Tomohiro Ando

Overview

Tomohiro Ando is a statistician and econometrician at Keio University (Graduate School of Business Administration, Yokohama). He works on Bayesian model selection criteria and Bayesian computation for multivariate regression models. He is known for the Bayesian Predictive Information Criterion (BPIC, 2007) and for the Direct Monte Carlo approach to Bayesian SUR analysis (with Zellner, 2010).

Key Contributions / Features

• Ando (2007): Bayesian Predictive Information Criterion (BPIC) — a model selection criterion for hierarchical and empirical Bayes models; approximable as $-2\int \log L(D|b,\Sigma) g(b,\Sigma|D) db\,d\Sigma + 2(\dim\{b\}+\dim\{\Sigma\})$ under standard asymptotics.
• Zellner and Ando (2010): Direct Monte Carlo (DMC) approach for Bayesian SUR inference exploiting the triangular reparameterization; eliminates MCMC convergence problems in high-dimensional settings. See Zellner and Ando (2010).

Related

• Seemingly Unrelated Regression
• Arnold Zellner
• Gibbs Sampler