Can Geospatial Data Improve House Price Indexes? A Hedonic Imputation Approach with Splines
研究结合享乐插补与非参数样条曲面利用地理空间数据，发现相比邮编虚拟变量，地理数据对悉尼房价指数精度提升有限，对资源有限的统计机构是好消息。
Determining how and when to use geospatial data (i.e. longitudes and latitudes for each house) is probably the most pressing open question in the house price index literature. This issue is particularly timely for national statistical institutes (NSIs) in the European Union, which are now required by Eurostat to produce official house price indexes. Our solution combines the hedonic imputation method with a flexible hedonic model that captures geospatial data using a non‐parametric spline surface. For Sydney, Australia, we find that the extra precision provided by geospatial data as compared with postcode dummies has only a marginal impact on the resulting hedonic price index. This is good news for resource‐stretched NSIs. At least for Sydney, postcodes seem to be sufficient to control for locational effects in a hedonic house price index.