住房价格统计中的样本代表性误差及其修正  

Sample error in housing price statistics and its adjustments

作　　者：吴璟[1] 郑思齐[1] 刘洪玉[1] 

机构地区：[1]清华大学,北京100084

出　　处：《土木工程学报》2009年第4期140-144,共5页China Civil Engineering Journal

基　　金：建设部研究开发项目(43906-K9-72);国家自然科学基金(70603017)

摘　　要：对住房价格变化的准确测量在当前背景下具有重要意义,样本代表性误差则是其中必须解决的重要问题。基于住房和住房市场的特殊性,分析住房价格统计过程中产生样本代表性误差的原因;将样本代表性误差划分为系统性误差和随机性误差两个组成部分,并比较其对我国住房价格统计的影响;提出评估价格法、固定权重法、特征价格法(包括截面建模形式和一次性建模形式)等四种针对样本代表性误差的修正方法,并通过北京市住房价格统计的数值模拟算例,比较各种修正方法的有效性和可行性。研究结果表明:样本代表性误差,尤其是其中的随机性误差部分,是当前制约我国住房价格统计结果质量的主要瓶颈之一,而一次性建模形式的特征价格法则是一种有效的修正该误差的方法。研究成果有助于我国住房价格统计的改进和完善。Price is a key indicator in housing market, and sample error is a major problem in housing price statistics. This paper analyzes the causes and effects of sample error in housing price statistics in China, and establishes four possible adjustment methods, namely, the appraisal method, the fixed-weight method, the cross-sectional hedonic method and the pooled hedonic method. Then the validity and feasibility of these methods are compared empirically by taking a simulated housing price index in Beijing as an example. The results suggest that the sample error, especially its random error component, significantly affects the accuracy of housing price indicator in China at present, and the pooled hedonic method is the most applicable and efficient method for error adjustment.

关 键 词：住房市场 住房价格统计 样本代表性误差 

分 类 号：F293.3[经济管理—国民经济]