The Advantages of Using CoreLogic CBSA Level HPI Stress-Testing Scenarios

The Federal Reserve’s Comprehensive Capital Analysis and Review program provides Home Price Index (HPI) stress-testing scenarios at the national level but does not include testing at the local metropolitan level. This approach lacks consideration of the substantial heterogeneity of home price cycles at more disaggregate geographic levels. To address this concern, CoreLogic has created HPI stress-testing scenarios at the Core Based Statistical Area (CBSA) level by combining historical home price index values with CoreLogic standard HPI forecasts to disaggregate the Federal Reserve’s national-level stress scenarios.¹

CoreLogic applies the CBSA-level HPI forecast stress-testing scenarios to the CoreLogic RiskModel Agency using Freddie Mac 2006 origination data. As referenced in the previous related blog²: A Closer Look at Freddie’s Single-Family Loan-Level Dataset; Freddie Mac has released its fixed-rate loan performance data (originations from 1999 to 2016) to the public³. To see the advantage of using the CBSA-level over national-level HPI stress-testing scenarios, we take California as an example and simulate forecasts for the three-year period from 2016 to 2018. We use Freddie’s 2006 origination sample because it has the most active loans, about 7,000 active Freddie Mac California accounts, and covers the bubble-burst and recovery periods. The California state baseline scenario forecasts 6 percent average annual home price appreciation over the three-year scenario horizon compared to a 3 percent increase for the Federal Reserve’s national baseline. In the adverse scenario, California home prices fall by 18 percent (compared to a U.S. 12 percent drop) before starting to recover. California home prices drop by 33 percent (compared to a U.S. 25 percent drop) in the severely adverse scenario. These CBSA HPI forecast scenarios are fully consistent with the Federal Reserve’s 2017 national HPI scenarios.

Figure 1 shows the conditional default rate (CDR) from the stress test forecasting over this sample. At the base scenario (blue group), the stress testing based on national and CBSA-level HPI forecasts shows low and similar default rates. As we move on to the adverse scenario (orange group), the test based on the CBSA-level HPI scenarios (orange, solid line) correctly predicts higher default rates than those at the national level (orange, dotted line). When it comes to the worst-case scenario (green group), we see an even larger gap between the two default rates. In particular, the test conducted at the more disaggregate CBSA level (green, solid line) yields substantially higher default rates than its counterpart at the national level (green, dotted line), implying a much bigger concern for California when hit by a severe crisis.

Figure 2 shows the conditional prepayment rate (CPR) over the same sample. Since our CBSA-level stress scenarios are based on predictions of home price index rather than interest rates, which are the most important driver of prepayments, the differences in prepayment rates from the different scenarios are subtle. Yet, for both adverse scenarios (orange and green groups), the stress test conducted at the CBSA level (solid line) predicts lower prepayment rates than those at the national level (dotted line). These differences are clearly visible at the one- and two-year ahead forecasts.

Using the stress testing at the more granular CBSA level can help banks, businesses, investors and policy makers customize their own portfolios and efficiently address their regulatory concerns.

Ling Chen contributed to this analysis.

© 2017 CoreLogic, Inc. All rights reserved.

CoreLogic.com