Bagging the Network

TL;DR

This paper introduces a new estimation framework for dyadic network formation models with fixed effects, applicable to both TU and NTU links, overcoming non-concavity and bias issues.

Contribution

It develops a robust, asymptotically normal estimator combining method-of-moments, a Le Cam refinement, and jackknife bagging, valid under general link functions.

Findings

Estimator is unbiased, asymptotically normal, and efficient.

Simulation results confirm theoretical properties under TU and NTU.

Empirical application reveals different factors influence network links in Thai villages and Nyakatoke.

Abstract

We develop a unified estimation and inference framework for dyadic network formation with individual fixed effects, covering both transferable-utility (TU) and nontransferable-utility (NTU) links under general link functions. Under NTU, bilateral consent makes the fixed effects non-additive and the log-likelihood non-concave in the high-dimensional fixed effects, so differencing and profile-likelihood methods fail. We combine a joint method-of-moments initial estimator, a Le Cam one-step refinement, and a split-network jackknife bagging step that removes the incidental parameter bias without inflating variance. The resulting homophily estimator is asymptotically normal, unbiased, and attains the Cram\'er--Rao lower bound without requiring the log-likelihood to be concave in the fixed effects; we extend the theory to average partial effects and establish robustness to link-function misspecification. Simulations under both TU and NTU designs confirm these predictions. Applied to Thai village networks (TU), kinship and wealth differences both increase linking; in the Nyakatoke risk-sharing network (NTU), wealth differences have no significant effect, mirroring the two regimes' distinct logics.