무조건부 분위회귀를 이용한 아파트 가격결정요인 분석 - 양재천 인근 아파트를 중심으로 -

Abstract

In estimating the effect of housing characteristics on housing prices, the ordinary least squares method has a limitation in that the result shows only a part of the influence on the conditional mean price, so the conditional quantile regression has been used as an alternative to measure the effect by price levels. However, the conditional quantile regression has a problem in which different price groups are connected to each other. It is recognized that such a problem arises remarkably when the heteroskedasticity occurs or when the shape between conditional distribution and the unconditional distribution are different. In order to get more reasonable effects by price levels, we apply the unconditional quantile regression based on the recenterd influence function for apartments near Yangjae Stream in Seoul. The result shows that the price effect is more sensitive to surrounding amenity factors like Yangjae Stream View, as it goes to expensive houses. This result is more consistent with common sense than the result of other methods of analysis. It is also confirmed that the utility of traffic types like the accessibility to subway stations may vary depending on the level of housing prices.