抄録
The aim of this paper is to analyze the factors of decision to stay in rural areas. The standard logit model has been often used for an analysis of rural recreation behavior. But I can not use this model, if the errors exhibit heteroscedasticity. Thus, I use the random parameter logit (RPL) model, which can be coped with heteroscedasticity.
To confirm whether the errors exhibit heteroscedasticity, I compare the RPL model with the logit model by likelihood ratio test. The result is that the RPL model is superior to logit model. Then, I hypothesize that the estimated parameter of the travel cost including staying cost is random. When the staying cost in Miyama Town is discounted by half, the probability of choice of staying rises to 9.1% and the social surplus is 0.10 billion yen.
