Estimating Loan‐to‐Value Distributions
估计了一个包含房价、贷款价值比和交易违约行为的模型，通过贝叶斯滤波纠正样本选择偏差，发现加州阿拉米达县35%的独栋住宅资不抵债，远高于传统方法估计的19%。
We estimate a model of house prices, combined loan‐to‐value ratios (CLTVs) and trade and foreclosure behavior. House prices are only observed for traded properties and trades are endogenous, creating sample‐selection problems for existing approaches to estimating CLTVs. We use a Bayesian filtering procedure to recover the price path for individual properties and produce selection‐corrected estimates of historical CLTV distributions. Estimating our model with transactions of residential properties in Alameda, California, we find that 35% of single‐family homes are underwater, compared to 19% estimated by existing approaches. Our results reduce the index revision problem and have applications for pricing mortgage‐backed securities.