베이지안 추정법을 이용한 주택선택의 다항프로빗 모형 분석[Analysis of housing choice using multinomial probit model – Bayesian estimation]

We employ a multinomial probit model to understand the housing choice of Koreans, more specifically to understand the decision factors that affect Koreans’ housing decisions. The data used are from surveys conducted in July, 2006, and we use the Gibbs sampling technique in analyzing them. Apparently the housing decision is based on three dimensions–types of housing, size of places, and home ownership. A household may make its decision by simultaneously considering these three dimensions given its income and decision factors. Therefore, it would be ideal to categorize the alternatives based on these three dimensions. Due to the limited data, we treated the home ownership decision as if it were given outside the model or, in other words, as if a household made the decision before taking the type and size into account. From the results, we can see the followings: Firstly, the estimate of the coefficient of the alternative specific variables is negative and significant in all equations. Secondly, a household which decides to purchase housing is more likely to buy AS rather than HS. This is understandable considering Koreans’ inclination toward apartments.