Answers are grounded in this paper and cite their sources.
This discussion explores how Sherman–Morrison–Woodbury matrices improve OLS estimation in network models, providing theoretical insights.
We discuss and motivate the use of specific network matrices with a Sherman–Morrison–Woodbury inverse form in econometrics. Such matrices feature a parsimonious structure dependent on only two vectors that fits many situations where link intensity is the product of two size-based measures—a common reduced-form approximation in many gravity and industrial organization contexts. Matrices of Sherman–Morrison–Woodbury form permit special treatment which is fundamentally different from customary network matrices. In particular, the use of such matrices can lead to network models where the key slope parameters can be estimated easily by OLS from the structural network model. We illustrate their properties theoretically and in simulations.
No takes yet. Share an insight, caveat, or question.
Egger et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: