모집단 가중치 적용여부에 따른 주택수요 추정결과 비교 및 시사점

다국어 초록 (Multilingual Abstract)

This study examines how the application of population weights affects the estimation of housing demand in Korea, employing the household-size–based Mankiw–Weil(M-W) model, a widely used framework in domestic housing demand analysis. Using the Korea Housing Survey data from 2017–2023, we compare weighted and unweighted estimations of the modified M-W model to assess how population weighting alters regression coefficients across periods and regions, and how it influences long-term housing demand. The analysis distinguishes between the Seoul Capital Area and non-capital regions to evaluate regional heterogeneity. Our findings indicate that the Housing Survey tends to under-sample one-person households and over-sample two-person households compared with the Population and Housing Census, suggesting that population weights may be necessary to mitigate sampling bias in estimating the M-W model. Estimation results from the modified model reveal statistically significant differences in regression coefficients between weighted and unweighted specifications in both the capital and non-capital regions, accompanied by substantial discrepancies in projected long-term housing demand. Notably, the omission of population weights appears to lead to an upward bias in long-term demand estimates for non-capital regions relative to the capital region. These results underscore the importance of applying population weights when estimating housing demand from sample survey data. In line with previous evidence, we recommend reporting and comparing both weighted and unweighted estimation results to enhance the accuracy and robustness of model inferences.