Hierarchical Bayesian Hedonic Regression Analysis of Japanese Rice Wine: Price is Right?
This research formulates a hedonic pricing model for Japanese rice wine, sake, via hierarchical Bayesian modeling, estimating it with a Markov chain Monte Carlo (MCMC) method. The data used in the estimation are obtained from Rakuten, the largest online shopping site in Japan. Flavor indicators, premium categories, rice breeds, and regional dummy variables are used as pricing factors. The Bayesian estimation of the model employs an ancillarity-sufficiency interweaving strategy to improve the sampling efficiency of MCMC. The estimation results indicate that Japanese consumers value sweeter sake more and the price reflects the cost of pre-processing rice only for the most luxurious category. No distinctive differences are identified among rice breeds or regions in the hedonic pricing model.