Federal Housing Finance Agency Print
Home / Media / Blog / Geographic Disaggregation of House Price Stress Paths

Geographic Disaggregation of House Price Stress Paths

Category: FHFA Stats
Published: 11/17/2023

​​​​​​​​​​Financial managers calculate the risk of holding single-family mortgages to understand what expected losses (or gains) they may anticipate in the future, and, when applicable, how much capital they should potentially set aside to cover future losses. Usually, this is done by running a pool of loans through a statistical model to estimate future cash flows, loan defaults, and loan prepayments. Macroeconomic scenarios provide a tool to test the impact of future house price movements on these cash flow estimates and potential loan outcomes. 

A reasonable risk manager may wonder if it's enough to specify a national house price path to perform these calculations. But real estate is about local markets. Would it be better to work with state level paths? Alternatively, some states have several metropolitan areas which can be starkly different leading to further house price variation. What would happen if we try to incorporate these more localized house price dynamics and work with metro level paths? A recently released working paper attempts to provide answers.

Figure 1 splits up historical house price declines at the state-level (solid maroon line) and metro-level (solid green line). As depicted by the metro-level density line extending more to the right-hand side of the graphic, the distribution of declines isn't the same—there is a greater frequency of more severe declines at the metro-level. The median price decline across state-level paths is 29 percent while across metro-level paths it is a larger 31 percent.  The metro-level distribution is also characterized by a fatter right-hand tail indicating more instances of large house price declines. Going down to a smaller level might just matter.

Figure 1: Comparing State-Level and Metro-Level House Price Declines

A figure that compares the distributions of state-level and metro-level house price declines. 

A common way to model house price paths is to see how they've done in the past and then create bounds for what might be reasonable to expect in the future. Figure 2 shows us how this can be done using the state of California as an example. We look back in time to estimate the typical price trend (dashed green line) and how much the house price index (HPI is shown with the solid black line) have dropped below trend. That trough (dashed red line) can then be projected across time so that an analyst is able to calculate how much prices might drop at any future date (solid gold line). 

Figure 2: House Price Paths for California

A figure that shows house price paths for a state (California). A house price path is shown along with a trend line, trough, and the CMAST path.

Figure 3 includes two panels that illustrate how much house price patterns can potentially differ, even within the same state.  For this example, we choose two different metro areas in California, San Jose (panel a) and Riverside (panel b). If prices had dropped suddenly at the start of this year, the state-level path for California would have anticipated a decline of 33 percent. However, if a risk manager were only interested in one of the two metro areas, they may expect a smaller decline of 19 percent in San Jose or a significantly larger decline of 49 percent in Riverside. In other words, location does matter and by creating disaggregated house price paths, analysts can be better equipped to measure historically based scenarios. 

Figure 3: House Price Paths for Metro Areas in California

​(a) San Jose   

A figure that shows house price paths for a metropolitan statistical area (San Jose). A house price path is shown along with a trend line, trough, and the CMAST path.

​  (b) Riverside

A figure that shows house price paths for a metropolitan statistical area (Riverside). A house price path is shown along with a trend line, trough, and the CMAST path.


The paper builds on this discussion to show how much risk management estimates might change across these different macroeconomic scenarios. The state-level and metro-level house price paths are contrasted to show the impact on portfolio-level stress losses and how different credit models might perform. There are a variety of other ways to construct macroeconomic shocks or to construct potential tail-event scenarios. However, the magnitude that house prices fall from their long-term trend has been remarkably consistent across different kinds of national and regional cycles. For that reason, we think it is useful to turn to historical scenarios and even more helpful to provide paths at the metro level as done in this paper.

For additional information on these results and further discussion, we invite you to read FHFA Working Paper 23-02 on "Geographic Disaggregation of House Price Stress Path." The paper can be viewed on the FHFA website and a published version is forthcoming at the Journal of Fixed Income.

Tagged: Research; House Prices; Federal Home Loan Banks; Source: FHFA

By: Alex Bogin

Principal Economist, Capital Requirements and AMA Modeling Section
Office of Risk Analysis and Modeling
Division of Research and Statistics

By:  Will Doerner

Supervisory Economist, Research Branch
Office of Research and Analysis
Division of Research and Statistics

© 2024 Federal Housing Finance Agency