My research lies at the intersection of Bayesian econometrics and empirical macroeconomics, with a primary focus on vector autoregressive (VAR) models. I study how modern Bayesian statistical techniques can be used to improve forecasting, structural analysis, and joint density estimation in high-dimensional macroeconomic systems.
My recent research examines the use of probabilistic network models and structured sparsity to characterize economic linkages in multivariate time series, methods to robustify simple conjugate Bayesian vector autoregressions against model misspecification and modeling techniques for micro- and macroeconomic interactions.
In my ongoing work, I examine functional vector autoregressions, local projections, Gaussian processes, and Dirichlet process mixtures, among other things.
Work in progress
- Tobias Scheckel. Microeconomic Heterogeneity and Macroeconomic Shocks -- A Nonlinear State-Space Approach (Draft coming soon)
Working papers
- Florian Huber, Massimiliano Marcellino and Tobias Scheckel. Coarsened Bayesian VARs -- Correcting BVARs for Incorrect Specification (2025). R&R: Journal of Business and Economic Statistics
- Florian Huber, Gary Koop, Massimiliano Marcellino and Tobias Scheckel. Bayesian modelling of VAR precision matrices using stochastic block networks (2024). R&R: Journal of Applied Econometrics