Author:
Alexander N. Bogin, Senior Economist; William M. Doerner, Senior Economist; William D. Larson, Senior Economist
Abstract:
We make two contributions to the study of house price index and mortgage credit modeling accuracy. First, we assess the predictive power of house price indices calculated at different levels of geographic aggregation. Lower levels of aggregation offer superior fit when appreciation rates vary substantially across submarkets and the indices are based on a sufficient number of transactions. Second, we estimate a competing options credit model using 15 years of mortgage performance data in the United States. Model accuracy is highest when using indices at a city or lower level of aggregation to construct current loan-to-value ratios. Fit is weaker when using state or national price indices. Overall, this research highlights the benefits of using more localized house price indices when predicting property values and mortgage performance.
Our FAQs address common questions about the indices. Please cite this working paper when using the local HPI data. The local HPI data can be downloaded on the HPI Downloadable Data page or with these following links
This research was selected as the best paper in 2016 in real estate market analysis by the American Real Estate Society. A revised version of this paper has undergone external peer-review and is published in an academic journal. Citation: Alexander N. Bogin, William M. Doerner, William D. Larson. 2019. "Missing the Mark: Mortgage Valuation Accuracy and Credit Modeling. "Financial Analysts Journal, 75(1), 32-47. https://www.tandfonline.com/doi/abs/10.1080/0015198X.2018.1547051
Related papers: FHFA Working Paper 16-01: Local House Price Dynamics and FHFA Working Paper 16-02: Local House Price Growth Accelerations