This annual report describes FHFA's accomplishments, as well as challenges, the agency faced in meeting the strategic goals and objectives during the past fiscal year.
Read about the agency’s 2019 examinations of Fannie Mac, Freddie Mac and the Home Loan Bank System.
Submit comments and provide input on FHFA Rules Open for Comment by clicking on Rulemaking and Federal Register.
Implement critical reforms that will produce a stronger and more resilient housing finance system.
FOSTER competitive, liquid, efficient, and resilient (CLEAR) national housing finance markets that support sustainable homeownership and affordable rental housing; OPERATE in a safe and sound manner appropriate for entities in conservatorship; and PREPARE for eventual exits from the conservatorships.
2019 Conservatorships Strategic Plan
FHFA experts provide reliable data, including all states, about activity in the U.S. mortgage market through its House Price Index, Refinance Report, Foreclosure Prevention Report, and Performance Report.
FHFA economists and policy experts provide reliable research and policy analysis about critical topics impacting the nation’s housing finance sector. Meet the experts...
The FHFA HPI suite is built on tens of millions of home sales and offers insights about house price fluctuations at the national, census division, state, metro area, county, ZIP code, and census tract levels. The FHFA HPI uses a fully transparent methodology based upon a weighted, repeat-sales statistical technique to analyze transaction data from Fannie Mae and Freddie Mac. The repeat-sales technique helps to control for differences in the quality of the houses comprising the sample and is the reason the FHFA HPI is referred to as a "constant quality" index. This distinction is important when comparing summary statistics like mean or median values (such as those provided by the American Community Survey, the Decennial Census, or Zillow) which inherently reflect changes to both the price and quantity of housing. By isolating price changes, the FHFA HPI is more reflective of the average appreciation in an area.
The FHFA HPI also provides an analytical tool that is useful for a range of housing finance topics such as updating the current value of residential property assets, estimating potential mortgage default and any losses given default, predicting prepayment speeds for financial securities, and gauging differences in housing affordability in specific geographic areas. Researchers have used the FHFA HPI data for hundreds of studies about business cycles, demographic challenges, environmental or disaster risks, labor markets, local government budgets, migration, political elections, property taxation, urban revival, and wealth creation.
FHFA HPI reports are released on a monthly and quarterly basis. Calendar dates are always posted at the bottom of the FHFA HPI main webpage or can be found by typing:
Release dates are announced at the end of every summer for the following calendar year. The reports and data are posted at 9AM on announced dates. Each month, consolidated reports highlight trends for certain regions (census divisions) as well as the entire country. Every three months, a more comprehensive report appends additional quarterly information about those areas while including new coverage for states and cities.
Most statistics in the reports reference price changes computed by FHFA's standard "purchase-only" HPI. In some cases, however, the statistics reference alternative price measures. These additional house price indexes use the same general methodology as the flagship index but have slight differences as follows:
Data constraints preclude the production of all types of indexes for every geographic area, but multiple index types are generally available. For individual states, for instance, several types of indexes are available. The various indexes tend to correlate closely over the long-term, but short-term differences can be significant.
The methodology used to construct the indexes is a modified version of the Case-Shiller® geometric weighted repeat-sales procedure. A detailed description of the HPI methodology is available at:
The repeat-sales procedure estimates price changes using repeat transactions on the same property units over time. This approach helps to control for differences in the quality of the houses comprising the sample and is the reason the FHFA HPI is referred to as a "constant quality" index.
The monthly indexes are calculated in the same way as the quarterly and annual indexes except the time period is controlled for more directly. In the monthly indexing model, all the transactions for the same month are aggregated and separate index values are estimated for each month. To construct the quarterly index, every transaction from the same quarter is aggregated and index values are estimated using the assigned quarters. The same approach is used with the annual indexes. For advanced data users, this means that dummy or constant variables are not included as additional time-varying controls.
The FHFA HPI is based on transactions for single-family properties involving conforming, conventional mortgages purchased or securitized by Fannie Mae or Freddie Mac.
Conforming refers to a mortgage that both meets the underwriting guidelines of Fannie Mae or Freddie Mac and that does not exceed the conforming loan limit. For 2020-acquired loans, the ceiling limit rose to $765,600 for one-unit homes in the contiguous U.S. More information on the conforming loan limits can be found at
Conventional mortgages are those loans that are neither insured nor guaranteed by the FHA, VA, or other federal government entities.
The flagship FHFA HPI eliminates non-conforming and non-conventional mortgages during index production. The sample also restricts data on condominiums, cooperatives, multi-unit properties, and planned unit developments. Additional filters are applied to remove property observations with potential recording errors like extremely low sales values, implausible appreciation changes, repeat sales within the same period, and incomplete date or physical property address information.
Historical estimates of the FHFA HPI revise for three primary reasons:
1) The FHFA HPI is based on repeat transactions. That is, the estimates of appreciation are based on repeated valuations of the same property over time. Therefore, each time a property "repeats" in the form of a sale or refinance, average appreciation since the prior sale/refinance period is influenced.
2) Fannie Mae and Freddie Mac (the Enterprises) purchase seasoned loans, providing new information about prior quarters.
3) Due to a 30- to 45-day lag from loan origination to Enterprise funding, FHFA receives data on new fundings for one additional month following the last month of the quarter. These fundings contain many loans originating in that most recent quarter, and especially the last month of the quarter. This will reduce with subsequent revisions, however data on loans purchased with a longer lag, including seasoned loans, will continue to generate revisions, especially for the most recent quarters.
In connection with the release of the 2012Q2 FHFA HPI results, a special revision was made to two historical HPI values. In prior releases, the all-transactions index values for Vermont-1976Q1 and West Virginia-1982Q1 were both reported to be 100.01. Those values were not correct; index values for those respective periods should have been set to missing because no modeling data were available in the underlying sample. The FHFA HPI releases for 2012Q2 and later periods reflect the change. With the release of the 2019Q1 FHFA HPI results, modeling data became available for Vermont-1976Q1. The FHFA HPI releases for 2019Q1 and later periods reflect the change.
Although both indexes employ the same fundamental repeat-valuations approach, there are several data and methodology differences.
Among the dissimilarities:
For details on these and other differences, consult the FHFA HPI Technical Description at
http://go.usa.gov/8BBT and the Case-Shiller methodology materials at
A paper that analyzes in detail the methodological and data differences between the two metrics can be accessed at
Many of the FHFA HPIs are provided in both seasonally adjusted and non-seasonally adjusted terms. The seasonal adjustment is designed to remove recurring calendar year patterns or seasonal variation from the HPI which would otherwise partially confound month-to-month or quarter-to-quarter comparisons of appreciation.
The Census Bureau's X-12 ARIMA procedure is used, as implemented in the SAS software package. The automated ARIMA model-selection algorithm in X-12 is employed, which searches through a series of seasonality structures and selects the first that satisfies the Ljung-Box test for serial correlation. To obtain more information on the FHFA HPIs contact us via the Data and Research Contact page at
Yes. The expanded-data index includes purchase-money mortgages from other sources to provide an index which tries to capture all loans which could potentially make up the complete conforming mortgage market. The approach to estimating the expanded-data HPI is detailed in the Highlights article published with the 2011Q2 FHFA HPI at
In general, the methodology is the same as is used in the construction of the standard purchase-only FHFA HPI, except a supplemented dataset is used for estimation. The augmented data include sales price information from Fannie Mae and Freddie Mac mortgages as well as two new information sources: (1) transactions records for houses with mortgages endorsed by FHA and (2) county recorder data licensed from CoreLogic. The licensed county recorder data do not include records in all U.S. counties-particularly rural ones. To ensure that the addition of the CoreLogic data does not unduly bias index estimates toward price trends in urban areas, the expanded-data index for certain states is estimated by weighting price trends in areas with CoreLogic coverage and other areas. Details on this sub-area weighting can be found in the text of the Highlights piece referenced above.
FHFA uses the revised Metropolitan Statistical Areas (MSAs) and Divisions as defined by the Office of Management and Budget (OMB) in September 2018. The delineations became effective with the 2018Q4 FHFA HPI release in February 2019. For information about the current MSAs, please visit:
Not all data sources use this same delineation when referring to Core Based Statistical Areas (CBSAs). Differences can arise when newer delineations are released by OMB or when sources may not be retroactively updated (data in prior years might reflect older geographic boundaries).
This is a tricky geospatial question. ZIP codes sometimes overlap county boundaries, a single ZIP code can be located partially inside and outside of a Metropolitan Area. Thus, the development of a crosswalk between ZIP codes and Metropolitan Areas is not a straightforward exercise. The Department of Housing and Urban Development has released a lookup table that maps ZIP codes to the Metropolitan Area(s) that they fall within. That lookup file, as well as a discussion of the underlying technical issues, can be found here:
Although FHFA has published HPIs for some ZIP codes, those indexes are annual (i.e. quarterly index values are not provided). Researchers needing quarterly values for ZIP codes may be interested in using index values for the applicable metropolitan area.
For many small- or medium-sized cities, the ZIP3 is almost synonymous with the MSA area. Some published mortgage datasets show the "three-digit" ZIP code of the included properties, where the three-digit ZIP code is merely the first three digits of the applicable five-digit code. For example, a property whose ZIP code is 91711 would be in the "917" three-digit ZIP code. Three-digit ZIP codes represent larger geographic areas than five-digit ZIP codes.
To aid modelers who have three-digit ZIP codes, FHFA has released a set of experimental HPIs for such areas. The underlying information used for calibrating the indexes is the "all-transactions" dataset, which includes information on both home purchases and refinances. FHFA HPIs can be used to approximate current home values by assuming that prices have changed by the same percentage as the applicable house price index for the local area. Given loan-level data, the same technique can be applied to update the underlying value of securitized mortgage pools. The choice of the applicable index used to update home values depends on the geographic information available in the data.
As discussed in the Highlights article accompanying the 2011Q1 FHFA HPI Release (available for download at
http://go.usa.gov/8k5d), the census division indexes are constructed from statistics for the component states. For the quarterly all-transactions and purchase-only indexes, the census division indexes are constructed from quarterly growth rate estimates for the underlying state indexes. Census division index estimates are "built-up" from quarterly growth rate estimates (monthly growth rates for the monthly index) for the component states.
The census division indexes are set equal to 100 in the relevant base periods. Then, the index values for subsequent periods are increased (or decreased) by the weighted average quarterly (or monthly) price change for the underlying states. Index values for periods before the base period are calculated in a similar fashion; beginning with the base period value, the preceding index values are sequentially determined so that the growth rate in each period always reflects the weighted average growth rate for the component states.
The national FHFA HPI is constructed in an analogous fashion, except that the weighted components are census divisions. Because the census divisions measures are themselves weighted averages of state metrics, the U.S. index is equivalent to a state-weighted metric.
The weights used in constructing the FHFA HPI are estimates for the shares of one-unit detached properties in each state. For years in which decennial census data are available, the share from the relevant census is used. For intervening years, a state's share is the weighted average of the relevant shares in the prior and subsequent censuses, where the weights are changed by ten percentage points each year. For example, California's share of the housing stock for 1982 is calculated as 0.8 times its share in the 1980 census plus 0.2 times its share in the 1990 census. For 1983, the Pacific Division's share is 0.7 times its 1980 share plus 0.3 times its 1990 share.
For years since 2000, state shares are calculated as follows:
The year-specific estimates of the state shares of U.S. detached housing stock can be accessed at
The FHFA HPI includes indexes for all nine census divisions, the 50 states and the District of Columbia, and every Metropolitan Statistical Area (MSA) in the U.S., excluding Puerto Rico. OMB recognizes 384 MSAs, 11 of which are subdivided into a total of 31 Metropolitan Divisions. As noted earlier, FHFA produces indexes for the divisions where they are available, in lieu of producing a single index for the MSA. In total, 404 indexes are released: 373 for the MSAs that do not have Metropolitan Divisions and 31 Division indexes. The starting dates for indexes differ and are determined by a minimum transaction threshold; index values are not provided for periods before at least 1,000 transactions have been accumulated.
In each release, FHFA publishes rankings and appreciation rates (quarterly, annual, and five-year). To be ranked, the MSAs or Metropolitan Divisions must have at least 15,000 transactions over the prior 10 years. For the remaining unranked MSA or MSAD areas, one-year and five-year rates of change are provided.
The "four-quarter" percentage change in home values is simply the price change relative to the same quarter one year earlier. For example, if the FHFA HPI release is for the second quarter, then the "four-quarter" price change reports the percentage change in values relative to the second quarter of the prior year. It reflects the best estimate for how much the value of a typical property increased over the four-quarter period. "One-year" and "annual" appreciation are used synonymously with "four-quarter" appreciation in the full quarterly FHFA HPI releases.
The "one-quarter" change estimates the price gains relative to the prior quarter. Please note that, in estimating the quarterly price index, all observations within a given quarter are pooled together; no distinction is made between transactions occurring in different months. As such, the "four-quarter" and "one-quarter" changes compare typical values throughout a quarter against valuations during a prior quarter. The appreciation rates do not compare values at the end of a quarter against values at the end of a prior quarter.
The index numbers alone (for census divisions and U.S., individual states, and MSAs) do not have significance. They have meaning in relation to previous or future index numbers, because you can use them to calculate appreciation rates using the formula below.
To calculate appreciation between any 2 quarters, use the formula:
(QUARTER 2 INDEX NUMBER - QUARTER 1 INDEX NUMBER) / QUARTER 1 INDEX NUMBER
You can generate annual numbers by taking the four-quarter average for each year or monthly numbers by finding the difference between two months. For ease, we have made a calculator tool that is available at:
© 2020 Federal Housing Finance Agency