This paper examined the forecasting peformance of the disaggregated data with spatial dependency and applied it to forecasting electricity demand in Japan. We compared the performance of the spatial autoregressive ARMA (SAR-ARMA) model with that of vector autoregressive (VAR) model from a Bayesian point of view. Compared the models by the log marginal likelihood and log predictive density, VAR(1) model had the better performances than SAR-ARMA(1,1) model. In the case of electricity demand in Japan, we can conclude that the VAR model with the contemporaneous aggregation had the better forecasting performance than the SAR-ARMA model.