Bayesian Shrinkage in High-Dimensional VAR Models: A Comparative Study
Harrison Katz, Robert E. Weiss · 2025
Read Bayesian Shrinkage in High-Dimensional VAR Models: A Comparative Study by Harrison Katz, Robert E. Weiss (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-05489.txt.
Description
Harrison Katz and Robert Weiss wrote this econometrics paper, published in two thousand twenty-five, on forecasting many linked time series at once. Vector autoregression models do this well but drown in parameters when the series count and lag count grow.
The authors stage a fair race among shrinkage methods that pull wild estimates back toward zero. Three Bayesian priors, horseshoe, lasso, and normal, face two frequentist rivals across three simulation setups: a small overfit model, a sparse large one, and a mix of both traps. The horseshoe prior wins on accuracy of estimates and honest coverage of its intervals. Ridge forecasts points well but understates its own doubt. A run on Canadian macroeconomic data echoes the simulations. The paper guides analysts who must choose a prior before they see the data.
Catalog record
| CREATOR | Harrison Katz, Robert E. Weiss |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-05489