Fast Computation of the Deviance Information Criterion for Latent Variable Models

Joshua Chan and Angelia Grant (2016)
Computational Statistics and Data Analysis, 100, 847-859
[ Journal Version | Working Paper | Code ]

This code computes three variants of the DIC for three latent variable models: static factor model, TVP-VAR and semiparametric regression. We recommend that only the observed-data DIC should be used.

Evaluation of the observed-data likelihood is based on fast band matrix routines.

The algorithms are illustrated using three empirical applications that involve returns on stock portfolios, US macroeconomic time series and female body mass index and wages.