A0206
Title: Dynamic network Poisson autoregression with an application to Covid-19 count data
Authors: Manabu Asai - Soka University (Japan) [presenting]
Amanda Chu - The Education University of Hong Kong (China)
Mike So - The Hong Kong University of Science and Technology (Hong Kong)
Abstract: There is a growing interest in accommodating network structure for panel data models. We consider dynamic network Poisson autoregressive (DN-PAR) models for panel count data, allowing time-varying network structure. We develop a Bayesian Markov chain Monte Carlo technique for estimating the DN-PAR model, and we conduct Monte Carlo experiments for examining the property of the posterior quantities to compare dynamic and constant network model. The Monte Carlo results indicate that the bias for the DN-PAR models is negligible, while the constant network model suffers from the bias when the true network is dynamic. We also suggest an approach for extracting the time-varying network from the data. The empirical results for the count data for the confirmed cases of Covid-19 in the United States indicate that the true and extracted dynamic network models outperform the constant network models regarding the deviance information criterion and out-of-sample forecasting.
