ABSTRACT

This study uses U.S. Census data on average home values in Southern California census tracts from 1960 to 2010. Using growth mixture modeling (GMM), 26 unique groups are detected capturing nonlinear change in neighborhood relative home values over this study period. There were seven broad patterns of changing home values: (1–3) decline and then rise (at high, mid, and low portions of the home value distribution); (4) rise and then decline; (5–6) a monotonic increase (either above or below the region average); and (7) a monotonic decrease. Multinomial regression models found that covariates exhibited a much stronger effect for distinguishing between the average level of home values in neighborhoods over the study period, rather than how home values changed over time.

摘要

这项研究使用了1960年至2010年南加州人口普查区域的平均房屋价值的美国人口普查数据。使用增长混合模型（GMM），检测到26个独特的组，以捕获该研究期间邻域相对房屋价值的非线性变化。有七种改变房屋价值的主要模式：（1-3）先下降后上升（在房屋价值分布的高，中，低部分）；（4）先升后降；（5-6）单调增加（高于或低于区域平均值）；（7）单调下降。多项回归模型发现，在研究期间，协变量显示出更好的区分邻居平均房价水平的效果，而不是随着时间的推移房价的变化。