준모수방법에 의한 주택가격지수 추정에 관한 연구 = A Study on the Semiparametric Estimation of the Housing Price Indexes

다국어 초록 (Multilingual Abstract)

A semiparametric estimator is used to analyze the hedonic housing price indexes in a sample of 1,284 sales prices of apartments in the southern Seoul area. The hedonic price functions are expected to be nonlinear, but choosing the appropriate functional form is particularly difficult for location variables and time trends because it tends to be unique to the place and time. A semiparametric estimation offers important advantages for the hedonic housing price function estimation. It combines the benefits of the parametric and nonparametric estimation. The parametric portion of the model includes standard housing characteristics such as dwelling unit size, the size of the community, age of building, types of heating system and fuel material, school districts, among others. Distances to the CBD, several subcenters, and the nearest subway station are estimated nonparametrically. The sales year is modeled nonparametrically, as well. The coefficients have the expected signs and are stable across parametric and semiparametric specifications. The semiparametric estimation has higher coefficient of determination than two parametric OLS estimations. Hausman (1978) tests for the specification of the hedonic price functional form prefer to the semiparametric estimation. At last, the housing price indexes for four subway station regions are estimated and compared.