Affiliation:
Brunel University of London
Contact:
Email: vasilis.sarafidis [the usual symbol] brunel.ac.uk
Research Field:
Econometric Theory and Practice, Panel Data Analysis, Spatial Models and Networks
What's new?
March 2026: I’m happy to share a new working paper “Network Effects in Corporate Emissions: Evidence from a Data-Dependent Spatial Panel Model” (with Stylianos Asimakopoulos, George Kapetanios and Alexia Ventouri). We study spillover effects in corporate toxic emissions using a heterogeneous panel network of U.S. industrial facilities during 2000–2023. Rather than imposing a network structure a priori, we uncover an unobserved web of influence directly from the data using recent advances in high-dimensional network econometrics. Indirect effects transmitted through the estimated network account for around 28% of the total impact of key firm balance-sheet characteristics. By contrast, distance-based networks generate no statistically discernible spillovers, while a priori firmor industry-based networks substantially overstate within-group spillins relative to the data-driven network.
February 2026: I’m happy to share a new working paper “Model Selection in High-Dimensional Linear Regression using Boosting with Multiple Testing” (with George Kapetanios and Alexia Ventouri). The paper addresses a simple but important question: how can we reliably identify the truly relevant variables when data are high-dimensional, highly correlated, and potentially driven by latent factors? In this paper we propose Boosting with Multiple Testing (BMT), a new approach that combines stepwise learning with rigorous statistical testing to recover true signals while avoiding spurious ones. Across simulations and empirical applications, the method produces sparse, interpretable models and strong predictive performance, even in challenging environments with multicollinearity, latent structure, and persistence. The underlying idea is broadly applicable across many empirical settings where reliable variable selection is essential.
January 2026: I am pleased to share that I will teach a course titled Network and Spatial Panel Data Analysis using Stata at the 2026 Thessaloniki Econometrics Summer School (Th.E.S.S.), hosted by the University of Macedonia in Thessaloniki, Greece, on July 13 to 17, 2026. The course focuses on modelling spatial interactions and network effects in panel data, with empirical applications in economics and finance. Practical lab sessions will be led by Professor Theodore Panagiotidis. More information and applications: https://thess.uom.gr/
December 2025: It was a real honour to take part in a tribute session for Hashem Pesaran at CFE 2025, hosted at Birkbeck, University of London, and contribute a paper, especially with Hashem present. His landmark work across time series, panel data, macroeconometric modelling and forecasting has opened new literatures, created space for hundreds of others to follow, and shaped my own thinking through his writings and later conversations. This is new work on model selection in high-dimensional regression, joint with George Kapetanios and Alexia Ventouri. Slides from my presentation are here.
November 2025: I am grateful to Enrique Pinzon, Stata Corporation and everyone who joined the 2025 Stata Economics Virtual Symposium. My talk on “Spatial dynamic panel data models with interactive effects,” drew more than 200 attendees and generated a lively online discussion at the end. My presentation slides are available in two different formats: static, and dynamic.