Inferring Price Information from Mortgage Payment Behavior: A Latent Index Approach
Posted: 24 Jun 2016
Date Written: June 22, 2016
Existing price indices are based on real estate sales. This approach encounters problems when (1) sales are infrequent or (2) when these differ systematically from the overall market (selection bias). Relative to the number of properties sold on the market, a much greater number of properties have borrowers who need to make monthly mortgage payment decisions. Therefore, each month borrowers cast a vote of confidence of no confidence in their price relative to the loan balance. Based on this behavior, we invert the relation between mortgage performance and prices to derive a latent price index. Using a large sample of individual mortgages across the 10 cities investigated, the latent index in each city has a high correlation with the respective Case-Shiller index. In addition, the latent index is partially explained by the housing expectations (derived from futures on the respective Case-Shiller index) which indicates that it is not a purely reactive measure. Overall, the results show that the latent index has potential to boost the information resources in tracking the important real estate sector.
Keywords: House price index; Mortgage; Mortgage default; Latent index
Suggested Citation: Suggested Citation