GLAMLE: inference for multiview network data in the presence of latent variables, with application to commodities trading

TL;DR

This paper introduces a novel latent variable model for multiview networks, enabling inference on economic trade data with unobservable factors, and demonstrates its effectiveness through simulations and real-world commodities trading analysis.

Contribution

The paper proposes the GGLLVM model and the GLAMLE inference method, providing a new approach to analyze multiview network data with latent variables.

Findings

GLAMLE achieves fast, accurate inference in simulations.

Application reveals import/export propensities among countries.

Model uncovers latent factors influencing trade relationships.

Abstract

The statistical analysis of import/export data is helpful to understand the mechanism that determines exchanges in an economic network. The probability of having a commercial relationship between two countries often depends on some unobservable (or not easy-to-measure) factors, like socio-economical conditions, political views, level of the infrastructures. To conduct inference on this type of data, we introduce a novel class of latent variable models for multiview networks, where a multivariate latent Gaussian variable determines the probabilistic behavior of the edges. We label our model the Graph Generalized Linear Latent Variable Model (GGLLVM) and we base our inference on the maximization of the Laplace-approximated likelihood. We call the resulting M-estimator the Graph Laplace-Approximated Maximum Likelihood Estimator (GLAMLE) and we study its statistical properties. Numerical experiments on simulated networks illustrate that the GLAMLE yields fast and accurate inference. A real data application to commodities trading in Central Europe countries unveils the import/export propensity that each node of the network has toward other nodes, along with additional information specific to each traded commodity.