Researcher Nicolas Debarsy (CNRS) as a Speaker at the Spatial Warsaw Center Seminar

EDIT

The meeting with the guest speaker will take place on April 14, 2025, at 15:00. Please note that the event will be held exclusively online via the Zoom platform.

Link to the meeting: https://uw-edu-pl.zoom.us/j/93484071905?pwd=CPfHxRCS6aBh8L4voGWeU9OPunt5F7.1

[Meeting ID: 934 8407 1905

Passcode: 049400]

We warmly encourage you to participate!

---

Professor Nicolas Debarsy, Centre National de la Recherche Scientifique (CNRS) will present his study titled “Efficient estimation of regression models with spillovers: flexible parametric and semi-parametric approaches” (co-authors: Vincenzo Verardi and Catherine Vermandele).

We invite you to attend the meeting on April 9, 2025 at 15:00. It will be held in a hybrid format: in-person at Room F ( at the Faculty of Economic Sciences) and remotely via the Zoom platform.

Link to the meeting: https://uw-edu-pl.zoom.us/j/93484071905?pwd=CPfHxRCS6aBh8L4voGWeU9OPunt5F7.1

[Meeting ID: 934 8407 1905
Passcode: 049400]

We encourage you to review the abstract provided below.

---
In this presentation, we explore alternative methods for efficiently fitting regression models with spillover effects, which cannot typically be handled by ordinary least squares because of the simultaneity caused by interactions between individuals. Common estimation methods, such as two-stage least squares, generalized method of moments, or (quasi)maximum likelihood, are widely used but may not be the most efficient except in some specific cases. To address this, we propose two approaches, both based on Le Cam’s local asymptotic normality (LAN) theory but differing in how they handle the error distribution. The first is a semiparametrically efficient estimator based on residual ranks and signs that requires only strong unimodality of the error distribution. The second is a parametric estimator designed to account for skewness and heavy tails, using flexible distributions such as Tukey’s g-and-h and Pewsey and Jones’ sinh-arcsinh. Both estimators perform well in Monte Carlo simulations, frequently outperforming commonly used estimators. Finally, we demonstrate the practical implications of our approach with a trade regression from Behrens et al. (2012), illustrating how empirical results can change  significantly when Gaussian assumptions are relaxed.