifications? Which reasons did they give publicly for how they voted? Are these reasons and their logical underpinnings more similar within countries or within party groups in the European parliament? How did the answers to these questions evolve over time and how do they depend on the specific and their contents? The seminar will discuss how we approach these questions and how we plan to measure the variables of interest, focusing on the logical structure of the speeches. The research project is at a very early stage and no results will be presented. This is a joint project with Daniel Gros and Giorgio Presidente.
Title: Beyond Correlation, Beyond Validation, Beyond the Point Estimate: Recovering Expert-Calibrated LLM-Imputed Party Preferences from Manifestos
Abstract: TBA
Title: 16th century Florentine banks' ledgers: digitizing handwritten text using Transkribus
Abstract: TBA
Title: Randomness in Large Language Models: What Researchers Need to Know (and Report)
Event summary: Large language models are increasingly used to generate data for research. Their outputs can vary across repeated requests even when the prompt and settings are unchanged, and setting the temperature to zero removes only part of that variation. The talk looks at where the variation comes from, how it affects downstream regressions, and proposes a reporting standard for articles and replication packages.
Joint with Guillaume Coqueret, Joan Llull, Christophe Pérignon, Christoph Scheuch and Lars Vilhuber.