Skip to main content

Deriving Local Trend Factors for Fair Market Rent Estimation

HUD.GOV HUDUser.gov
 
Research
HUD USER Home > PD&R Edge Home > Research
 

Deriving Local Trend Factors for Fair Market Rent Estimation

Deriving Local Trending Factors for Fair Market Rent Estimation

The U.S. Department of Housing and Urban Development (HUD) estimates Fair Market Rents (FMRs) for approximately 600 metropolitan areas and 2000 non-metropolitan counties in the United States to set housing assistance payment limits for public programs that address the housing needs of low-income families. As of FY 2019, the calculation of FMRs were simplified to three steps: estimating of gross rents paid by recent movers from the American Community Survey (ACS), applying an inflation adjustment measured using shelter rent and utility components of the Consumer Price Index (CPI), and applying a national trend factor to make gross rents current as of the effective fiscal year.

HUD has traditionally forecasted changes in components of gross rent using OMB projections of housing market drivers at the national level to derive a national trend factor. This step is necessary because reliable rental statistics lag the period for which current FMRs are effective. However, the use of a national trend factor means that FMRs do not reflect the most recent demand and supply drivers unique to local housing markets. In recent years, the most prevalent comments concerning FMRs are that they should incorporate more local and timelier data. A Senate Appropriations Committee report accompanying the 2018 Consolidated Appropriations Act raised concerns about the accuracy of FMRs, particularly in areas that have experienced rapid rent increases and have sought to conduct local rent surveys as an alternative to using FMRs.

This study investigates statistical approaches for HUD to add geographic resolution to the trend factor to address concerns regarding how well FMRs reflect local market conditions. The framework developed in this study assesses alternative FMR trend factor methodologies and proposes an approach for calculating FMRs that more accurately reflect local housing market trends. The research was divided into two phases. Phase I presents a statistical approach for deriving local trend factors, using extensions of HUD’s methodology in areas where local CPI data are available. Phase II develops an empirical framework for analyzing and utilizing third-party rental data to augment local market conditions for the calculation of FMRs.

Research Findings

Phase I of this study proposes and evaluates an approach for deriving local trend factors that builds upon HUD’s approach for calculating a national trend factor. The national trend factor incorporates national forecasted inputs from economic assumptions used in the formulation of the President's Budget and consists of two independently forecasted components of the CPI at the national level: Rent of Primary Residence and Housing - Fuels and Utilities. The study analyzes three alternative approaches for forecasting these components of the CPI in addition to replacing national CPI data with local CPI data for 13 of 22 metropolitan areas where historical CPI data is available and 4 regions (North, South, East, West) of the U.S. These three models are highlighted below:

  • National Input Model (NIM) - forecasts of change in local rent of primary residence from the Bureau of Labor Statistics (BLS) are informed by the forecast of national residential fixed investment (Bureau of Economic Analysis); forecasts of change in local fuels and utilities are informed by national forecasts of the price per barrel of West Texas Intermediate Crude Oil, the price per short ton of bituminous coal, and the seasonally adjusted Consumer Price Index, All Urban Consumers (CPI-U).
  • Local Input Model (LIM) - forecasts of change in local rent of primary residence are informed by historical trends in local variables such as local building permits (U.S. Census Bureau) and local area employment (BLS); forecasts of change in local fuels and utilities are informed by local electricity prices for utilities (U.S. Energy Information Administration).
  • Pure Time Series Model (PTS) - local forecasts are based upon previous values of the variable of interest: either the local rent of primary residence or fuel and utilities index.

For each local area, this study tests the performance of these forecast models by comparing actual data to an in-sample forecast (or validation period). Models are estimated using approximately 20 years of quarterly observations up to 2016 (Q1) and forecasted out through 2018 (Q1). These comparisons reveal how close rent and utility predictions of the validation period are as measured by the Root Mean Square Error Statistic (RMSE). Models yielding the lowest RMSE are determined to provide the most accurate estimates. For many areas the NIM model performed the best for forecasting changes in rent of primary residence, whereas the PTS model often performed best for forecasting changes in fuels and utilities. Research from Phase I concludes that local forecasts of indexed gross rent components are statistically different from a national forecast and show that local trend factors capture variation in local housing markets that is not captured in national forecasts.

Phase II of the research explores the possibility of expanding the number of local geographies for which trend factors can be calculated by utilizing a methodology that incorporates third-party rental data. Using such sources offer rental prices which are timelier and cover many more metropolitan areas than those analyzed in Phase I of the study. Changes in rental data from Axiometrics and Zillow are compared to changes observed in the CPI and ACS to determine the comparability and reliability of the third-party sources. The research shows that changes in Axiometrics data is consistent with local rent trends derived from the U.S. Bureau of Labor Statistics (BLS) and the ACS, however, recommends that further analysis of the data is needed to show that use of third-party data will improve FMR predictions.

Policy Implications

Overall, this study concludes that migrating away from a single national trend factor to local trend factors using the approach presented in Phase I is sound policy. However, the overall empirical impact of moving in this direction is relatively small. Less than one percent of FMR areas have an FY 2019 FMR that changed by more than two percent using the proposed methodology. Approximately 90 percent of areas experience an FMR change less than $15. The greatest change in rents across FMR areas for FY 2019 using the proposed methodology is 5 percent and mostly experienced in western portions of the country (e.g. Oakland, CA, San Francisco, CA, Phoenix, AZ, Denver, CO).

In calculating FY2020 FMRs, HUD adopted the Phase I approach of using local CPI data and the best performing forecast models in the trend factor calculation. As a result, approximately 42 percent of households participating in the Housing Choice Voucher program will have an FMR that incorporates a local trend factor with the remaining households benefiting from a regionally forecasted trend factor. Metropolitan areas that now receive a local trend factor as a result of this study are:


Boston-Cambridge-Newton, MA-NHWashington-Arlington-Alexandria, DC-VA-MD-WV
New York-Newark-Jersey City, NY-NJ-PABaltimore-Columbia-Towson, MD
Philadelphia-Camden-Wilmington, PA-NJ-DE-MDMiami-Fort Lauderdale-West Palm Beach, FL
Chicago-Naperville-Elgin, IL-IN-WIAtlanta-Sandy Springs-Roswell, GA
Detroit-Warren-Dearborn, MITampa-St. Petersburg-Clearwater, FL
Minneapolis-St.Paul-Bloomington, MN-WIDallas-Fort Worth-Arlington, TX
St. Louis, MO-ILLos Angeles-Long Beach-Anaheim, CA
Houston-The Woodlands-Sugar Land, TXSan Francisco-Oakland-Hayward, CA
Phoenix-Mesa-Scottsdale, AZSeattle-Tacoma-Bellevue, WA
Denver-Aurora-Lakewood, COSan Diego-Carlsbad, CA
Urban AlaskaUrban Hawaii

Find the full report on deriving local trend factors here.

 
 
Published Date: 9 September 2019


The contents of this article are the views of the author(s) and do not necessarily reflect the views or policies of the U.S. Department of Housing and Urban Development or the U.S. Government.