As More Households Rent, the Poorest Face Affordability and Housing Quality Challenges (GAO-20-427)

Summary

A report to congressional requesters by the United States Government Accountability Office, GAO-20-427, issued May 2020, titled Rental Housing: As More Households Rent, the Poorest Face Affordability and Housing Quality Challenges. GAO analyzed survey data from 2001 through 2017 and finds that in 2017 almost 7 million more households rented than in 2001, raising the share of households that rent from an estimated 34 percent to 36 percent. It finds that 48 percent of renter households were rent burdened in 2017, paying over 30 percent of income for rent, 6 percentage points higher than in 2001, with rent burden most common among households at 80 percent or less of area median income. The report notes that the federal government subsidizes rents for around 4.4 million households per year and that it examines trends prior to the COVID-19 pandemic.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Full text

Text recognized from a scanned document; may contain errors. Page 53 of 96 is not shown because the text could not be recognized reliably.

United States Government Accountability Office

Report to Congressional Requesters

May 2020

RENTAL HOUSING

As More Households
Rent, the Poorest
Face Affordability and
Housing Quality
Challenges

GAO-20-427
GAO
Highlights

Highlights of GAO-20-427, a report to
congressional requesters

Why GAO Did This Study

Since the 2007—2009 financial crisis,
growth in the share of renter households
has reversed a decades-long trend
toward homeownership. This change
has underscored concerns about the
availability, affordability, and condition of
rental housing, especially for low-
income households. The federal
government subsidizes rents for around
4.4 million households per year, but
more households qualify for assistance
than receive it.

GAO was asked to provide a
comprehensive assessment of the
housing market. This report examines
trends in the housing market prior to the
COVID-19 pandemic and does not
account for the profound impact it will
likely have on renter households. This
report, one of several GAO plans to
issue, focuses on rental housing from
2001 through 2017 and analyzes (1) the
share of households that rent, (2) the
affordability of rental housing, and (3)
rental housing conditions.

GAO analyzed American Community
Survey and American Housing Survey
data from 2001 through 2017 (the most
recent data available at the time of this
review) at the national level and for
different types of localities. GAO also
reviewed recent reports by the
Department of Housing and Urban
Development (HUD), research
organizations, and academic
researchers on rental housing and
obtained views from a variety of
stakeholders selected for their
knowledge of these issues, including
federal agency officials, academic
experts, research organizations, and
industry groups.

View GAO-20-427. For more information,
contact Daniel Garcia-Diaz at (202) 512-4529 or
garciadiazd@gao.gov.

RENTAL HOUSING

As More Households Rent, the Poorest Face
Affordability and Housing Quality Challenges

What GAO Found

In 2017, almost 7 million more households rented their homes than in 2001,
which brought the share of households that rent from an estimated 34 percent to
36 percent. Renting became more common after the 2007—2009 financial crisis
as foreclosures and changes in household characteristics reduced the proportion
of homeowners. Renting was more prevalent across most age and race/ethnicity
groups in 2017 than in 2001, with notable increases among higher-income
households.

Rental affordability declined from 2001 to 2017. In 2017, 48 percent of renter
households were rent burdened—that is, they paid over 30 percent of income for
rent—which is 6 percentage points higher than in 2001. Rent burden was most
common and most severe among lower-income households (80 percent or less
than area median income), with almost three-quarters of extremely low-income
households (30 percent or less than area median income) paying over half of
their income in rent (see figure). Affordability declined because of a range of
factors, including more households competing for rental units and the supply of
low-cost rental units not keeping up with demand.

Estimated Percentage of Renter Households with Rent Burdens by Income in 2017

Percentage of renter households

100
90
80
70 No rent burden — rent costs are at or below
60 30 percent of household income
50 94% Moderate rent burden — rent costs are more
40 than 30 and up to 50 percent of household
30 income
20 Severe rent burden — rent costs exceed 50

percent of household income

10

Extremely Very low Low Moderate Higher
low income income’ income’ income
income

Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development. | GAO-20-427

Note: Estimates in this figure have a margin of error of +2 percentage points or fewer, at
the 95 percent confidence level.

An estimated 15 percent of rental units in 2017—more than 5 million—had
substantial quality issues (such as cracked walls and the presence of rodents) or
lacked essential components of a dwelling (such as heating equipment or hot
and cold running water), according to GAO’s analysis of American Housing
Survey data. The share of units with deficiencies was relatively stable from 2001
to 2017. Serious deficiencies more often affected households with extremely low
incomes or rent burdens. In addition, lower-income households rented
approximately two-thirds of the units with substantial quality issues and nearly 80
percent of units lacking essential components.

United States Government Accountability Office
Contents

Letter 1
Background 4
Renting Became More Common after the 2007-2009 Financial
Crisis but Varied by Demographic Group and Location 5
Rent as Share of Income Increased from 2001 through 2017, with
Serious Consequences for the Poorest 14
About 15 Percent of Rental Units Had Serious Deficiencies in
2017 25
Agency Comments 32
Appendix | Objectives, Scope, and Methodology 34
Appendix II Statistical Analysis of Rental Housing Conditions 40
Appendix III Additional Information on Rentership and Affordability Trends 54
Appendix IV Estimated Rent Burden in State-wide Rural Areas 65
Appendix V Comparison of GAO Housing Conditions Indexes and HUD Adequacy
Index 74
Appendix VI Additional Information on Rental Housing Conditions 79
Appendix VII GAO Contact and Staff Acknowledgments 89
Tables

Table 1: Types of Localities by Population Growth 2000—2017 and
Density, as of 2017 4

Page i GAO-20-427 Rental Housing
Table 2: Condition of Rental Housing Units Based on GAO’s

Quality Indexes, 2017 27
Table 3: GAO Rental Housing Conditions Indexes 38
Table 4: GAO Rental Housing Conditions Index Variables 41
Table 5: Renter Household Responses to AHS Questions about

Rental Unit Conditions, 2017 42
Table 6: National Distribution of Completeness Index by AHS

Year, 2001-2017 43
Table 7: Polychoric Correlations between Variables in Quality

Factor Model, 2001-2017 47
Table 8: Factor Loadings Model Estimates 48
Table 9: Distribution of Quality Index by AHS Year, 2001-2017 50
Table 10: Most Frequent Profiles of Quality Issues by Quality

Level, 2017 51
Table 11: Estimated Percentage of Households That Rent by Age,

Race/Ethnicity, Locality Type, and Income, 2001-2017 54
Table 12: Estimated Number of Renter Households by Age,

Race/Ethnicity, Locality Type, and Income, 2001-2017 56
Table 13: Estimated Number of Households with Rent Burdens by

Age, Race/Ethnicity, and Locality Type, 2001-2017 57
Table 14: Estimated Percentage of Households with Rent Burdens

by Age, Race/Ethnicity, and Locality Type, 2001-2017 60
Table 15: Estimated Number of Households with Rent Burdens by

Income, 2001—2017 62
Table 16: Estimated Percentage of Households with Rent Burdens

by Income, 2001-2017 63
Table 17: Estimated Rent-to-Income Ratio by Demographic Group

and Locality Type, 2001-2017 64
Table 18: GAO Quality and Completeness Indexes, Rental Units

by Housing Condition and Income, 2017 79
Table 19: GAO Quality and Completeness Indexes, Rental Units

by Housing Conditions and Rent Burden, 2017 80
Table 20: GAO Quality and Completeness Indexes, Rental Units

by Condition and Race/Ethnicity, 2017 81
Table 21: GAO Quality and Completeness Indexes, Rental Units

by Condition and Age, 2017 82
Table 22: GAO Quality and Completeness Indexes, Rental Units

by Condition and Structure Age, 2017 83
Table 23: GAO Quality and Completeness Indexes, Rental Units

by Condition and Unit Type, 2017 84

Page ii GAO-20-427 Rental Housing
Figures

Figure 1: Estimated Percentage of Households That Rent, 2001-—

2017 6
Figure 2: Estimated Rentership Rates by Age Group, 2001—2017 8
Figure 3: Estimated Number of Renter Households by Age Group,

2001 and 2017 AS)
Figure 4: Estimated Rentership Rates by Race and Ethnicity,

2001-2017 10
Figure 5: Estimated Number of Renter Households by Household

Income, 2001-2017 12
Figure 6: Estimated Change in Renter Households by Income and

Locality Type, 2001-2017 13
Figure 7: Estimated Percentage of Renter Households by Rent

Burden in 2001 and 2017 15
Figure 8: Rent Burden among Selected Demographic Groups and

Locality Types in 2017 18
Figure 9: Estimated Percentage of Renter Households with Rent

Burdens by Income in 2017 19
Figure 10: Estimated Number of Renter Households by Rent

Burden and Household Income, 2001—2017 21
Figure 11: Rent Burden for Families of Four Who Paid Fair Market

Rent in the St. Louis and San Francisco Areas, 2017 23
Figure 12: Estimated Median Rent-to-Income Ratio in 2001 and

2017 25

Figure 13: GAO Quality and Completeness Indexes, Estimated
Percentage of Rental Units by Housing Condition, 2001-
2017 28
Figure 14: GAO Quality and Completeness Indexes, Estimated
Percentage of Rental Units with Serious Deficiencies by

Housing Type, 2001-2017 31
Figure 15: Distribution of Continuous Quality Index, 2001, 2005,

2009, 2013, and 2017 46
Figure 16: Estimated Number of Renter Households in Rural

Areas by State, 2013-2017 66
Figure 17: Estimated Percentage of Renter Households in Rural

Areas by State, 2013-2017 67
Figure 18: Estimated Percentage of Rural Renter Households with

Income Less than $35,000 by State, 2013-2017 69
Figure 19: Estimated Percentage of Rural Renter Households with

Rent Burdens by State, 2013-2017 71

Page iii GAO-20-427 Rental Housing
Figure 20: Estimated Percentage of Rural Renter Households with

Income Less than $35,000 That Were Rent Burdened by

State, 2013-2017 72
Figure 21: Estimated Percentage of Rural Renter Households with

Income $35,000 or Greater That Were Rent Burdened by

State, 2013-2017 74
Figure 22: Comparison of the Department of Housing and Urban

Development's (HUD) Adequacy Index and GAO’s

Housing Conditions Indexes 76
Figure 23: Comparison of Findings from the Department of

Housing and Urban Development's (HUD) Adequacy

Index and GAO’s Housing Conditions Indexes, 2017 78
Figure 24: Estimated Percentage of Crowded Renter Households

by Race/Ethnicity, 2001-2017 85
Figure 25: Estimated Percentage of Crowded Renter Households

by Age, 2001-2017 86
Figure 26: Estimated Percentage of Crowded Renter Households

by Income, 2001—2017 87
Figure 27: Estimated Percentage of Crowded Renter Households

by Locality Type, 2001-2017 88

Abbreviations

AHS American Housing Survey

HAMFI Department of Housing and Urban Development area
median family income

HUD Department of Housing and Urban Development

RUCA rural-urban commuting area

This is a work of the U.S. government and is not subject to copyright protection in the
United States. The published product may be reproduced and distributed in its entirety
without further permission from GAO. However, because this work may contain
copyrighted images or other material, permission from the copyright holder may be
necessary if you wish to reproduce this material separately.

Page iv GAO-20-427 Rental Housing
GAO U.S. GOVERNMENT ACCOUNTABILITY OFFICE

441 G St. N.W.
Washington, DC 20548

May 27, 2020
Congressional Requesters

Since the 2007-2009 financial crisis, growth in rentership—that is, the
share of households that rent—has reversed a decades-long trend toward
homeownership. Although a majority of U.S. households still own their
homes, the larger number of renters has led to increased competition for
rental housing. These changes have given rise to concerns about the
availability and affordability of rental housing, especially for households
with low incomes. While the federal government subsidizes rents for
around 4.4 million households per year, housing assistance is not an
entitlement, and more households qualify for assistance than receive it.1

You requested that we conduct a comprehensive assessment of the U.S.
housing market, including whether the housing market is meeting the
needs of the American people. This report examines trends in the
housing market prior to the Coronavirus Disease 2019 pandemic and
does not account for the profound impact it will likely have on renter
households. We have ongoing work that will examine implementation of
foreclosure and eviction protections authorized in recent legislation. This
report, one of several we plan to issue, focuses on rental housing from
2001 through 2017, the most recent year for which data were available at
the time of our review. Specifically, this report analyzes trends in (1) the
share of households that rent and their characteristics, (2) the affordability
of rental housing, and (3) rental housing conditions.

To describe trends in the share of households that rent, their
characteristics, and the affordability of rental housing, we analyzed the
Census Bureau’s American Community Survey data from 2001 through
2017, reviewed recent reports on rental housing, and obtained views from
a variety of stakeholders, including federal agency officials, academic
experts, research organizations, and industry groups. For renter
household characteristics, we analyzed data on renter household age,
race/ethnicity, and income at the national level and by locality type (that

1See GAO, Rental Housing Assistance: HUD Should Strengthen Physical Inspection of
Properties and Oversight of Lead Paint Hazards, GAO-20-277T (Washington, D.C.: Nov.
20, 2019) and Housing Choice Vouchers: Options Exist to Increase Program Efficiencies,
GAO- 12-300 (Washington, D.C.: Mar. 19, 2012).

Page 1 GAO-20-427 Rental Housing
is, areas we grouped by population density and growth rate).2 For
affordability, we analyzed data on household rent costs and household
incomes. Appendix | provides more detail on the data used in our
analysis. Consistent with other research and Department of Housing and
Urban Development (HUD) policies, we defined rent burden as spending
more than 30 percent of household income on rent and severe rent
burden as spending more than 50 percent of household income on rent.

To describe trends in rental housing conditions, we analyzed American
Community Survey data and American Housing Survey data from HUD
and Census, reviewed recent reports by federal agencies and research
organizations on rental housing conditions, and obtained views from
federal agency officials, academic experts, and research organizations.
We created two indexes to analyze American Housing Survey variables
related to housing conditions and identified trends at the national level
and for different types of localities. HUD and others have analyzed
aspects of housing quality from American Housing Survey data. Their
research helped inform our methodology for developing two indexes to
more specifically define the range of rental housing conditions. Appendix
Il provides more information about our indexes.

2We defined four head-of-household age categories as younger (20—34 years old), early
middle age (35—49 years old), late middle age (50-64 years old), and older (65 and older).
The Census Bureau defines race as a person’s self-identification with one or more social
groups. An individual can report as White, Black or African American, Asian, American
Indian and Alaska Native, Native Hawaiian and Other Pacific Islander, or some other race.
Survey respondents may report multiple races. Ethnicity determines whether a person is
of Hispanic origin or not and is broken out in two categories, Hispanic or Latino and Not
Hispanic or Latino. Hispanics may report as any race. For ease of analysis, we combined
some Census categories and reported on five race/ethnicity categories: White; Black;
Hispanic (an ethnicity that applies to individuals of any racial background); Asian (includes
Asian, Native Hawaiian, and Other Pacific Islander); and Other (includes American Indian,
Alaska Native, two or more races, and some other race). We defined five income
categories: extremely low income (up to 30 percent of HUD area median family income
(HAMFI)); very low income (more than 30, up to 50 percent of HAMFI); low income (more
than 50, up to 80 percent of HAMFI); moderate income (more than 80, up to 120 percent
of HAMFI) and higher income (greater than 120 percent of HAMFI).

3Rent burden is the share of household income used to pay gross rent, which includes
rent and utility costs.

4We determined that two indexes were needed to describe rental housing unit conditions
based on American Housing Survey data, as relevant variables fell into two categories
that required different statistical treatment and interpretation. We developed a quality
index based on a factor analysis of 13 quality-related variables and a completeness index
based on the presence of nine variables we identified as essential components of a
dwelling.

Page 2 GAO-20-427 Rental Housing
We assessed the reliability of the American Community Survey data and
American Housing Survey data by reviewing technical information for
each survey and interviewing HUD and Census Bureau officials. We
determined the data were sufficiently reliable for purposes of reporting at
the national level on some renter household characteristics and rental
housing conditions for 2001 through 2017, but we identified limitations to
reporting for smaller geographies. To address this limitation, we obtained
aggregated Census Bureau data from HUD’s Office of Policy
Development and Research. To assess the reliability of the data, we
analyzed the underlying programming code and related documentation
from agency officials and reviewed the data for missing elements,
outliers, and errors. We determined that the data were sufficiently reliable
for purposes of analyzing renter household characteristics, rent burden,
and rental housing conditions from 2001 through 2017 at the national
level and for different types of localities.

To describe common trends across similar types of localities for all of our
objectives, we developed metro area groupings based on population
change over our review period and population density, as of 2017. We
identified three growth categories (high, moderate, and negative) and
further categorized the moderate growth group by density (high and
moderate). We also identified a group of nonmetro areas consisting of all
counties in selected states that are outside the boundaries of any metro
area.® See table 1 fora summary of the different types of areas we used
for our analysis. Appendix | contains more information on our approach
for grouping metro areas.

5We analyzed American Community Survey data for core-based statistical areas by
population in 2000 and 2017 and by population density as of 2017. To identify the
nonmetro group, we analyzed 2017 American Community Survey data by county and
calculated nonmetro county population information for each state.

Page 3 GAO-20-427 Rental Housing
Table 1: Types of Localities by Population Growth 2000-2017 and Density, as of

2017

Locality type

Definition

Description

High-growth metro areas

Metro areas that grew in
population by more than 35
percent between 2000 and
2017

Fast-growing, moderate-
density metro areas where new
housing has been
comparatively less costly to
build than in other areas.

Moderate-growth/high-
density metro areas

Metro areas with population
growth of up to 35 percent
between 2000 and 2017,
with more than 950 people
per square mile

Growing, older, and dense
cities where new housing has
generally been costly to
develop.

Moderate-growth/
moderate-density metro
areas

Metro areas with population
growth of up to 35 percent
between 2000 and 2017,
with fewer than 950 people
per square mile

A diverse group of growing
metro areas with a wide range
of housing types, availability,
and cost.

Negative-growth metro
areas

Metro areas that lost
population or had no
population growth between
2000 and 2017

Generally older cities that may
have a surplus of existing,
older housing.

Nonmetro areas

Counties within each state
that are outside the
boundaries of any metro
area

Micropolitan areas, rural areas,
and small towns.

Source: GAO analysis of American Community Survey data. | GAO-20-427

We conducted this performance audit from February 2018 to May 2020 in
accordance with generally accepted government auditing standards.
Those standards require that we plan and perform the audit to obtain
sufficient, appropriate evidence to provide a reasonable basis for our
findings and conclusions based on our audit objectives. We believe that
the evidence obtained provides a reasonable basis for our findings and
conclusions based on our audit objectives.

Background

Federal policy for rental housing has traditionally focused on assisting
low-income households through rental assistance and incentives for the
development of housing with below-market rents. In fiscal year 2020,
Congress appropriated about $43.9 billion for HUD’s three largest federal
rental assistance programs: public housing, Housing Choice Vouchers,
and Project-Based Rental Assistance. These programs make rents
affordable to eligible households, generally by paying the difference
between the unit’s rent and 30 percent of a household’s adjusted income.

Page 4

GAO-20-427 Rental Housing
Renting Became
More Common after
the 2007—2009
Financial Crisis but
Varied by
Demographic Group
and Location

Unlike certain other means-tested programs, federal rental assistance
programs are not entitlements. The number of households that the
programs can assist is limited by the amount of budget authority that HUD
requests and Congress provides through the annual appropriations
process.§® Historically, appropriations for rental assistance programs have
not been sufficient to assist all households that HUD has identified as
having worst case housing needs—that is, renter households that (1)
have very low incomes; (2) do not receive housing assistance; and (3)
use more than one-half of their income to pay for rent, live in severely
inadequate conditions, or both.” In 2017, HUD reported that 8.3 million
households had worst case needs in 2015, an increase from 7.7 million in
2013. HUD reported that among very low-income renters in 2015, 25
percent of them received rental assistance, and an additional 43 percent
had worst case needs.

To determine program eligibility and identify populations in need of
assistance, many federal rental assistance programs have specific
income eligibility requirements. HUD sets income limits that determine
eligibility for its assistance programs based on median family income and
market rent estimates. These income limits can vary across different
types of localities.

Renting Expanded after
the Financial Crisis

The national rentership rate increased from 2001 through 2017 (see fig.
1). In 2004, the estimated rentership rate fell below 33 percent, the lowest

SBudget authority is the authority federal law provides to enter into financial obligations
that will result in immediate or future outlays involving federal government funds.

7These households have incomes no more than 50 percent of the area median income.

Department of Housing and Urban Development, Worst Case Housing Needs: 2017
Report to Congress (Washington, D.C.: August 2017).

Page 5 GAO-20-427 Rental Housing
in U.S. history, then climbed to 37 percent in 2013, a rate not seen since
the 1960s. By 2017, almost 7 million more households rented their homes
than in 2001, which brought the rentership rate to an estimated 36
percent.

Figure 1: Estimated Percentage of Households That Rent, 2001-2017

Percentage of households that rent
40

39
38
37

36 >.

35

34
33

32

31
4

7
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Year

Financial crisis
Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development. | GAO-20-427

Note: Estimates in this figure have a margin of error of +1 percentage point or less, at the 95 percent
confidence level.

This increase of 7 million households reflects both overall growth in the
population as well as the net shift from owning to renting. Many
households experienced lasting financial effects of the financial crisis,
such as impaired credit or loss of income, which hampered their ability to
enter into or transition back into homeownership. Although the national
foreclosure rate has slowed significantly in recent years, past research
has shown that most households struggle to return to homeownership

Page 6 GAO-20-427 Rental Housing
after foreclosure.8 Further, median home prices have risen faster than
median incomes nationally, which makes achieving homeownership more
challenging. Specifically, the gap between rising home prices and wage
growth has likely contributed to increases in rentership in many metro
areas.9

Renting Became More
Prevalent among Most
Age Groups, with Notable
Increases among Middle-
Aged Households

Nationally, the rentership rate increased from 2001 through 2017 across
all age categories we analyzed, except for older households (65 years or
older), as shown in figure 2. The greatest increase was among early
middle-aged households (35-49 years old), an estimated increase of
nearly 8 percentage points. In addition, rentership for late middle-aged
(50-64 years old) and younger (20—34 years old) households increased
by 5 percentage points.

8According to one study, from 2007 through 2017, only one-quarter of foreclosed
households were able to become homeowners again, taking an average of 4 years to do
so. See T. Piskorski and A. Seru, “Debt Relief and Slow Recovery: A Decade After
Lehman,” National Bureau of Economic Research Working Paper No. 25403 (2018). In
addition, the National Association of Realtors forecasted that less than one-third of
households who lost homes to foreclosure between 2006 and 2014 would return to
homeownership. See Ken Fears, “Return Buyers: Many Already Here, Many More to
Come,” Economists’ Outlook (National Association of Realtors: Apr. 17, 2015).

9For example, according to one study, renting a three-bedroom property was less
expensive than purchasing a median-priced home in 59 percent of U.S. metro areas
studied. See ATTOM Data Solutions. “Renting a Home More Affordable Than Buying in 59
Percent of U.S. Housing Markets.” January 8, 2019. Accessed October 4, 2019.
https://www.attomdata.com/news/most-recent/attom-data-solutions-201 9-rental-affordabilit
y-report/.

Page 7 GAO-20-427 Rental Housing
Figure 2: Estimated Rentership Rates by Age Group, 2001-2017

Percentage of households that rent
70

60
50

40

30 ee

20 =

10

0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Year

Younger (20-34 years old)
=== Early middle age (35-49 years old)
Mam Late middle age (50-64 years old)
====== Older (65 years and older)
Financial crisis

Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development. | GAO-20-427

Note: Estimates in this figure have a margin of error of +1 percentage point or less, at the 95 percent
confidence level.

Renters are, on average, older than they previously were. The late
middle-aged group (50-64 years) experienced the largest estimated
increase in the number of renter households—an increase of 4 million
households—and accounted for more than half of the total increase in
renter households from 2001 through 2017 (see fig. 3). Many of these
households have not recovered from the financial crisis, and this group
has lower incomes and higher rentership rates than in previous
generations, Harvard’s Joint Center for Housing Studies has reported. We
previously reported that the homeownership rate for the poorest older

Page 8 GAO-20-427 Rental Housing
households was significantly lower after the financial crisis than before

Figure 3: Estimated Number of Renter Households by Age Group, 2001 and 2017

Number of renter households (in millions)

15
12
9
6
3
0
Younger Early middle Late middle Older (65
(20-34 age (35-49 age (50-64 years and
years old) years old) years old) older)
Age group

Wy 2001
|| 2017

Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development. | GAO-20-427

Note: Estimates in this figure have relative margins of error of +2 percentage points or fewer, at the
95 percent confidence level.

10Specifically, we analyzed the estimated percentage of older households (those in which
the household head or any spouses or partners were aged 55 or older) that owned a
home from 1989 through 2016. The homeownership rate for households in the bottom
quintile in each year of our analysis was consistently much lower than for the other
quintiles—ranging between 18 and 32 percent. Further, the homeownership rate for
households in the bottom 20 percent was significantly lower in 2016 (19 percent) than in
2007 (28 percent). Differences in the percentage of households that owned a home from
2007 through 2016 were statistically significant at the 95 percent confidence level for the
bottom two quintiles of our analysis. See GAO, Retirement Security: Income and Wealth
Disparities Continue through Old Age, GAO-19-587 (Washington, D.C.: Aug. 9, 2019).

Page 9 GAO-20-427 Rental Housing
Renting Became More
Common among Black
Households and Declined
for Hispanic and Asian
Households

Black households had higher estimated rentership rates than White,
Hispanic, and Asian households, and rentership among Black households
increased from 54 percent in 2001 to 58 percent in 2017 (fig. 4). In
contrast, rentership among White households was lowest among the
race/ethnicity groups and remained generally stable during our analysis
period (ranging from 26 to 29 percent from 2001 through 2017). While
rentership among Hispanic and Asian households increased slightly in the
aftermath of the financial crisis, as of 2017, their rentership rates had
returned to levels below those of 2001, although these rates were still
higher than those of White households. "1

Figure 4: Estimated Rentership Rates by Race and Ethnicity, 2001-2017

Percentage of households that rent
60

50

45 So

==

35

30

=-
---—-

2001 2002 2003 2004 2005 2006

Year

Hispanic

---- Asian
Other
Financial crisis

2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development. | GAO-20-427

Note: Estimates in this figure have a margin of error of +3 percentage points or fewer, at the 95
percent confidence level.

11The estimated rentership rates for Black, Hispanic, and Asian households in 2017 were
all significantly higher than the rate for White households at the 95 percent confidence
level.

Page 10 GAO-20-427 Rental Housing
As of 2017, high-growth and moderate-growth/high-density metro areas
we analyzed tended to have more racially diverse renter populations than
other areas, and renters in these metro areas were mostly from minority
groups. For example, in Dallas, Texas, which is high-growth, an
estimated 59 percent of renter households were minority households, and
in Miami, Florida, which is moderate-growth/high-density, an estimated 75
percent of renter households were minority households. 12

Higher-Income Renter
Households Increased
Substantially after 2010

The most significant change in rentership from 2001 through 2017 by
income group was for higher-income households (more than 120 percent
of area median income), with the greatest change between 2010 and
2017. Nationally, higher-income households were the second smallest
renter group in 2001, with an estimated 6.6 million households, or 17
percent of renter households. In 2017, higher-income households were
the second largest renter group, with approximately 10.3 million
households, approximately 20 percent of renter households (see fig. 5).

12According to the Census Bureau, these estimates have a margin of error of +0.7
percentage points, at the 90 percent confidence level.

Page 11 GAO-20-427 Rental Housing
Figure 5: Estimated Number of Renter Households by Household Income, 2001-2017

Number of households that rent (in millions)

12

10

— = =

4
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Year
Mmmm =Extremely low income
— Very low income
ms mem Low income
eeeee Higher income

Financial crisis

Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development (HUD). | GAO-20-427

Notes: Estimates in this figure have a relative margin of error of +2 percentage points or fewer, at the
95 percent confidence level.

Extremely low income is defined as up to 30 percent of HUD area median family income (HAMFI);
very low income as more than 30, up to 50 percent of HAMFI; low income as more than 50, up to 80
percent of HAMFI; moderate income as more than 80, up to 120 percent of HAMFI, and higher
income as greater than 120 percent of HAMFI.

HUD revised the extremely low-income definition in 2014 to include all households below the federal
poverty threshold. This classified some very low-income households as extremely low-income.

Consistent with national trends, in all locality types—that is, those with
higher and lower population density or rates of growth—the estimated
number and proportion of higher-income renter households increased
from 2001 through 2017 (see fig. 6). The greatest increase occurred in
high- and moderate-growth metro areas. This trend could reflect (1) a
change in income, (2) relocation from moderate-growth/high-density
metro areas, and (3) consolidation of households—such as having
multiple roommates, extended families occupying one housing unit, or

Page 12 GAO-20-427 Rental Housing
households doubling up with relatives or others to make ends meet. 13
There were modest changes in the number and proportion of low-income
households during the same period. Rural areas and metro areas with
shrinking populations had the highest proportion of renter households
with low incomes as of 2017—for example, an estimated 63 percent of
renters in negative-growth metro areas had low to extremely low incomes.

Figure 6: Estimated Change in Renter Households by Income and Locality Type, 2001-2017

Change in renter households (in millions)
1.50

58% 60%

1.25

1.00

0.75

0.50

35% 13% 12%

28%
0.25

0.00
-0.25

-0.50

-0.75 -32%
Extremely low Very low Low income Moderate Higher income
income income income
High-growth
Moderate-growth/high-density

Moderate-growth/moderate-density

Negative-growth

Nonmetro

Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development (HUD). | GAO-20-427

Notes: We categorized locality types based on population growth and density. Estimates of the
percent change in renter households in this figure have a margin of error of +11 percentage points or
fewer at the 95 percent confidence level. The largest margins of error occur in the higher income and
nonmetro categories.

Bar height denotes the absolute change in renter households 2001-2017; data point at the end of
each bar denotes percent change in that group of renter households 2001-2017.

Extremely low income is defined as up to 30 percent of HUD area median family income (HAMFI);
very low income as more than 30, up to 50 percent of HAMFI; low income as more than 50, up to 80

13We were unable to identify changes in income or relocation information at the household
level from the aggregated Census Bureau data we analyzed.

Page 13 GAO-20-427 Rental Housing
Rent as Share of
Income Increased
from 2001 through
2017, with Serious
Consequences for the
Poorest

percent of HAMFI; moderate income as more than 80, up to 120 percent of HAMFI, and higher
income as greater than 120 percent of HAMFI.

HUD revised the extremely low-income definition in 2014 to include all households below the federal
poverty threshold. This classified some very low-income households as extremely low-income.

Population growth and two other factors appear to have contributed to the
growth in higher-income renter households. First, many homeowners who
experienced foreclosure during the financial crisis became renters.
Second, with rising housing costs, there has been a trend toward
consolidated households.'4 The share of households with three or more
adults was higher in 2017 than in 2001.15 Some of these households may
have chosen to combine as an alternative to eviction or homelessness,
and they may have overcrowded or unstable living arrangements. '6

The Percentage of Rent-
Burdened Households
Increased from 2001
through 2017

Most renter households paid a larger share of their income in rent in 2017
than in 2001. Federal housing policy generally considers rents at or below
30 percent of household income to be affordable, and households that
pay more than 30 percent of income in rent are considered to be rent
burdened. We found that by 2017, an estimated 48 percent of renter
households were rent burdened, 6 percentage points higher than in 2001.
Severe rent burden, where more than 50 percent of household income is
paid in rent, also became more common. Of the households that were

14For example, Harvard’s Joint Center for Housing Studies reported that the number of
shared renter households doubled from 4.6 million to 9.2 million between 1987 and 2017,
based on an analysis of Current Population Survey data. See Joint Center for Housing
Studies of Harvard University, The State of the Nation’s Housing (Cambridge, Mass.: Joint
Center for Housing Studies of Harvard University, 2018).

15Household income is determined by combining the incomes of all adults in a rental unit.
For example, a group of low-income adults living together could have a combined
household income in the higher-income category.

16For example, see Giselle Routhier, “Beyond Worst Case Needs: Measuring the Breadth

and Severity of Housing Insecurity Among Urban Renters,” Housing Policy Debate, vol.
29, no. 2 (2019): 235-249.

Page 14 GAO-20-427 Rental Housing
rent burdened in 2017, about half were severely rent burdened. These
households represented 24 percent of all renter households—an increase
of 4 percentage points from 2001 (see fig. 7).

Figure 7: Estimated Percentage of Renter Households by Rent Burden in 2001 and
2017

Percentage of renter households
100

90
80
52%
70 58%

60

50

40

30

20

10

2001 2017

Year

[ No rent burden — rent costs are at or below 30 percent of household income
i Moderate rent burden — rent costs are more than 30 and up to 50 percent of household income
| | Severe rent burden — rent costs exceed 50 percent of household income

Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development. | GAO-20-427

Note: Estimates in this figure have a margin of error of +2 percentage points or fewer, at the 95
percent confidence level.

The rising rent burden is part of a long-term trend in rental unaffordability,
as supply has not kept pace with demand for rental units. With fewer
affordable apartments available, rent burdens increased among lower-
income households, who were forced to spend a greater proportion of
income on rent. Government, academic, and industry research has
identified several factors that contribute to this trend:

e Local regulation and geography have long constrained where and

how much rental housing can be built. Cities have adopted zoning and
land use regulations that can prohibit or increase costs for new rental

Page 15 GAO-20-427 Rental Housing
units. Metro areas, particularly those in coastal or mountainous
regions, have limited available land for new housing.

Construction of new rental units has been limited since the 2007—
2009 financial crisis, in part because developers struggled to rebuild
workforce capacity after layoffs of skilled construction workers. As a
result, since 2009, the construction industry has focused on building
luxury apartments, which have higher profit margins, and produce few
units affordable to lower-income households. Conversion of lower-rent
units to higher-rent units through renovation also reduced the number
and share of rental units affordable to lower-income households.

Demographic changes, particularly the aging of the millennial and
baby boomer generations, have increased demand for rental units. As
previously discussed, we found that renters were, on average, older in
2017 than in 2001. In addition, Harvard’s Joint Center for Housing
Studies has reported that late middle-age renters (50—64 years) have
lower incomes and higher rentership rates than previous generations.
Populations with higher rentership rates—including minority
households—are forecasted to continue growing through 2030.17

The spike in foreclosures during the financial crisis resulted in millions
of households entering the rental market, increasing competition for
available units. Tighter credit standards after the financial crisis have
kept many of those who lost their home due to foreclosure from
qualifying for a new mortgage.

In the United States, rent burden has been most common among
minorities and older adults and in dense metropolitan areas (see fig. 8):

Rent burden was about 10 percentage points more common among
Black and Hispanic households than White households in 2017. This
disparity was due to sizable differences in median income. In 2017,
estimated median income was $63,704 for White households,
$49,793 for Hispanic households, and $40,232 for Black households.

Rent burden was more than 10 percentage points more common
among older adult (65 and over) households than working-age (20-—
64) households in 2017. This disparity was also due to sizable
differences in median income, as older adults were less likely to be in

17Laurie Goodman, “The Demographics of Demand,” Mortgage Banking (October 2015).

Page 16 GAO-20-427 Rental Housing
the workforce. In 2017, median income was $69,459 for households
age 25-64 and $43,735 for households age 65 and over. ‘8

e Rent burden was nearly 10 percentage points more common among
renters in high-density metro areas than in nonmetro areas in 2017.
According to the Urban Institute, the shortage of affordable rental
housing was more acute in urban areas than rural areas in 2014.19

See appendixes III and IV for more detailed information on rent burden by
age, race/ethnicity, and locality type.

18According to the Census Bureau, household income includes not just earnings for each
household member but also income from Social Security, interest, dividends, and many
other sources.

19|n 2014, rural counties had 69 adequate and affordable units available for every 100
extremely low-income renters, compared with 42 units in metropolitan counties. See Liza
Getsinger, Lily Posey, Graham MacDonald, Josh Leopold, and Katya Abazajian, The
Housing Affordability Gap for Extremely Low-Income Renters in 2014 (Urban Institute,
April 2017).

Page 17 GAO-20-427 Rental Housing
Figure 8: Rent Burden among Selected Demographic Groups and Locality Types in

2017
Estimated percentage of renter households
100

90

80 41%

56% 45% 46% 549 48%

70 57%

60

50

40

30

20

32% 27%
10 ° 21%
0
White Black Hispanic Working Older Moderate- Nonmetro
age (65 years growth/
(20-64 and high-
years old) older) density

Demographic or locality group

[ No rent burden — rent costs are at or below 30 percent of household income
i Moderate rent burden — rent costs are more than 30 and up to 50 percent of household income
7 Severe rent burden — rent costs exceed 50 percent of household income

Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development. | GAO-20-427

Note: Estimates in this figure have a margin of error of +2 percentage points or fewer, at the 95
percent confidence level.

Lower-Income Households In 2017, moderate and severe rent burdens were common among low- to
Commonly Experienced extremely-low income households and relatively rare among moderate- to

High Rent Burdens from higher-income households (see fig. 9).
2001 through 2017

Page 18 GAO-20-427 Rental Housing
Figure 9: Estimated Percentage of Renter Households with Rent Burdens by
Income in 2017
Percentage of renter households

100
90
80
70
60
50
40
30
20
10

0

94%

Extremely Very low Low Moderate Higher
low income income income’ income
income
[ No rent burden — rent costs are at or below 30 percent of household income
ia Moderate rent burden — rent costs are more than 30 and up to 50 percent of household income

| | Severe rent burden — rent costs exceed 50 percent of household income

Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development. | GAO-20-427

Note: Estimates in this figure have a margin of error of +2 percentage points or fewer, at the 95
percent confidence level.

From 2001 through 2017, the estimated number of renters with moderate
or severe rent burdens increased across all income levels, but the
increase was more pronounced among lower-income groups (see fig. 10).
Specifically, we found the following:

The estimated number of higher-income renters increased by more
than 3.6 million households from 2001 through 2017, but relatively few
of these households experienced rent burden. In contrast, the
numbers of low-income, very low-income, and extremely-low income
renters also increased over this period, and these groups saw
significant increases in rent burden.

In more recent years, the estimated number of extremely low-income
renter households with severe burden actually decreased—from 7.4
million in 2011 to 6.6 million in 2017. This decrease, however, does
not necessarily indicate improved conditions for these households
because it was not accompanied by a corresponding increase in
either (1) the number of extremely low-income households that were
less burdened or (2) the number of very low-income households (the
next highest income group). An increase in either of these groups
could indicate that the poorest, most burdened households

Page 19 GAO-20-427 Rental Housing
experienced either an increase in income or a decrease in rent
burden. However, because these other groups did not increase, it is
possible that some of these extremely low-income, severely burdened
households moved in with other households or experienced some
other form of homelessness.

Page 20 GAO-20-427 Rental Housing
Figure 10: Estimated Number of Renter Households by Rent Burden and Household Income, 2001-2017

Year
2017
2015
2013
2011
2009
2007
2005
2003
2001
2017
2015
2013
2011
2009
2007
2005
2003
2001
2017
2015
2013
2011
2009
t 2005

t 2003

i 2001
t 2017
[ 2015
Moderate [ 2013
income i 2011
i 2009
i 2007
i 2005
i 2003

t 2001
t 2017
i 2015
[ 2013

Higher L 2009
income t 2007
t 2005
t 2003

t 2001

10 8 6 4 2 0 2 4 6 8 10
Non-rent burdened households (in millions) Rent burdened households (in millions)
Source: GAO analysis of American Community Survey data from the Department of Housing and Urban Development. | GAO-20-427

Note: Estimated numbers of rental households in this figure include estimates for each year from
2001 through 2017. The estimates have a relative margin of error of +16 percent or less, at the 95

Extremely
low
income

Very low
income

Low
income

il

No rent burden — rent costs are at or below 30
percent of household income

Moderate rent burden — rent costs are more than 30
and up to 50 percent of household income

Severe rent burden — rent costs exceed 50 percent
of household income

Page 21 GAO-20-427 Rental Housing
percent confidence level, with the exception of higher-income rent-burdened households, which have
a relative margin of error of +32 percent or less due to small sample size.

Rent burdens affect households differently depending on their income.
Households with lower incomes may pay the same percentage of income
in rent as moderate- or higher-income households but have less income
left over for other necessities. Even relatively inexpensive units may not
leave enough money for lower-income households to cover other
necessities like food, clothing, transportation, or medical care. These
households may also be sensitive to shocks, such as job loss and health
emergencies, and may be at heightened risk of eviction and
homelessness.

Challenges that lower-income households face can vary across cities and
regions due to differences in local market rents and incomes. For
example, as figure 11 shows, in the San Francisco area in 2017 a very
low-income family of four would experience a severe rent burden if it paid
the fair market rent for a two-bedroom apartment ($3,018 per month). 2°
Such a family would struggle to pay the rent and afford other necessities
even with two or three full-time minimum wage jobs.21 In contrast, a very
low-income family of four in the St. Louis area in 2017 would experience a
moderate or no rent burden if it paid the fair market rent for a two-
bedroom apartment ($896). Such a family with at least two full-time
minimum wage jobs would have relatively more money left over for other
necessities.22 See appendixes III and IV for more detailed information on
rent burden by household income.

20F air market rents are calculated annually by HUD and generally defined as the 40th
percentile gross rent in a local area—typically a metropolitan area or smaller area defined
by HUD. Gross rent includes the cost of rent and utilities.

21We defined the 2017 hourly minimum wage in San Francisco as the city-level minimum
wage for San Francisco. We calculated annual income for a minimum wage job assuming
a 40-hour work week for 52 weeks per year.

22We defined the 2017 hourly minimum wage in St. Louis as the state-level minimum

wage for Missouri. We calculated annual income for a minimum wage job assuming a 40
hour work week for 52 weeks per year.

Page 22 GAO-20-427 Rental Housing
Figure 11: Rent Burden for Families of Four Who Paid Fair Market Rent in the St.
Louis and San Francisco Areas, 2017

Household income (dollars)

120,000
Between 3
minimum wage
jobs and median
1 00,000 household wage
Low income
80,000
Between 2 and 3
minimum wage
Between 3 jobs
minimum wage
jobs and i
60,000 household wage
Very low income
Low income Between 1 and 2
Between 2 and3 minimum wage
minimum wage jobs
40,000 Hees
Very low income
age Extremely low
20,000 ob income
0
St. Louis San Francisco
Moderate rent burden — rent costs are more than 30 and up to 50 percent of household

income

|| Severe rent burden — rent costs exceed 50 percent of household income

Source: GAO analysis of Census and Department of Housing and Urban Development (HUD) data. | GAO-20-427

Notes: We define median household wage for St. Louis and San Francisco as the HUD fiscal year
2017 estimated median family income for those metro areas, which are calculated based on 2014
American Community Survey 5-year estimates. Fair market rents are calculated annually by HUD and
generally defined as the 40th percentile gross rent in a local area—typically a metro area or smaller
area defined by HUD. Gross rent includes the cost of rent and utilities. We defined the 2017 hourly
minimum wage in St. Louis as the state-level minimum wage for Missouri and the hourly minimum
wage in San Francisco as the city-level minimum wage for San Francisco. We calculated annual
income for a minimum wage job assuming a 40-hour work week for 52 weeks per year.

For moderate-income households, the consequences of rent burden are
less dire than for lower-income households, but they are still significant.
For example, a family of four earning the median income in San
Francisco that paid fair market rent for a two-bedroom apartment would
be rent burdened. A housing unit that would be considered affordable to
them would cost at least $135 per month below fair market rent (or
approximately $2,882 or less). Money that a family could save on a unit
below fair market rate could help reduce household debt, add to
retirement savings, or pay for necessities like child care. Rent burden

Page 23 GAO-20-427 Rental Housing
among moderate-income households tends to be more common in large
cities with strong economies and significant geographic and regulatory
constraints on new housing, such as San Francisco and New York.

The lowest-income households face challenges securing affordable rental
units. There are not enough rental units that are affordable to the lowest-
income households without rental assistance. Specifically, according to
HUD, lower-income households face competition from moderate- or high-
income households to rent affordable units.23 HUD’s analysis showed that
although there were enough affordable units nationwide to house 66
percent of extremely-low income renters in 2015, 43 percent of those
units were occupied by renters with higher incomes.

We also found that for all income groups, rents rose faster than incomes
and therefore became less affordable to varying degrees. Specifically,
estimated median rent-to-income ratios, which indicate the median
proportion of income devoted to rent, generally increased from 2001
through 2017, according to our analysis (see fig. 12). For the lowest-
income households, even small declines in affordability have a big impact
because these households face the highest rent burdens and have the
fewest options in the housing market. See appendix III for more detailed
information on rent-to-income ratios.

23See Department of Housing and Urban Development, Worst Case Housing Needs 2017
Report to Congress.

Page 24 GAO-20-427 Rental Housing
About 15 Percent of
Rental Units Had
Serious Deficiencies
in 2017

Figure 12: Estimated Median Rent-to-Income Ratio in 2001 and 2017

Median rent-to-income ratio

0.8
0.6
0.4
Extremely Very low Low Moderate Higher
low income income income income
income

Household income

Source: : GAO analysis of American Community Survey data from the Department of Housing and Urban
Development (HUD). | GAO-20-427

Notes: Estimates in this figure have a margin of error of +1 percentage point or less, at the 95 percent
confidence level.

Extremely low income is defined as up to 30 percent of HUD area median family income (HAMFI);
very low income as more than 30, up to 50 percent of HAMFI; low income as more than 50, up to 80
percent of HAMFI; moderate income as more than 80, up to 120 percent of HAMFI, and higher
income as greater than 120 percent of HAMFI.

Based on two indexes we created to analyze rental housing conditions
using American Housing Survey data, we found that an estimated 15
percent of renter households—more than 5 million—lived in units with

Page 25 GAO-20-427 Rental Housing
serious deficiencies in 2017.24 Specifically, an estimated 12 percent of
renter households (more than 4 million households) lived in units with
substantial quality issues. These units typically had a combination of
issues, such as cracked walls and the presence of rodents, or multiple
heating problems and the presence of rodents.25 An additional 3 percent
of renter households (more than 1 million households) lived in incomplete
units—that is, units lacking essential components of a dwelling (Such as
heating equipment or hot and cold running water). Further, an estimated
28 percent of households—nearly 10 million—rented units with less
substantial quality issues. Table 2 presents these findings and how our
indexes described different types of rental housing conditions.

24VWWe determined that two indexes were needed to describe rental housing unit conditions
based on American Housing Survey data, as relevant variables fell into two categories
that required different statistical treatment and interpretation. We developed a quality
index based on a factor analysis of 13 quality-related variables, and we developed a
completeness index based on the presence of nine variables we identified as essential
attributes of a rental unit. App. Il includes more detailed information on index variables and
our methodology. In addition, app. V describes how our indexes compare to an index that
HUD uses to measure housing adequacy.

25To avoid double-counting, we excluded incomplete units from our analysis of units in
2017 with substantial quality issues. HUD has noted that a portion of severe physical
inadequacies reported in the survey likely result from or reflect maintenance or upgrade
activity occurring in occupied housing units. App. II includes more detailed information on
index variables and our methodology.

Page 26 GAO-20-427 Rental Housing
Table 2: Condition of Rental Housing Units Based on GAO’s Quality Indexes, 2017

Category

Description

Estimated rental units

Number Percentage
(in millions)

Incomplete units

Rental units that lacked one or more essential components of a
dwelling: electricity; heating equipment; bathroom sink; toilet; tub or
shower; kitchen sink; refrigerator; cooking appliance; or hot and cold
running water.

1.06 3

Substantial quality
issues

Rental units that were complete but had at least one deficiency among
13 quality-related variables and scored above our modeled statistical
threshold for substantial quality issues. The most common profiles
were the presence of (1) cracked walls and rodents, (2) uncomfortably
cold periods, heating equipment breakdowns, and rodents, or (3)
cracked walls and water leaks.

4.34 12

Less substantial quality
issues

Rental units that were complete but had one or several deficiencies
among the quality-related variables and scored below our modeled
statistical threshold for substantial quality issues. The most common
profiles were (1) the presence of rodents, (2) the presence of water
leaks, or (3) instances of blown fuses or tripped circuit breakers in the
last 3 months.

9.88 28

No quality issues

Rental units that were complete and without any deficiencies among
the 13 quality-related variables we identified.

20.41 57

Source: GAO analysis of American Housing Survey data. | GAO-20-427
Notes: Estimates of the number of housing units in this table have a relative margin of error of +6

percent of the estimated number. Estimates of the percentage of housing units in this table have a
margin of error of +2 percentage points or fewer, at the 95 percent confidence level.

We refer to the combination of issues affecting units with substantial or less substantial quality issues

as profiles.

The proportion of rental units with the three types of deficiencies—
substantial quality issues, less substantial quality issues, and absence of
essential components of a dwelling—generally remained stable from
2001 through 2017 (see fig. 13). The proportion of rental units that had at
least one of these deficiencies ranged from an estimated 39 to 47 percent

from 2001 through 2017.

Page 27

GAO-20-427 Rental Housing
Figure 13: GAO Quality and Completeness Indexes, Estimated Percentage of Rental
Units by Housing Condition, 2001—2017

Estimated percentage of rental units
70

. ~

50
40

30 pee

20

10

2001 2003 2005 2007 2009 2011 2013 2015 2017
Year

MEE No quality issues

=a ma == Less substantial quality issues

eeeeeee Substantial quality issues

---- Incomplete housing units
Source: GAO analysis of American Housing Survey data. | GAO-20-427
Notes: We developed two indexes to describe the range of housing conditions based on American
Housing Survey data. The quality index, based on 13 quality-related variables, identified units as
having no quality issues, less substantial quality issues, and substantial quality issues. The

completeness index, based on nine variables we identified as essential components of a dwelling,
identified units as being incomplete or complete.

Estimates in this figure have a margin of error of +3 percentage points or fewer, at the 95 percent
confidence level.

Page 28 GAO-20-427 Rental Housing
Serious Deficiencies More
Often Affected Lower-
Income and Rent-
Burdened Households

We analyzed rental housing conditions by renter household and rental
unit characteristics.26 Households with low incomes (those with low, very
low, or extremely low incomes) or with rent burdens comprised half or
more of renters living in units with substantial quality issues and
incomplete housing units (those lacking essential components of a
dwelling). Although incomplete housing units represented a small
percentage of rental units overall (about 3 percent), there were more than
an estimated 1 million such units in 2017.

Low-income renters have fewer affordable options and, as a result, may
end up in units with deficiencies out of necessity. Households with low,
very low, or extremely low incomes represented an estimated 62 percent
of renters overall in 2017. These households occupied an estimated 67
percent of units that had substantial quality issues and nearly 80 percent
of incomplete units. Similarly, rent-burdened households represented an
estimated 50 percent of renters overall in 2017 and occupied an
estimated 53 percent of units with substantial quality issues and 60
percent of incomplete units.

There were some notable differences in housing conditions by age and
race/ethnicity. Older households (65 and older) were the most likely age
group to live in rental units with no deficiencies in 2017.27 About half of
renting households were White in 2017, and White households comprised
the largest share of renters in each quality or completeness category we
analyzed. The proportions of Hispanic and Asian households that rented
incomplete units (estimated at 31 percent and 11 percent, respectively)
were higher than the overall proportions of Hispanic and Asian renter
households (estimated at 20 percent and 6 percent, respectively). In
addition, the proportion of Black households that rented units with
substantial quality issues (estimated at 24 percent) was slightly higher
than the overall proportion of Black renter households (estimated at 21
percent).

26With our indexes, we analyzed the following renter household characteristics: household
income and affordability (rent burden), age, and race/ethnicity. In addition, based on an
analysis of American Community Survey data, we analyzed household crowding, another
renter household characteristic. With our indexes, we analyzed the following rental unit
characteristics: unit age and location (locality type) and type of unit—that is, multifamily (2
~9 unit buildings and buildings with 10 or more units), single-family (detached and
attached), or mobile home.

27We did not include accessibility features (such as grab bars in showers and wheelchair-
accessible entrances) in our analysis, as available data were limited.

Page 29 GAO-20-427 Rental Housing
Older and Single-Family
Rental Units Were More
Likely to Have
Deficiencies

Rental housing conditions by unit age or type were generally consistent
from 2001 through 2017. As expected, units built after 2000 had fewer
deficiencies than those built before. Older rental housing—units built prior
to 1980—were more likely to have substantial quality issues than those
built after. An estimated 63 percent of units in high-growth metro areas
had no quality issues as of 2017, compared to an estimated 55 to 57
percent of units in other types of localities. There was little other variation
in housing conditions by locality type.

We also found that detached single-family homes and mobile homes
were somewhat more likely to have serious deficiencies than multifamily
units.28 The proportion of units with these deficiencies remained relatively
steady from 2001 through 2017. One reason for this is single-family units
lack on-site building managers and other benefits of shared maintenance
that multifamily units may provide. Some researchers and industry
participants have noted possible maintenance challenges for a growing
number of investor-owners of single-family rentals that manage
thousands of properties of varying size, age, and condition.29 From 2001
through 2017, the proportion of single-family units with serious
deficiencies (rental units lacking essential components of a dwelling or
units with substantial quality issues) ranged from around 13 to 20 percent
(see fig. 14).8° During the same period, the proportion of single-family
units with less substantial quality issues ranged from an estimated 28 to
34 percent.

28In 2017, single-family rental units represented an estimated 37 percent of all rental units
(more than 13 million units), while multifamily units represented an estimated 60 percent
of all units (more than 21 million units). Mobile homes represented an estimated 4 percent
(fewer than 1.3 million units). Some differences in quality may be overstated due to
American Housing Survey sampling error.

29See Emily Badger, “Would You Rather Pay Your Rent to a Bum Landlord or a Wall
Street Investor?” CityLab, accessed September 23, 2013,
https://www.citylab.com/life/2013/09/would-you-rather-pay-your-rent-bum-landlord-or-wall-
street-investor/6985/; William Cohan, “Why Wall Street Loves Houses Again,” The Atlantic
(October 2013); Alana Semuels, “When Wall Street Is Your Landlord,” The Atlantic
(February 2019).

30As previously mentioned, we excluded incomplete units from our analysis of units with
substantial quality issues to avoid double-counting.

Page 30 GAO-20-427 Rental Housing
Figure 14: GAO Quality and Completeness Indexes, Estimated Percentage of Rental
Units with Serious Deficiencies by Housing Type, 2001—2017

Percentage
25

20

10
5
0
2001 2003 2005 2007 2009 2011 2013 2015 2017

Year

m = &@ Multifamily rental units

Source: GAO analysis of American Housing Survey data. | GAO-20-427

Notes: We developed two indexes to define the range of housing conditions based on American
Housing Survey data. The quality index, based on 13 quality-related variables, identified units as
having no quality issues, less substantial quality issues, and substantial quality issues. The
completeness index, based on nine variables we identified as essential components of a dwelling,
identified units as being incomplete or complete. Units with serious deficiencies lacked essential
components or had substantial quality issues.

Estimates in this figure have a margin of error of +4 percentage points or fewer, at the 95 percent
confidence level.

We also analyzed household crowding trends based on American
Community Survey data. We defined crowded households as those
having more than two people per bedroom.31 From 2001 through 2017,
the incidence of renter household crowding decreased, with the greatest
percentage point declines for Hispanic households prior to the housing
crisis. Generally, households that were younger, Hispanic or Asian, or
had lower incomes were more likely to experience crowding. In addition,
crowded households were more common in high-density and high-growth

31A studio apartment occupied by two people did not meet our crowding definition.

Page 31 GAO-20-427 Rental Housing
Agency Comments

metro areas. Appendix VI includes information on household crowding by
race/ethnicity, age, household income, and locality type.

We provided a draft of this report to HUD for review and comment. HUD
officials told us that they had no comments on the draft report.

As agreed with your offices, unless you publicly announce the contents of
this report earlier, we plan no further distribution until 30 days from the
report date. At that time, we will send copies of this report to the
appropriate congressional committees, the Secretary of Housing and
Urban Development, and other interested parties. In addition, the report
will be available at no charge on the GAO website at http://Awww.gao.gov.

If you or your staff have any questions about this report, please contact
me at (202) 512-4529 or garciadiazd@gao.gov. Contact points for our
Offices of Congressional Relations and Public Affairs may be found on
the last page of this report. GAO staff who made key contributions to this
report are listed in appendix VII.

[Dds

Daniel Garcia-Diaz
Managing Director, Financial Markets and Community Investment

Page 32 GAO-20-427 Rental Housing
List of Requesters

The Honorable Susan M. Collins

Chairman

Subcommittee on Transportation, Housing and Urban Development, and
Related Agencies

Committee on Appropriations

United States Senate

The Honorable Michael F. Bennet
United States Senate

The Honorable Christopher A. Coons
United States Senate

The Honorable Lindsey O. Graham
United States Senate

The Honorable Tim Scott
United States Senate

Page 33 GAO-20-427 Rental Housing
Appendix |: Objectives, Scope, and

Methodology

The objectives of this report were to analyze trends in (1) the share of
households that rent and their characteristics, (2) the affordability of rental
housing, and (3) rental housing conditions.

Data Used in Our Analysis

We analyzed 2001-2017 data from the American Community Survey and
American Housing Survey to describe renter household characteristics,
rent affordability, rental housing conditions, and trends at the national
level and across different types of localities.

e The American Community Survey is an ongoing survey administered
by the Census Bureau of around 3.5 million households across the
United States; the data we used in our analysis were current as of
2017, the most recently available data at the time of our review. 32

e The survey collects data on the economic, social, housing, and
demographic characteristics of communities at various geographic
levels, including metropolitan areas, states, and counties.

e The American Housing Survey is a biennial survey sponsored by the
Department of Housing and Urban Development (HUD) and
administered by the Census Bureau that collects a range of housing
information, including the size and composition of the U.S. housing
inventory, physical condition of housing units, characteristics of
occupants, and other information.

Findings from each survey are subject to sampling errors. To assess the
reliability of the data, we reviewed technical information for each survey.
In addition, we interviewed HUD and Census Bureau officials to identify
differences across survey years and understand geographic limitations of
publicly available data. We determined that the surveys were sufficiently
reliable for purposes of reporting at the national level on renter household
characteristics. However, we determined that additional, nonpublic data
were needed from each survey to analyze renter household
characteristics, rent burden, and rental housing conditions for smaller
geographic units.

32The American Community Survey uses a series of monthly samples to produce annually
updated estimates for census tracts across the United States. The survey sample initially
included 800,000 addresses during a demonstration phase starting in 2000 and expanded
to nearly 3 million addresses after being fully implemented in 2005, and it had increased to
more than 3.5 million addresses as of 2017. In recent years, the Census Bureau has
interviewed around 2.2 million people for the survey annually through four different
methods: internet, mail, telephone, and personal visit.

Page 34 GAO-20-427 Rental Housing
Appendix I: Objectives, Scope, and
Methodology

To address this limitation, staff from HUD’s Office of Policy Development
and Research provided us with aggregated Census Bureau data. To
assess the reliability of these data, we analyzed the underlying
programming code and related documentation from agency officials and
reviewed for missing data, outliers, and errors. We determined that the
data were sufficiently reliable for purposes of analyzing renter household
characteristics, rent burden, and rental housing conditions from 2001
through 2017 at the national level and for different types of localities.

Locality Types

For all objectives, to describe common trends and differences across
localities—that is, localities with different population growth rates and
densities—we developed metro area groupings. The groupings provide a
general framework for describing metro areas that experienced varying
degrees of population growth from 2000 through 2017 and how trends in
renter household characteristics, rent affordability, and rental housing
conditions compared to trends in other types of areas.

To identify the locality types, we analyzed core-based statistical areas by
population growth from 2000 through 2017 and population density as of
2017.33 We identified three growth categories (high, moderate, and
negative) and further categorized the moderate growth group by density
(high and moderate). We also identified a group of nonmetro areas
consisting of all counties in each state that are outside the boundaries of
any metro area.*4 These areas included micropolitan areas, small towns,
and low-density rural areas.35 The five locality types were high-growth
metro areas, moderate-growth/high-density metro areas, moderate-

33Core-based statistical area refers collectively to metropolitan statistical areas and
micropolitan statistical areas, which are geographic entities that the Office of Management
and Budget defines for use by federal statistical agencies. A core-based statistical area
consists of the county or counties (or equivalent entities) associated with at least one core
(urbanized area or urban cluster) with a population of at least 10,000, plus adjacent
counties having a high degree of social and economic integration with the core as
measured through commuting ties.

34To identify the nonmetro group, we analyzed 2017 American Community Survey data by
county and calculated nonmetro county population information for each state.

35Micropolitan areas consist of the county or counties (or equivalent entities) associated
with at least one urban cluster with a population of at least 10,000 but less than 50,000,
plus adjacent counties having a high degree of social and economic integration with the
core as measured through commuting ties.

Page 35 GAO-20-427 Rental Housing
Appendix I: Objectives, Scope, and
Methodology

growth/moderate-density metro areas, negative-growth metro areas, and
nonmetro areas.

Renter Household
Characteristics

To describe trends in the share of households that rent and their
characteristics, we analyzed American Community Survey data from
2001 through 2017 at the national level and across different types of
localities, with a focus on renter household age, race/ethnicity, and
income.

We defined four head-of-household age categories: younger (20-34
years old), early middle age (35—49 years old), late middle age (50—64
years old), and older (65 years and older).3° We reported on five
race/ethnicity categories, combining some Census categories for our
analysis: White, Black, Hispanic (an ethnicity that applies to individuals of
any racial background), Asian (includes Asian, Native Hawaiian, and
Other Pacific Islander), and Other (includes American Indian, Alaska
Native, two or more races, and some other race).3” We defined five
income categories based on income ranges that HUD uses for
determining rental assistance eligibility or reporting to Congress on worst
case needs: extremely low income (up to 30 percent of HUD area median
family income (HAMFI)); very low income (more than 30, up to 50 percent
of HAMFI); low income (more than 50, up to 80 percent of HAMFI);
moderate income (more than 80, up to 120 percent of HAMFI) and higher
income (greater than 120 percent of HAMFI).38

36Our national-level analyses of renter household characteristics, rent affordability, and
rental housing conditions include renter households age 19 or younger in the total number
of renter households, but we did not analyze trends specific to that age group.

37The Census Bureau defines race as a person’s self-identification with one or more social
groups. An individual can report as White, Black or African American, Asian, American
Indian and Alaska Native, Native Hawaiian and Other Pacific Islander, or some other race.
Survey respondents may report multiple races. Ethnicity determines whether a person is
of Hispanic origin or not and is broken out in two categories, Hispanic or Latino and Not
Hispanic or Latino. Hispanics may report as any race.

38HUD sets income limits that determine eligibility for various rental assistance programs
(extremely low income, very low income, and low income). The agency develops income
limits based on median family income estimates and fair market rent area definitions for
each metropolitan area, parts of some metropolitan areas, and each nonmetropolitan
county. Since 2014, as a result of a legislative change, HUD has used for several of its
housing programs a definition of “extremely low income” that does not exceed the higher
of Department of Health and Human Services poverty guidelines or 30 percent of area
median income.

Page 36 GAO-20-427 Rental Housing
Appendix I: Objectives, Scope, and
Methodology

Rent Affordability

To describe trends in the affordability of rental housing, we analyzed
American Community Survey data on gross rent as a percentage of
household income from 2001 through 2017 at the national level and
across different types of localities. Consistent with other housing research
and HUD policies, we defined rent burden as spending more than 30
percent of household income on rent, moderate rent burden as spending
more than 30 and up to 50 percent of household income on rent, and
severe rent burden as spending more than 50 percent of household
income on rent.°9 Further, as described in appendix IV, we developed a
supplementary analysis of rental housing affordability for rural areas by
state.40

Rental Housing Conditions

To describe trends in rental housing conditions, we analyzed data from
the American Community Survey and American Housing Survey from
2001 through 2017 at the national level and across different types of
localities.41 HUD designed the American Housing Survey to include
indicators of housing quality. HUD analyzes and reports periodically on a
housing adequacy measure as part of its worst case housing needs
assessments for Congress.42 HUD’s adequacy measure and related
research informed our methodology for developing two indexes to

39Rent burden is the share of household income used to pay gross rent, which includes
rent and utility costs.

40There is no single agreed-upon definition of rural areas and, for purposes of this report,
we defined rural areas based on the Department of Agriculture’s 2010 rural-urban
commuting area codes 7, 8, 9, 10, and 10.3 (as revised July 3, 2019). We obtained renter
household data for these areas from the American Community Survey’s 5-year estimates
for 2013-2017.

41HUD revised some American Housing Survey variables and expanded the survey
sample in 2015.

42HUD’s adequacy index categorizes housing units as severely inadequate, moderately
inadequate, or adequate based on whether a surveyed housing unit meets certain
conditions or criteria. The conditions and criteria are related to the presence of bathroom
and kitchen facilities, whether a unit has heat in the winter, and the presence of other
hazards to health and well-being, such as water leaks, wall cracks, and rodents.

Page 37 GAO-20-427 Rental Housing
Appendix I: Objectives, Scope, and
Methodology

analyze rental housing conditions.4s We developed the indexes to more
specifically define the range of housing conditions. The two indexes
include 13 quality-related variables and nine variables we identified as
essential components of a dwelling from the American Housing Survey,
described in table 3.44 Appendix II includes more detailed information
about our methodology. Appendix V includes information on the
similarities and differences between HUD’s adequacy index and the
indexes we developed for this report.

Table 3: GAO Rental Housing Conditions Indexes

Category Description

Incomplete units Rental units that lacked one or more essential components of a dwelling: electricity; heating equipment;
bathroom sink; toilet; tub or shower; kitchen sink; refrigerator; cooking appliance; or hot and cold running
water.

Substantial quality Rental units that were complete but had at least one deficiency among 13 quality-related variables and

issues scored above our modeled statistical threshold for substantial quality issues. The most common profiles

were the presence of (1) cracked walls and rodents, (2) uncomfortably cold periods, heating equipment
breakdowns, and rodents, or (3) cracked walls and water leaks.

Less substantial quality Rental units that were complete but had one or several deficiencies among the quality-related variables and

issues scored below our modeled statistical threshold for substantial quality issues. The most common profiles
were (1) the presence of rodents, (2) the presence of water leaks, or (3) instances of blown fuses or tripped
circuit breakers in the last 3 months.

No quality issues Rental units that were complete and were without any deficiencies among the 13 quality-related variables we
identified.

Source: GAO analysis of American Housing Survey data. | GAO-20-427

Note: We refer to the combination of issues affecting units with substantial or less substantial quality
issues as profiles.

43Researchers have analyzed HUD’s adequacy index and tested several other indexes for
measuring housing quality. For example, see F.J. Eggers and F. Moumen, American
Housing Survey: A Measure of (Poor) Housing Quality, (Bethesda, MD: Econometrica,
Inc., 2013); F.J. Eggers and F. Moumen, American Housing Survey: Housing Adequacy
and Quality As Measured by the AHS (Bethesda, MD: Econometrica, Inc., 2013); and
Sandra Newman and C. Scott Holupka, The Quality of America’s Assisted Housing Stock:
Analysis of the 2011 and 2013 American Housing Surveys, a report prepared for the
Department of Housing and Urban Development (Falls Church, VA: Economic Systems
Inc., 2017). There is no agreed-upon definition of housing quality. Based on limitations of
other types of housing quality indexes, such as challenges reporting trends over time and
subjective scoring techniques, we developed our own statistical analysis of American
Housing Survey variables to analyze rental housing conditions.

44We determined that two indexes were needed to describe rental housing unit conditions

based on American Housing Survey data, as relevant variables fell into two categories
that required different statistical treatment and interpretation.

Page 38 GAO-20-427 Rental Housing
Appendix I: Objectives, Scope, and
Methodology

With our indexes, we analyzed trends in rental housing conditions by
renter household characteristics and rental unit characteristics. The renter
household characteristics we analyzed included household income and
affordability, race/ethnicity, and age. The rental unit characteristics we
analyzed included location, age, and structure type. In addition, from
American Community Survey data, we analyzed household crowding as
another aspect of housing conditions. Further, we reviewed reports and
studies on housing conditions and interviewed stakeholders including
federal agency officials, academic experts, and research organizations.

To further describe trends in renter household characteristics, rent
affordability, and rental housing conditions during our review period, we
reviewed reports and studies by federal agencies and research
organizations and interviewed a variety of stakeholders selected for their
knowledge of these issues, including federal agency officials from HUD,
the Census Bureau, Congressional Research Service, the Department of
Agriculture, the Federal Housing Finance Agency, and the Department of
the Treasury; academic experts, including researchers from Harvard’s
Joint Center for Housing Studies and others; research organizations that
included the Bipartisan Policy Center, various researchers associated
with the Board of Governors of the Federal Reserve System, Brookings
Institute, Center on Budget and Policy Priorities, Housing Assistance
Council, National League of Cities, National Rural Housing Coalition,
Urban Land Institute, and Urban Institute; and industry groups that
included the National Association of Home Builders, National Association
of Realtors, and the National Housing Conference.

We conducted this performance audit from February 2018 to May 2020 in
accordance with generally accepted government auditing standards.
Those standards require that we plan and perform the audit to obtain
sufficient, appropriate evidence to provide a reasonable basis for our
findings and conclusions based on our audit objectives. We believe that
the evidence obtained provides a reasonable basis for our findings and
conclusions based on our audit objective.

Page 39 GAO-20-427 Rental Housing
Appendix II: Statistical Analysis of Rental
Housing Conditions

This appendix provides additional details on our analysis of the conditions
of the national rental housing stock between 2001 and 2017.

Data To assess rental housing conditions, we used data from the national
American Housing Survey (AHS), which is administered by the Census
Bureau and conducted every odd year. Specifically, we considered two
concepts, unit completeness and unit quality, and relied on questions
which were consistently asked over the 2001 to 2017 period to define
nine completeness and 13 quality variables.45

These are described in table 4.

45See Department of Housing and Urban Development and Census Bureau, American
Housing Survey Codebook, accessed February 4, 2020,
https://www.census.gov/data-tools/demo/codebook/ahs/ahsdict.html.

Page 40 GAO-20-427 Rental Housing
Appendix II: Statistical Analysis of Rental
Housing Conditions

Table 4: GAO Rental Housing Conditions Index Variables

GAO index Variable
Completeness Electricity

Hot and cold running water

Heating equipment

A bathroom sink

A bathroom toilet

A bathroom tub or shower
A kitchen sink
A kitchen refrigerator

Any kitchen cooking appliance (a stove or range with oven, burners, or a microwave)

Quality Any 24 hour or longer period during which the unit was uncomfortably cold in the past winter

The number of times the main heating equipment broke down for 6 hours or more in the past winter

The number of times the unit was completely without running water for 6 hours or more in the past 3 months

The number of toilet breakdowns lasting 6 hours or more in the past 3 months

The number of sewer breakdowns lasting 6 hours or more in the past 3 months

Any water leaks from outside the unit in the past 12 months

Any water leaks from inside the unit in the past 12 months

Any open holes or cracks wider than the edge of a dime in the inside walls or ceilings

Any holes in the floors big enough for someone to catch their foot on

Any areas of peeling paint or broken plaster larger than 8 inches by 11 inches

Evidence of rodents inside the unit in the past 12 months

The number of times fuses blew or circuit breakers tripped in the past 3 months

Any electrical wiring in the finished areas neither concealed inside walls nor under protective metal or plastic
coverings

Source: GAO analysis of American Housing Survey data. | GAO-20-427

The survey questions underlying uncomfortably cold periods and heating
equipment breakdowns were only asked of respondents who occupied
their unit in the winter prior to the survey year, so our main analysis of
rental unit quality only considered cash-rent housing units occupied by
households since the prior winter, while the analysis of unit completeness
considered all cash-rent units.4° See table 5 for the distributions of the
completeness and quality variables in 2017.

46Renter-occupied units include those that are rented for cash and those with no cash
rent, such as a life tenancy or units that come free with a job. We included only cash-rent
units in our analysis.

Page 41 GAO-20-427 Rental Housing
Appendix II: Statistical Analysis of Rental
Housing Conditions

Table 5: Renter Household Responses to AHS Questions about Rental Unit Conditions, 2017

Number of issues

GAO index Estimated percentage of 0 1 2 3 4 No Number of | Number of

renter households with: (No) (Yes) response observations interviews
in analysis

Completeness Missing electricity 99.9 0.1 —- —- — —_— 22,701 22,701
Missing hot and cold running water 99.5 05 —- — — —_— 22,701 22,701
Missing heating equipment 99.0 10 — — — —_— 22,701 22,701
Missing bathroom sink, toilet, or 99.7 0.0 0.0 0.2 —_— —_— 22,701 22,701
tub/shower
Missing kitchen sink, fridge, or cooking 98.9 0.7 O02 0.2 —_— —_— 22,701 22,701
appliance

Quality Uncomfortably cold winter periods 90.2 88 — — — 1.0 19,613 19,613
Heating equipment breakdowns 96.7 1.6 0.7 O23 0.7 —_— 19,613 19,613
Running water breakdowns 95.8 2.1 06 0.3 0.5 0.7 19,613 22,701
Toilet breakdowns 97.6 13 05 01 04 _— 19,613 22,701
Sewer breakdowns 98.4 0.9 04 O01 0.2 _— 19,613 22,701
Outside leaks 90.0 10.0 — — — _— 19,613 22,701
Inside leaks 88.7 11.3 — — — _— 19,613 22,701
Floor holes 98.2 180°0—- — — _— 19,613 22,701
Wall cracks 92.7 730- => — _— 19,613 22,701
Peeling paint 96.9 3.10 —- — — _— 19,613 22,701
Rodents inside 87.2 128 — — — _— 19,613 22,701
Blown fuses 91.9 40 19 08 1.4 _— 19,613 22,701
Exposed electrical wiring 96.8 320° — — — —_— 19,613 22,701

Legend: — = not applicable.

Source: GAO analysis of American Housing Survey (AHS) data. | GAO-20-427

Note: All estimates in this table have a margin of error of less than 1 percentage point. The heating
equipment, running water, and toilet breakdown variables all allowed respondents answers of up to 8
or more breakdowns. Survey responses of 5, 6, 7, and 8 or more breakdowns were pooled together
with responses of only 4 breakdowns for reporting simplicity, so that the percentage of responses in
this category always corresponds to 4 or more issues. However, all other analysis used the original
responses.

Methodology and Results — We used each set of variables to construct two indexes, one of unit
completeness (an indicator) and one of unit quality (a continuous
measure). We collapsed the continuous quality index into three
categories (no quality issues, less substantial quality issues, and
substantial quality issues) to facilitate a summary of rental housing quality
trends.

Page 42 GAO-20-427 Rental Housing
Appendix II: Statistical Analysis of Rental
Housing Conditions

Completeness

To determine rental unit completeness, we first summed the number of
missing components contained in the nine completeness components for
each rental unit in the surveys. We obtained an estimate of Cronbach's
alpha associated with this sum of 0.40, which was low enough to suggest
that a simple indicator would be a more appropriate summary measure.4?
We therefore determined that each of the nine completeness components
was essential for us to consider a unit livable, and assigned a
completeness score to each rental unit based on the absence of any of
them. The resulting index therefore measured incompleteness, where a
score of 1 indicated that a unit was missing one or more of the essential
components. See table 6 for the distribution of the completeness index in
the survey years between 2001 and 2017.

Table 6: Distribution of Completeness Index by AHS Year, 2001-2017

Year Estimated percentage of Estimated percentage of Number of
complete rental units incomplete rental units observations in

analysis

2001 97.2 2.8 12,954
2003 97.8 2.2 14,821
2005 98.0 2.0 13,034
2007 98.0 2.0 11,715
2009 98.3 1.7 14,133
2011 98.0 2.0 20,544
2013 98.3 1.7 23,358
2015 97.2 2.8 24,223
2017 97.5 2.5 22,701

Source: GAO analysis of American Housing Survey (AHS) data. | GAO-20-427

Note: All 2001 to 2007 estimates in this table have a margin of error of less than 3 percentage points,
and all 2009 to 2017 estimates have a margin of error of less than 1 percentage point. The essential
components we used to determine whether a rental unit was complete are electricity; hot and cold
running water; heating equipment; a bathroom sink, toilet, and tub or shower; and a kitchen sink,
refrigerator, and cooking appliance.

47The 95 percent confidence interval for this estimate is [0.39, 0.40].

Page 43 GAO-20-427 Rental Housing
Appendix II: Statistical Analysis of Rental
Housing Conditions

Quality

To determine rental unit quality, we conducted a factor analysis, a
statistical method that exploits correlations among variables related to
unobserved factors to obtain estimates for these factors. Given our
selection of variables, we interpreted the unobserved factor as rental unit
quality, and we specified the following measurement models:

where ¢; is an error term satisfying E[e;] = 0 Vi and cov(¢;,n) =

cov(€, &) = 0 vi # j, and A; are factor loadings capturing the linear effect
of unobserved quality n on y;. Each of the 13 observed categorical
variables, y;, has a latent continuous variable, y;, that determines the
values realized in the data, such that:

1,a, <y; <a

*
Ci,Ac, < Vi < ©

where c;+1 is the number of categories each y; can take, and ay, dz, ..., A,
are thresholds determining the mapping between each unobserved latent
continuous indicator y; and its observed categorical counterpart y;.48 This
model was estimated in two steps. We first estimated the thresholds and
polychoric correlations between the 13 unobserved continuous variables
by maximum likelinood.42 Then, given the vector of the 91 unique
polychoric correlation coefficients and the identification assumption 2, =
1, we obtained estimates of the model’s parameters (the factor loadings
and the variances of the latent quality and error terms) via diagonally-
weighted least squares.5° We used this estimator (in conjunction with the
full weights matrix in estimating standard errors and fit statistics) to allow

48c.+1 can be either four or eight (depending on AHS variable topcoding) for variables
counting a number of occurrences, such as the number of heating equipment or toilet
breakdowns, or two for variables that are just indicators of an issue, such as the water
leaks or evidence of rodents. Note that for these latter indicators, c; = 1 and a, = o. See
table 5 for more details.

49F or details, see U. Olsson, “Maximum Likelihood Estimation of the Polychoric
Correlation Coefficient,” Psychometrika, vol. 44 (1979): pp. 443-460.

50In particular, the factor loading set to one was that from the equation in which the

endogenous variable is the continuous counterpart to the indicator for whether a unit was
ever uncomfortably cold for a period of 24 hours or more in the previous winter.

Page 44 GAO-20-427 Rental Housing
Appendix II: Statistical Analysis of Rental
Housing Conditions

for inference that is robust to non-normal distributions of the latent
continuous variables. Finally, we obtained quality score estimates by
empirical Bayes, which selects the mode of the posterior distribution
ply) evaluated at the estimates of the model’s parameters. Note that
because all quality variables increased in the presence or number of
issues, the quality index correspondingly increased in poor quality.5'

We first estimated two variants of the factor model, one accounting for the
sample weights assigned to units in each survey (our preferred
specification), and one that did not account for these weights. The robust
root mean square error of approximation from the latter of 0.020
suggested that our single factor model provided an appropriate
representation of the AHS data.52 Estimates of the polychoric correlation
matrix and of the factor loadings are reported in tables 7 and 8
respectively. We assessed stability by estimating the model on each AHS
year separately and broadly found that factor loadings estimates varied
little over time. Given the estimates from our preferred specification,
which accounts for the survey design, we then relied on empirical Bayes
estimation to assign a quality score to each unit for which responses to all
13 quality variables were observed. The full distributions of the resulting
quality index in the survey years between 2001 and 2017 are reported in
figure 15.

51This is similar to an index previous researchers constructed by factor analysis. These
researchers showed that their index was very highly correlated with indexes using similar
underlying variables constructed by unweighted addition, ad hoc weighting, or with
weights derived from a consumer rating model. See Sandra Newman and C. Scott
Holupka, The Quality of America’s Assisted Housing Stock: Analysis of the 2011 and 2013
American Housing Surveys, a report prepared for the Department of Housing and Urban
Development (Falls Church, VA: Economic Systems Inc., 2017).

52The 90 percent confidence interval for the root mean square error is [0.019, 0.021]. The
statistical package (“lavaan’” in R) we used to perform the factor analysis could not directly
estimate our preferred specification accounting for sampling weights. We obtained this
specification's factor loadings estimates by approximating the survey design, but our
method did not allow us to obtain consistent estimates of standard errors or model fit
statistics, which is why we relied on the estimates from the unweighted specification to
assess model fit.

Page 45 GAO-20-427 Rental Housing
Appendix II: Statistical Analysis of Rental
Housing Conditions

Figure 15: Distribution of Continuous Quality Index, 2001, 2005, 2009, 2013, and 2017

Density

12

0.0

Quality score
Year

Mm 2001
= om 2009

2013

Source: GAO analysis of American Housing Survey data. | GAO-20-427

Note: This relative frequency polygon has an interval width of 0.05. The quality index is constructed
by factor analysis and is increasing in quality issues. The quality score associated with no quality
issues is approximately -0.23. Quality variables include: whether there was any 24 hour or longer
period during which the unit was uncomfortably cold in the past winter and, if so, the number of times
the main heating equipment broke down for 6 hours or more; the number of times the unit was
completely without running water for 6 hours or more in the past 3 months; the number of toilet and
sewer breakdowns lasting 6 hours or more in the past 3 months; whether there were any outside or
inside water leaks in the past 12 months; whether there were any holes in the floors large enough to
catch someone’s foot, any open holes or cracks wider than the edge of a dime in the inside walls or
ceilings, or any areas of peeling paint or broken plaster larger than 8 inches by 11 inches; whether
there was evidence of rodents inside the unit in the past 12 months; the number of times fuses blew
or circuit breakers tripped in the past 3 months; and whether any electrical wiring in the finished areas
was neither concealed inside walls nor under protective metal or plastic coverings.

Page 46 GAO-20-427 Rental Housing
Appendix Il: Statistical Analysis of Rental Housing Conditions

distinguish between units without any quality issues, units with a quality score indicating the presence of less substantial issues, and units with a score denoting more
substantial issues (those with either a combination of some of the most severe issues as determined by the model, or a large number of issues of varying severities).

Table 8: Factor Loadings Model Estimates

Variable Factor loading Factor loading Standard error, 95 percent 95 percent confidence
estimate, model estimate, model model without confidence interval interval upper bound,
with sample design without sample sample design lower bound, model model without sample
design without sample design
design
Uncomfortably cold winter periods 1.00 1.00 — — —
Heating equipment breakdowns 0.88 0.89 0.01 0.87 0.91
Running water breakdowns 0.78 0.78 0.02 0.75 0.81
Toilet breakdowns 1.05 1.05 0.02 1.02 1.08
Sewer breakdowns 1.07 1.07 0.02 1.04 1.11
Outside leaks 0.90 0.91 0.01 0.89 0.94
Inside leaks 0.93 0.95 0.01 0.93 0.98
Floor holes 1.21 1.21 0.02 1.18 1.24
Wall cracks 1.28 1.28 0.01 1.25 1.31
Peeling paint 1.21 1.21 0.01 1.18 1.24
Rodents inside 0.85 0.86 0.01 0.84 0.88
Blown fuses 0.82 0.82 0.01 0.80 0.85
Exposed electrical wiring 0.42 0.41 0.02 0.37 0.45
Legend: — = not applicable.

Source: GAO analysis of American Housing Survey data. | GAO-20-427

Note: The renter household universe in this table is that of cash-rent households that occupied their units since at least the winter prior to the survey interview. In 2017,

an estimated 14.8 percent of cash-rent households did not occupy their units in the winter prior to the interview, and we excluded them because they were not asked

about having uncomfortably cold winter periods or heating equipment breakdowns. The factor loading on the uncomfortably cold winter periods variable is set to 1 for
model identification. We used the R package "lavaan" to perform the factor analysis. Because this package cannot estimate our preferred specification accounting for
sampling weights, we obtained this specification's factor loadings estimates by approximating the sample design, but we did not estimate associated standard errors.

Page 48 GAO-20-427 Rental Housing
Appendix II: Statistical Analysis of Rental
Housing Conditions

We then selected thresholds in the distribution of the continuous quality
index to distinguish between units without any quality issues, units with a
quality score indicating the presence of less substantial issues, and units
with a score denoting more substantial issues (those with either a
combination of some of the most severe issues as determined by the
model, or a large number of issues of varying severities).

The first threshold between units with no quality issues and units with at
least one issue occurred at a score of -0.2280. Units with no issues
represented between 54 and 62 percent of the rental units to which we
were able to assign quality scores. To further separate units experiencing
any issues into two groups, we inspected the quality score distribution for
local minima in its density to find a score around which small
perturbations in threshold choice would have little effect on the share of
units falling into each of the two groups. We examined all quality issue
profiles experienced in units with scores in the region around two
candidates where the density nearly reached 0, and selected a score of
0.5240 as the second threshold, immediately above which were units with
one or more holes in the floor large enough to catch someone’s foot. All
units with a quality score of 0.5240 or higher were therefore considered to
have substantial issues.53

Table 9 reports the share of cash-rent, previous-winter-occupied units for
each quality level in the survey years between 2001 and 2017, and table
10 reports the most common quality issue profiles in 2017.

53Recall that the index is increasing in quality issues.

Page 49 GAO-20-427 Rental Housing
Appendix Il: Statistical Analysis of Rental Housing Conditions

Table 9: Distribution of Quality Index by AHS Year, 2001-2017

Year Estimated Estimated Estimated Estimated Less Less Substantial Substantial Number of
percentage percentage percentage percentage substantial substantial issues, first issues, third observations in
of of of of issues, first issues, third quartile quartile analysis

rental units rental units rental units rental units quartile quartile continuous continuous

with no with less with without continuous continuous qualityscore quality score

quality substantial substantial a quality quality quality score estimate estimate

issues quality quality score score estimate
issues issues estimate

2001 52.8 28.1 15.3 3.8 0.15 0.41 0.65 1.01 9,817
2003 55.8 27.5 12.8 4.0 0.15 0.41 0.65 1.01 12,105
2005 55.8 27.7 13.5 2.9 0.15 0.44 0.65 0.96 10,687
2007 55.6 27.2 13.3 3.9 0.15 0.44 0.65 0.96 10,400
2009 56.0 28.2 12.7 3.1 0.15 0.44 0.65 0.96 12,604
2011 52.4 30.4 13.3 3.9 0.15 0.41 0.65 0.99 16,841
2013 59.5 26.2 10.7 3.6 0.15 0.41 0.65 0.94 21,245
2015 57.0 26.6 13.5 2.9 0.15 0.44 0.65 0.97 21,922
2017 57.8 28.0 12.4 1.7 0.15 0.41 0.65 0.95 19,613

Source: GAO analysis of American Housing Survey (AHS) data. | GAO-20-427

Note: All 2001 to 2007 estimated percentages in this table have a margin of error of less than 3 percentage points, and all 2009 to 2017 estimated percentages have a
margin of error of 1 percentage point or less. The renter household universe in this table is that of cash-rent households that occupied their unit since at least the winter
prior to the survey interview. In 2017, an estimated 14.8 percent of cash-rent households did not occupy their unit in the winter prior to the interview, and we excluded
them because they were not asked about having uncomfortably cold winter periods or heating equipment breakdowns. The quality index is constructed by factor
analysis, and scores are then mapped to one of the three quality levels. The quality score associated with no quality issues is approximately -0.23. Quality variables

include: whether there was any 24 hour or longer period during which the unit was uncomfortably cold in the past winter and, if so, the number of times the main heating
equipment broke down for 6 hours or more, the number of times the unit was completely without running water for 6 hours or more in the past 3 months; the number of
toilet and sewer breakdowns lasting 6 hours or more in the past 3 months; whether there were any outside or inside water leaks in the past 12 months; whether there
were any holes in the floors large enough to catch someone's foot, any open holes or cracks wider than the edge of a dime in the inside walls or ceilings, or any areas of
peeling paint or broken plaster larger than 8 inches by 11 inches; whether there was evidence of rodents inside the unit in the past 12 months; the number of times fuses
blew or circuit breakers tripped in the past 3 months; and whether any electrical wiring in the finished areas was neither concealed inside walls nor under protective
metal or plastic coverings

Page 50 GAO-20-427 Rental Housing
Appendix Il: Statistical Analysis of Rental Housing Conditions

Table 10: Most Frequent Profiles of Quality Issues by Quality Level, 2017

Quality level Quality issues profile Estimated number 95 percent
of rental units, confidence level
2017 margin of error (in
(in thousands) thousands of units)
Less Evidence of rodents inside the unit in the past 12 months 1,740 153
substantial Inside water leaks in the past 12 months 1,328 128
quality issues
Outside water leaks in the past 12 months 1,247 130
One or more instances of fuses blowing or circuit breakers tripping in the past 3 months 881 103
Any 24 hour or longer period during which unit was uncomfortably cold in the past winter 638 82
Electrical wiring in the finished areas neither concealed inside walls nor under protective 617 77
metal or plastic coverings
Substantial Open holes or cracks wider than the edge of a dime in the inside walls or ceilings, 130 36
quality issues and evidence of rodents inside the unit in the past 12 months
Any 24 hour or longer period during which unit was uncomfortably cold, one or more main 117 39
heating equipment breakdowns lasting 6 hours or more in the past winter, and evidence of
rodents inside the unit in the past 12 months
Outside water leaks in the past 12 months, and open holes or cracks wider than the edge 111 34
of a dime in the inside walls or ceilings
Inside water leaks in the past 12 months, and open holes or cracks wider than the edge of 93 30
a dime in the inside walls or ceilings
Holes in the floors large enough to catch someone's foot 87 33
Any 24 hour or longer period during which unit was uncomfortably cold, and three or more 73 28

main heating equipment breakdowns lasting 6 hours or more in the past winter

Source: GAO analysis of American Housing Survey data. | GAO-20-427

Note: The renter household universe in this table is that of cash-rent households which occupied their unit since at least the winter prior to the survey interview. In 2017,
an estimated 14.8 percent of cash-rent households did not occupy their unit in the winter prior to interview, and we excluded them because they were not asked about
having uncomfortably cold winter periods or heating equipment breakdowns. The quality index is constructed by factor analysis, and scores are then mapped to one of
three quality levels (no quality issues, less substantial quality issues, or substantial quality issues).

Page 51

GAO-20-427 Rental Housing
Appendix II: Statistical Analysis of Rental
Housing Conditions

Limitations

Our analysis is subject to several limitations. In determining both unit
completeness and quality, we were limited to the variables consistently
available across all survey years. Therefore, we could not include
features not observed in the AHS which could be deemed to be important
components of either unit completeness (Such as a unit’s access to an
internet service provider) or quality (such as the presence of major
defects in the structure of the unit's building).

In the quality factor model, we assumed that quality was uncorrelated with
the error term from each measurement equation and that the error terms
were uncorrelated with each other to obtain estimates of the model’s
parameters, and ultimately the quality scores. A violation of these
assumptions would bias the estimates. For example, if rental units located
in regions with harsh weather were of systematically worse quality than
units in fairer weather regions, the estimated effect of poor quality ona
variable like the number of outdoor leaks could be overstated, which
would in turn overweight the importance of outdoor leaks in the estimation
of the quality scores, resulting in overly poor quality score estimates for
units experiencing outdoor leaks. Conversely, if units in harsh weather
regions were of systematically better quality than those in fairer weather
regions (e.g. as a measure of resilience) the estimated effect of poor
quality on outdoor leaks would be understated, biasing down the
importance of outdoor leaks in the estimation of quality scores. In general,
any systematic linear relationship between latent quality (7) and the
unobserved factors (¢;) affecting one of the 13 unobserved latent
continuous variables, or between the unobserved factors themselves,
would be a violation of the model’s assumptions.

Since the two quality variables recording uncomfortably cold periods and
heating equipment breakdowns were only asked of respondents who
occupied their unit in the winter prior to the survey year, we could not
assign quality scores to the 10 to 25 percent of rental units across years
which were occupied by recent movers. To assess potential biases on the
quality distribution of the full cash-rent-occupied rental housing stock
introduced by excluding this group, we therefore compared both groups
along the remaining 11 dimensions

Of the 11 observable quality variables, three exhibited an incidence of
issues that differed meaningfully across the two groups. These
differences were persistent throughout survey years and consistent in
their direction: units whose respondents moved in later than the winter
prior to the interview were between 5 to 10 percentage points less likely
to experience any outside leaks, inside leaks, and to report evidence of

Page 52 GAO-20-427 Rental Housing
Appendix II: Statistical Analysis of Rental
Housing Conditions

rodents. The differences were meaningful in that they corresponded to
over a halving of the incidence of the evidence of rodents, and up toa
halving of the incidence of both types of leaks in the recent-mover units
relative to the units for which all quality variables were available.

To evaluate the effect of these differences on the quality distribution of
the full universe of cash-rent units, we estimated a modified quality factor
model in which we dropped the uncomfortably cold periods and heating
equipment breakdowns variables. This allowed us to obtain quality scores
for both the units with the original scores and the recent-mover units. The
distributions of the modified quality indices in the two groups reached
their largest difference at the share of units without any of the set of 11
quality issues, and we estimated that across all survey years, 1.3 to 2.4
percentage points more units would likely have no measured quality
issues in the full cash-rent universe than we found in that which excludes
the recent movers. *4

Furthermore, the distributions of the modified indices truncated to exclude
the respective units without any of the 11 quality issues were largely
comparable. In the full universe of cash-rent units, we would therefore
expect decreases in each of the shares of less substantial issues and
substantial issues units proportional to their respective shares in the
partial universe, and in sum corresponding to the magnitude of the
increase in units with no issues each year.*> The alternative of including
recent movers in our main model at the expense of the uncomfortably
cold periods and heating equipment breakdowns variables would have
yielded a share of units without any other quality issues that we estimated
to be 3 to 4 percentage points higher than the share calculated using the
original index in the partial universe. Because we believed that these
variables should ultimately be included in the quality index, and because
we considered the biases we estimated to be relatively small, we retained
the original index.

54This approach assumes there would be no differences in the incidences of the omitted
uncomfortably cold periods and heating breakdowns variables if these were observable in
both groups. The estimated bias for units without quality issues in the 2017 survey was
+1.5 percentage points.

55The approximate biases for units with less substantial quality issues and substantial
quality issues in the 2017 survey were respectively -1 and -0.5 percentage points.

Page 53 GAO-20-427 Rental Housing
Appendix III: Additional Information on
Rentership and Affordability Trends

In this appendix we present our analysis of rentership and housing affordability by
age, race/ethnicity, locality type, and income from 2001 through 2017. The data on
renter households are from the American Community Survey’s 1-year estimates.

Table 11: Estimated Percentage of Households That Rent by Age, Race/Ethnicity, Locality Type, and Income, 2001-2017

Category 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
All households? 34 34 33 33 33 33 33 33 34 35 35 36 37 37 37 37 36
20-34 years 61 60 60 60 60 59 60 61 63 64 65 67 67 68 68 67 66
35—49 years 32 31 31 31 31 31 32 33 34 35 37 38 39 40 40 40 39
50-64 years 22 21 21 21 21 22 22 23 23 24 25 25 26 27 27 27 27
65 years and older 22 21 21 21 21 21 21 21 21 21 21 21 22 22 22 22 22
White 28 27 27 26 26 26 26 27 27 27 28 29 29 29 29 29 28
Black 54 54 53 54 54 53 53 54 55 56 56 57 58 59 59 59 58
Hispanic 55 54 53 52 52 51 50 51 52 53 54 54 55 55 55 55 53
Asian 47 45 44 43 41 40 40 41 41 42 43 43 43 43 43 42 41
Other 47 45 45 45 45 45 45 45 46 46 48 48 49 50 49 50 49

High-growth metro 34 33 33 32 34 33 33 33 34 35 36 37 37 38 38 38 37
areas

Moderate- 41 40 39 39 39 38 38 39 40 41 42 42 42 43 43 43 42
growth/high-

density metro

areas

Moderate- 34 33 33 32 32 32 32 33 34 34 35 35 36 36 36 36 35
growth/moderate-

density metro
areas

Negative-growth 30 30 29 29 29 29 29 30 30 31 32 32 32 33 33 33 32
metro areas
Nonmetro areas 27 27 27 28 27 27 27 28 28 28 28 29 29 29 29 29 29

Page 54 GAO-20-427 Rental Housing
Appendix III: Additional Information on Rentership and Affordability Trends

Category 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Extremely low 64 64 63 63 64 64 65 66 66 67 67 67 67 67 67 66 65
income

Very low income 50 50 49 49 49 49 51 51 51 51 51 52 54 54 54 55 54
Low income 41 40 39 39 40 40 41 41 41 41 41 42 44 45 45 46 45
Moderate income 31 29 28 27 28 28 29 30 29 29 30 31 32 34 34 35 35
Higher income 17 16 16 15 15 15 16 16 16 16 17 18 19 20 20 21 20

Source: GAO analysis of American Community Survey data. | GAO-20-427
Notes: Estimates in this table have a margin of error of +3 percentage points or fewer, at the 95 percent confidence level.

We defined four head-of-household age categories as younger (20-34 years old), early middle age (35-49 years old), late middle age (50-64 years old), and older (65
and older). We reported on five race/ethnicity categories, combining some Census categories for ease of analysis: White, Black, Hispanic (an ethnicity that applies to
individuals of any racial background), Asian (includes Asian, Native Hawaiian, and Other Pacific Islander), and Other (includes American Indian, Alaska Native, two or
more races, and some other race). We identified five locality types based on population growth and density: high-growth metro areas, moderate-growth/high-density
metro areas, moderate-growth/moderate-density metro areas, negative-growth metro areas, and nonmetro areas. We defined extremely low income as up to 30 percent
of Department of Housing and Urban Development area median family income (HAMFI); very low income as more than 30, up to 50 percent of HAMFI; low income as
more than 50, up to 80 percent of HAMFI; moderate income as more than 80, up to 120 percent of HAMFI; and higher income as greater than 120 percent of HAMFI.

*While “all households” includes households age 19 or younger, that category is not broken out in this table.

Page 55 GAO-20-427 Rental Housing
Appendix III: Additional Information on Rentership and Affordability Trends

Table 12: Estimated Number of Renter Households by Age, Race/Ethnicity, Locality Type, and Income, 2001-2017

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

ina 36,453,560 36,064,820 36,000,845 36,147,915 36,771,635 36,530,915 36,862,875 37,755,695 38,773,225 39,694,045 40,727,290 41,850,285 42,447,170 43,267,430 43,701,740 43,837,495 43,378,800
olds

20-34 years 14,657,365 14,495,425 14,422,510 14,310,735 14,277,800 13,600,650 13,565,570 13,899,025 14,422,705 14,279,560 14,409,675 14,807,250 14,816,305 14,871,815 14,882,110 14,830,305 14,602,505
35—49 years 11,150,435 10,983,960 10,900,750 10,843,670 11,076,265 11,183,085 11,259,065 11,366,900 11,508,735 11,772,120 12,142,535 12,388,120 12,474,905 12,678,540 12,757,705 12,674,285 12,467,695
50-64 years 5,357,955 5,402,595 5,601,200 5,883,990 6,177,070 6,495,005 6,732,390 7,078,020 7,372,380 7,960,730 8,386,745 8,704,015 8,977,470 9,279,500 9,386,980 9,433,840 9,371,005
65 years and older 4,740,145 4,634,445 4,579,125 4,610,005 4,757,415 4,831,400 4,884,300 5,004,570 5,056,245 5,327,920 5,461,905 5,658,030 5,877,705 6,145,700 6,391,090 6,607,015 6,642,105
White 21,795,050 21,273,010 21,103,110 20,839,170 21,095,330 20,967,000 21,113,195 21,531,865 22,018,000 22,172,740 22,615,485 23,093,880 23,285,870 23,402,345 23,457,255 23,280,720 22,810,395
Black 6,566,880 6,592,640 6,685,325 6,846,025 7,007,590 6,901,125 6,963,515 7,103,920 7,337,000 7,481,070 7,656,885 7,850,680 7,982,555 8,222,335 8,367,410 8,434,815 8,437,030
Hispanic 5,480,165 5,625,590 5,686,495 5,881,150 6,059,140 6,092,460 6,169,660 6,416,875 6,620,870 7,014,140 7,318,550 7,609,895 7,792,590 8,080,700 8,254,445 8,372,530 8,360,830
Asian 1,627,145 1,639,790 1,696,015 1,721,070 1,700,310 1,683,495 1,687,215 1,766,005 1,794,030 1,956,240 2,013,955 2,125,560 2,159,205 2,283,580 2,319,610 2,362,125 2,374,115
Other 984,315 933,790 829,895 860,505 909,265 886,840 929,285 937,030 1,003,325 1,069,855 1,122,415 1,170,265 1,226,960 1,278,470 1,303,015 1,387,305 1,396,430
High-growth metro areas 5,214,910 5,217,735 5,256,040 5,366,465 5,553,815 5,560,495 5,681,165 5,852,375 6,019,640 6,303,795 6,602,285 6,871,950 7,035,710 7,315,665 7,445,710 7,542,830 7,413,225

Moderate-growth/high-density 11,298,355 11,045,075 10,948,685 10,756,010 10,972,075 10,809,300 10,847,845 11,094,375 11,320,730 11,705,920 11,958,210 12,187,135 12,331,215 12,529,265 12,644,620 12,703,825 12,542,745
metro areas

Moderate-growth/moderate- 12,949,480 12,801,005 12,773,895 12,856,615 13,186,965 13,187,875 13,351,910 13,610,795 14,095,025 14,393,820 14,720,790 15,193,830 15,449,550 15,694,125 15,856,305 15,900,780 15,780,605
density metro areas

Negative-growth metro areas 2,392,345 2,354,310 2,317,785 2,320,585 2,312,505 2,227,490 2,237,425 2,300,425 2,384,910 2,379,970 2,465,980 2,517,380 2,522,170 2,560,545 2,577,675 2,573,505 2,553,070

Nonmetro areas 4,598,470 4,646,695 4,704,440 4,848,245 4,746,275 4,745,755 4,744,525 4,897,725 4,952,920 4,910,540 4,980,025 5,079,990 5,108,530 5,167,830 5,177,430 5,116,555 5,089,160
Extremely low income 8,598,680 9,103,535 9,539,315 9,813,355 10,025,070 9,667,425 9,298,735 9,564,285 10,408,310 11,001,665 11,546,810 11,487,555 10,982,505 11,763,215 11,477,735 11,125,320 10,932,075
Very low income 6,379,230 6,581,020 6,848,550 6,865,060 6,896,590 6,771,495 6,748,655 6,843,985 7,297,655 7,654,605 7,742,675 7,790,465 7,634,500 6,478,500 6,576,705 6,386,005 6,310,255
Low income 7,844,370 7,605,610 7,724,880 7,827,590 7,852,615 7,767,175 7,780,390 7,936,26 8,103,550 8,205,795 8,221,850 8,366,205 8,427,900 8,543,420 8,666,710 8,631,125 8,614,585
Moderate income 6,990,285 6,672,220 5,905,215 5,932,885 5,981,835 6,131,805 5,871,135 6,210,385 6,207,730 6,168,115 6,219,235 6,566,910 6,786,580 7,069,865 7,261,360 7,271,340 7,204,190
Higher income 6,640,995 6,102,435 5,982,890 5,709,030 6,015,530 6,193,015 7,163,955 7,200,785 6,755,980 6,663,865 6,996,720 7,639,155 8,615,685 9,412,430 9,719,230 10,423,700 10,317,700

Source: GAO analysis of American Community Survey data. | GAO-20-427

Notes: Estimates in this table have a relative margin of error of +8 percent or less of the estimated number of renter households, at the 95 percent confidence level.

We defined four head-of-household age categories as younger (20-34 years old), early middle age (35-49 years old), late middle age (50-64 years old), and older (65 and older). We reported on five race/ethnicity categories, combining some Census categories for ease of analysis: White, Black, Hispanic (an
ethnicity that applies to individuals of any racial background), Asian (includes Asian, Native Hawaiian, and Other Pacific Islander), and Other (includes American Indian, Alaska Native, two or more races, and some other race). We identified five locality types based on population growth and density: high-growth
metro areas, moderate-growth/high-density metro areas, moderate-growth/moderate-density metro areas, negative-growth metro areas, and nonmetro areas. We defined extremely low income as up to 30 percent of Department of Housing and Urban Development area median family income (HAMFI); very low
income as more than 30, up to 50 percent of HAMFI; low income as more than 50, up to 80 percent of HAMFI; moderate income as more than 80, up to 120 percent of HAMFI; and higher income as greater than 120 percent of HAMFI.

Page 56 GAO-20-427 Rental Housing
Appendix III: Additional Information on Rentership and Affordability Trends

Table 13: Estimated Number of Households with Rent Burdens by Age, Race/Ethnicity, and Locality Type, 2001-2017

Category Rent 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
burden
All Not rent 19,400,475 18,870,075 18,146,700 17,784,480 17,537,620 17,320,230 17,680,275 17,919,035 17,792,375 17,663,230 17,945,020 18,997,220 19,467,475 19,762,080 20,430,350 20,904,835 20,764,470

households burdened

Moderately 7,338,405 7,337,550 7,518,725 7,710,960 8,005,730 8,056,150 98,165,655 8,375,775 8,754,050 9,089,335 9,285,405 ,388,680 9,550,435 9,878,400 9,862,900 9,766,450 9,703,580
rent
burdened

Severely 6,891,520 7,080,315 7,537,355 7,813,230 8,408,115 8,358,460 8,238,905 8,704,340 9,323,565 9,909,010 10,359,780 10,300,910 10,230,495 10,368,230 10,125,975 9,990,055 9,774,080
rent
burdened

20-34 Not rent 8,112,310 7,792,420 7,449,415 7,260,090 6,99,275 6,573,820 6,657,720 6,761,870 6,786,950 6,483,225 6,498,520 6,901,155 7,017,675 7,059,590 7,261,860 7,436,665 7,363,115
years burdened

Moderately 2,941,875 2,924,85 3,029,830 3,026,705 3,150,320 3,059,545 3,036,805 3,125,275 3,273,350 3,269,320 3,316,430 3,334,965 3,331,990 3,420,195 3,375,065 3,306,805 3,270,255
rent
burdened

Severely 2,691,555 2,823,565 2,950,820 3,089800 3,295,195 3,094,105 3,029,070 3,188,265 3,492,900 3,630,165 3,676,670 3,635,125 3,551,960 3,486,485 3,352,810 3,238,780 3,129,700
rent
burdened

35-49 Not rent 6,439,600 6,215,265 5,926,220 5,684,020 5,706,640 5,750,550 5,803,045 5,840,950 5,650,040 5,618,480 5,758,175 6,091,875 6,200,560 6,281,730 6,491,075 6,577,790 6,486,185
years burdened

Moderately 2,078,000 2,114,340 2,177,030 2,290,790 2,332,385 2,369,630 2,442,460 2,458,875 2,546,205 2,654,745 2,706,965 2,705,455 2,735,190 2,815,205 2,786,110 2,739,750 2,726,755
rent
burdened

Severely 1,819,825 1,905,960 2,041,315 2,079,695 2,269,780 2,313,615 2,262,640 2,366,160 2,566,335 2,719,225 2,866,690 2,798,690 2,738,450 2,768,655 2,666,970 2,589,675 2,498,155
rent
burdened

50-64 Not rent 2,854,455 2,910,765 2,907,010 2,974,900 3,069,060 3,171,740 3,325,855 3,443,925 3,453,875 3,616,760 3,739,685 3,937,630 4,104,150 4,195,595 4,312,475 4,422,360 4,406,110
years burdened

Moderately 1,050,140 1,070,535 1,106,965 1,189,375 1,275,810 1,372,090 1,403,790 1,495,760 1,617,590 1,766,715 1,839,030 1,890,470 1,964,185 2,047,140 2,071,395 2,036,680 2,023,265
rent
burdened

Severely 973,085 958,640 1,103,265 1,201,530 1,296,670 1,395,025 1,427,890 1,555,755 1,667,495 1,890,450 2,079,935 2,121,335 2,129,750 2,217,825 2,179,220 2,174,820 2,141,190
rent
burdened

65 years Not rent 1,813,600 1,771,340 1,710,150 1,717,810 1,726,320 1,713,945 1,776,295 1,765,775 1,797,455 1,867,835 1,883,455 1,999,250 2,076,905 2,154,505 2,290,215 2,395,085 2,429,145
and older burdened

Moderately 1,151,515 1,126,400 1,095,230 1,104,935 1,143,100 1,170,050 1,200,485 1,212,980 1,236,760 1,331,145 1,359,630 1,406,160 1,459,130 1,541,180 1,576,005 1,623,245 1,622,225
rent
burdened

Severely 1,222,255 1,182,465 1,256,535 1,247,475 1,339,980 1,379,540 1,342,615 1,419,420 1,412,010 1,505,455 1,579,375 1,611,445 1,677,325 1,770,030 1,811,545 1,866,030 1,890,195
rent
burdened

White Not rent 12,123,525 11,708,190 11,215,675 10,870,315 10,728,040 10,529,830 10,759,335 10,894,870 10,778,000 10,487,045 10,707,240 11,281,475 11,451,860 11,532,585 11,790,315 11,898,610 11,708,240
burdened

Moderately 4,130,000 4,036,415 4,161,040 4,149,805 4,276,650 4,358,605 4,356,230 4,450,270 4,668,810 4,822,930 4,848,430 4,883,705 4,903,560 5,012,485 4,937,830 4829895 4,741,485
rent
burdened

Page 57 GAO-20-427 Rental Housing
Appendix III: Additional Information on Rentership and Affordability Trends

Category Rent 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
burden

Severely 3,642,520 3,683,360 3,934,405 4,008,765 4,276,640 4,260,770 4,195,915 4,409,450 4,719,345 4,965,950 5,123,370 4,996,915 4,980,715 4,913,475 4,784,980 4,690,370 4,509,790
rent
burdened

Black Not rent 3,131,950 3,048,580 2,954,260 2,937,675 2,903,745 2,832,640 2,893,190 2,914,780 2,895,950 2,861,815 2,852,820 3,047,550 3,133,360 3,188,165 3,360,250 3,543,715 3,532,930
burdened

Moderately 1,427,640 1,478,900 1,498,205 1,575,175 1,622,100 1,625,585 1,655,905 1,704,890 1,771,120 1,827,420 1,852,870 1,847,245 1,919,130 2,014,645 2,020,610 2,016,220 2,032,435
rent
burdened

Severely 1,577,805 1,637,890 1,745,850 1,847,790 1,999,500 1,965,325 1,945,515 2,020,440 2,167,720 2,268,885 2,387,100 2,382,670 2,350,180 2,431,195 2,378,810 2,297,470 2,303,375
rent
burdened

Hispanic Not rent 2,740,035 2,764,195 2,702,150 2,716,310 2,642,175 2,692,900 2,725,780 2,771,130 2,746,665 2,869,415 2,903,300 3,090,740 3,233,285 3,333,725 3,491,345 3,612,320 3,638,935
burdened

Moderately 1,294,750 1,351,985 1,372,135 1,472,000 1,575,690 1,546,445 1,605,260 1,655,205 1,724,355 1,819,865 1,928,285 1,983,410 2,040,545 2,121,310 2,163,520 2,150,840 2,167,810
rent
burdened

Severely 1,144,020 1,214,275 1,313,200 1,371,355 1,533,300 1,552,020 1,532,420 1,676,225 1,817,875 1,957,320 2,108,035 2,140,170 2,117,530 2,195,695 2,156,770 2,164,945 2,125,625
rent
burdened

Asian Not rent 883,980 869,515 875,955 873,660 846,070 859,895 869,255 906,330 921,125 983,910 1,011,620 1,078,285 1,107,795 1,147,250 1,190,820 1,214,080 1,233,285
burdened

Moderately 306,515 299,610 307,385 327,050 342,805 336,490 340,685 360,645 367,515 385,520 398,735 413,045 415,950 446,525 452,210 456,040 454,055
rent
burdened

Severely 320,005 339,865 368,680 373,140 370,870 362,970 354,230 371,055 371,190 431,575 439,000 463,620 465,855 494,185 490,205 501,520 503,435
rent
burdened

Other Not rent 520,980 479,590 398,660 386,525 417,595 404,960 432,715 431,925 450,630 461,050 470,040 499,165 541,180 560,350 597,615 636,115 651,080
burdened

Moderately 179,490 170,640 179,965 186,920 188,490 189,030 207,580 24,765 222,250 233,595 257,080 261,270 271,235 283,435 288,730 313,450 307,805
rent
burdened

Severely 207,175 204,930 175,215 212,185 227,800 217,375 210,825 227,170 247,435 285,280 302,280 317,540 316,220 333,680 315,205 335,755 331,850
rent
burdened

High-growth Not rent 2,828,930 2,734,215 2,632,935 2,630,020 2,659,945 2,678,345 2,776,780 2,821,560 2,766,675 2,824,620 2,960,225 3,193,695 3,337,770 3,442,030 3,603,085 3,679,940 3,648,740
metro areas burdened

Moderately 1,113,880 1,151,230 1,201,975 1,252,300 1,298,455 1,302,430 1,345,540 1,366,100 1,441,875 1,513,560 1,566,930 1,594,895 1,657,735 1,743,645 1,725,575 1,751,480 1,713,340
rent
burdened

Severely 926,680 995,035 1,076,200 1,121,240 1,238,345 1,229,320 1,208,195 1,304,920 1,436,360 1,563,290 1,629,555 1,646,325 1,591,490 1,677,965 1,649,685 1,641,355 1,607,715
rent
burdened

Not rent 6,033,905 5,737,950 5,469,875 5,201,235 5,111,225 5,043,550 5,084,360 5,150,540 5,163,765 5,138,015 5,158,140 5,381,595 5,450,100 5,496,935 5,644,685 5,797,675 5,738,405
burdened

Page 58 GAO-20-427 Rental Housing
Appendix III: Additional Information on Rentership and Affordability Trends

Category _ Rent 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
burden
Moderate- Moderately 2,343,065 2,359,330 2,401,870 2,387,820 2,535,740 2,486,705 2,539,690 2,583,915 2,643,230 2,788,525 2,822,600 2,832,430 2,889,260 2,976,685 2,963,035 2,924,575 2,915,715
growth/high- rent
density burdened
metro Severely 2,339,770 2,391,805 2,519,985 2,592,460 2,751,115 2,734,010 2,685,665 2,818,325 2,943,755 3,152,750 3,333,405 3,317,055 3,321,230 3,356,460 3,313,110 3,290,520 3,191,465
rent
burdened
Moderate— Not rent 6,909,090 6,766,240 6,515,025 6,412,780 6,368,465 6,308,095 6,474,340 6,547,980 6,550,450 6,488,285 6,571,905 6,960,105 7,190,200 7,275,155 7,548,080 7,734,955 7,659,105
growth burdened
lonsity Moderately 2,617,885 2,609,080 2,626,200 2,756,780 2,850,010 2,917,735 2,929,810 3,022,660 3,190,140 3,286,635 3,367,425 3,437,685 3,486,815 3,594,950 3,623,405 3,558,635 3,566,185
rent
metro areas burdened
Severely 2,435,430 2,490,730 2,658,935 2,717,695 2,988,060 2,966,820 2,960,220 3,088,835 3,325,925 3,554,290 3,697,790 3,698,440 3,651,560 3,694,540 3,576,485 3,523,135 3,493,680
rent
burdened
Negative- Not rent 1,317,855 1,264,030 1,210,075 1,158,165 1,096,295 1,048,830 1,055,780 1,073,655 1,078,810 1,059,900 1,078,255 1,147,725 1,160,280 1,181,380 1,212,050 1,249,675 1,246,935
growth burdened
metro areas “\oderately 451,885 441,895 463,210 458,930 465,855 457,775 467,010 478,430 494,570 504,880 515,930 517,115 523,185 535,530 539,595 533,335 519,900
rent
burdened
Severely 443,565 454,460 461,815 517,200 557,270 536,015 526,880 557,315. 609,405 612,200 651,625 628,730 623,870 615,105 599,980 587,070 572,345
rent
burdened
Nonmetro Not rent 2,310,695 2,367,635 2,318,790 2,382,280 2,301,690 2,241,405 2,289,010 2,325,295 2,232,675 2,152,405 2,176,490 2,314,095 2,329,130 2,366,580 2,422,450 2,442,590 2,471,280
areas burdened
Moderately 811,685 776,020 825,470 855,125 855,680 891,500 883,605 924,680 984,235 995,735 1,012,515 1,006,545 993,445 1,027, 1,011, 998,420 988,450
rent 595 295
burdened
Severely 746,075 748,285 820,415 864,640 873,320 892,295 857,945 934,940 1,008,120 1,026,480 1,047,410 1,010,370 1,042,345 1,024,160 986,715 947,980 908,865
rent
burdened

Source: GAO analysis of American Community Survey data. | GAO-20-427

Notes: Estimates in this table have a relative margin of error of +32 percent or less of the estimated number of renter households, at the 95 percent confidence level.

We defined four head-of-household age categories as younger (20-34 years old), early middle age (35-49 years old), late middle age (50-64 years old), and older (65 and older). We reported on five race/ethnicity categories, combining some Census categories for ease of analysis: White, Black, Hispanic (an
ethnicity that applies to individuals of any racial background), Asian (includes Asian, Native Hawaiian, and Other Pacific Islander), and Other (includes American Indian, Alaska Native, two or more races, and some other race). We identified five locality types based on population growth and density: high-growth
metro areas, moderate-growth/high-density metro areas, moderate-growth/moderate-density metro areas, negative-growth metro areas, and nonmetro areas. Rent burden is the share of household income used to pay gross rent, which includes rent and utility costs. We defined rent burden as spending more than 30
percent of household income on rent, moderate rent burden as spending more than 30 and up to 50 percent of household income on rent, and severe rent burden as spending more than 50 percent of household income on rent.

Page 59 GAO-20-427 Rental Housing
Appendix III: Additional Information on Rentership and Affordability Trends

Table 14: Estimated Percentage of Households with Rent Burdens by Age, Race/Ethnicity, and Locality Type, 2001-2017

Category Rent burden 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

All households Not rent burdened 58 57 55 53 52 51 52 51 50 48 48 49 50 49 51 51 52
Moderately rent 22 22 23 23 24 24 24 24 24 25 25 24 24 25 24 24 24
burdened
Severely rent 20 21 23 23 25 25 24 25 26 27 28 27 26 26 25 25 24
burdened

20—34 years Not rent burdened 59 58 55 54 52 52 52 52 50 48 48 50 50 51 52 53 53
Moderately rent 21 22 23 23 24 24 24 24 24 24 25 24 24 24 24 24 24
burdened
Severely rent 20 21 22 23 25 24 24 24 26 27 27 26 26 25 24 23 23
burdened

35—49 years Not rent burdened 62 61 58 57 55 55 55 55 52 51 51 53 53 53 54 55 55
Moderately rent 20 21 21 23 23 23 23 23 24 24 24 23 23 24 23 23 23
burdened
Severely rent 18 19 20 21 22 22 22 22 24 25 25 24 23 23 22 22 21
burdened

50—64 years Not rent burdened 59 59 57 55 54 53 54 53 51 50 49 50 50 50 50 51 51
Moderately rent 22 22 22 22 23 23 23 23 24 24 24 24 24 24 24 24 24
burdened
Severely rent 20 19 22 22 23 23 23 24 25 26 27 27 26 26 25 25 25
burdened

65 years and older Not rent burdened 43 43 42 42 41 40 41 40 40 40 39 40 40 39 40 41 41
Moderately rent 27 28 27 27 27 27 28 28 28 28 28 28 28 28 28 28 27
burdened
Severely rent 29 29 31 31 32 32 31 32 32 32 33 32 32 32 32 32 32
burdened

White Not rent burdened 61 60 58 57 56 55 56 55 53 52 52 53 54 54 55 56 56
Moderately rent 21 21 22 22 22 23 23 23 23 24 23 23 23 23 23 23 23
burdened
Severely rent 18 19 20 21 22 22 22 22 23 24 25 24 23 23 22 22 22
burdened

Black Not rent burdened 51 49 48 46 44 44 45 44 42 41 40 42 42 42 43 45 45
Moderately rent 23 24 24 25 25 25 25 26 26 26 26 25 26 26 26 26 26
burdened
Severely rent 26 27 28 29 31 31 30 30 32 33 34 33 32 32 31 29 29
burdened

Hispanic Not rent burdened 53 52 50 49 46 46 46 45 44 43 42 43 44 44 45 46 46
Moderately rent 25 25 25 26 27 27 27 27 27 27 28 27 28 28 28 27 27
burdened
Severely rent 22 23 24 25 27 27 26 27 29 29 30 30 29 29 28 27 27
burdened

Page 60 GAO-20-427 Rental Housing
Appendix III: Additional Information on Rentership and Affordability Trends

Category Rent burden 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Asian Not rent burdened 59 58 56 56 54 55 56 55 55 55 55 55 56 55 56 56 56
Moderately rent 20 20 20 21 22 22 22 22 22 21 22 21 21 21 21 21 21
burdened
Severely rent 21 23 24 24 24 23 23 23 22 24 24 24 23 24 23 23 23
burdened
Other Not rent burdened 57 56 53 49 50 50 51 50 49 47 46 46 48 48 50 49 50
Moderately rent 20 20 24 24 23 23 24 24 24 24 25 24 24 24 24 24 24
burdened
Severely rent 23 24 23 27 27 27 25 26 27 29 29 29 28 28 26 26 26
burdened
High-growth metro Not Rent 58 56 54 53 51 51 52 51 49 48 48 50 51 50 52 52 52
areas Burdened
Moderately rent 23 24 24 25 25 25 25 25 26 26 25 25 25 25 25 25 25
burdened
Severely rent 19 20 22 22 24 24 23 24 25 26 26 26 24 24 24 23 23
burdened
Moderate- Not rent Burdened 56 55 53 51 49 49 49 49 48 46 46 47 47 46 47 48 48
growth/high- Moderately rent 22. 22 23 23 24 2 25 24 2 2 2 2 2 2 2 2 25
density metro burdened
areas
Severely rent 22 23 24 25 26 27 26 27 27 28 29 29 28 28 28 27 27
burdened
Moderate- Not rent burdened 58 57 55 54 52 52 52 52 50 49 48 49 50 50 51 52 52
growth/moderate- “\iggerately rent 22. 22 #22 #23 23 2 2 24 24 #25 25 424 24 #25 25 24 24
density metro burdened
areas
Severely rent 20 21 23 23 24 24 24 24 25 27 27 26 25 25 24 24 24
burdened
Negative-growth Not rent burdened 60 59 57 54 52 51 52 51 49 49 48 50 50 51 52 53 53
metro areas Moderately rent 20 20 22 22 22 22 23 «©230©«©2306©623)06.23006230623062380623806l6U23 C22
burdened
Severely rent 20 21 22 24 26 26 26 26 28 28 29 27 27 26 26 25 24
burdened
Nonmetro areas Not rent burdened 60 61 58 58 57 56 57 56 53 52 51 53 53 54 55 56 57
Moderately rent 21 20 21 21 21 22 22 22 23 24 24 23 23 23 23 23 23
burdened
Severely rent 19 19 21 21 22 22 21 22 24 25 25 23 24 23 22 22 21
burdened

Source: GAO analysis of American Community Survey data. | GAO-20-427

Notes: Estimates in this table have a margin of error of +7 percentage points or fewer, at the 95 percent confidence level.

We defined four head-of-household age categories as younger (20-34 years old), early middle age (35—49 years old), late middle age (50-64 years old), and older (65 and older). We reported on five race/ethnicity categories, combining some Census categories for ease of analysis: White, Black, Hispanic (an
ethnicity that applies to individuals of any racial background), Asian (includes Asian, Native Hawaiian, and Other Pacific Islander), and Other (includes American Indian, Alaska Native, two or more races, and some other race). We identified five locality types based on population growth and density: high-growth
metro areas, moderate-growth/high-density metro areas, moderate-growth/moderate-density metro areas, negative-growth metro areas, and nonmetro areas. Rent burden is the share of household income used to pay gross rent, which includes rent and utility costs. We defined rent burden as spending more than 30
percent of household income on rent, moderate rent burden as spending more than 30 and up to 50 percent of household income on rent, and severe rent burden as spending more than 50 percent of household income on rent.

Page 61 GAO-20-427 Rental Housing
Appendix III: Additional Information on Rentership and Affordability Trends

Table 15: Estimated Number of Households with Rent Burdens by Income, 2001-2017

House Rent burden 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
hold
income
All Not rent 19,400,475 18,870,075 18,146,700 17,784,480 17,537,620 17,320,230 17,680,275 17,919,035 17,792,375 17,663,230 17,945,020 18,997,220 19,467,475 19,762,080 20,430,350 20,904,835 20,764,470
households burdened

Moderately 7,338,405 7,337,550 7,518,725 7,710,960 8,005,730 8,056,150 8,165,655 8,375,775 8,754,050 9,089,335 9,285,405 9,388,680 9,550,435 9,878,400 9,862,900 9,766,450 9,703,580

rent burdened
Severely rent 6,891,520 7,080,315 7,537,355 7,813,230 8,408,115 8,358,460 8,238,905 8,704,340 9,323,565 9,909,010 10,359,780 10,300,910 10,230,495 10,368,230 10,125,975 9,990,055 9,774,080

burdened
Extremely low Not rent 1,001,900 1,085,755 1,101,705 1,089,645 1,054,340 997,190 956,810 921,355 974,645 970,340 1,009,245 1,009,450 956,215 1,025,970 1,020,965 1,011,590 1,034,635
income burdened

Moderately 1,325,055 1,460,040 1,483,880 1,524,645 1,442,470 1,378,645 1,311,330 1,269,870 1,392,400 1,412,150 1,479,210 1,459,625 1,320,990 1,718,485 1,692,080 1,605,005 1,551,190

rent burdened
Severely rent 5,145,530 5,396,200 5,697,495 5,882,210 6,242,240 6,049,730 5,788,275 6,119,045 6,639,640 7,076,880 7,407,895 7,362,050 7,032,985 7,293,040 7,016,505 6,828,000 6,636,100

burdened
Very low Not rent 1,610,435 1,687,395 1,745,790 1,698,635 1,498,870 1,386,235 1,340,895 1,275,210 1,369,525 1,413,925 1,373,845 1,394,975 1,245,330 1,012,040 1,070,480 1,029,265 1,012,340
income burdened

Moderately 2,998,565 3,129,125 3,219,630 3,233,085 3,287,865 3,173,585 3,126,000 3,180,810 3,408,005 3,584,365 3,600,420 3,665,240 3,522,905 2,900,150 2,923,710 2,783,870 2,747,545

rent burdened
Severely rent 1,355,770 1,350,340 1,472,735 1,521,950 1,699,625 1,801,990 1,884,570 1,997,660 2,110,645 2,241,245 2,336,920 2,317,010 2,467,970 2,226,365 2,243,560 2,242,510 2,236,160

burdened
Low income Not rent 4,764,700 4,788,070 4,793,225 4,738,290 4,547,865 4,312,220 4,186,280 4,172,395 4,265,735 4,254,930 4,184,015 4,311,265 4,063,455 3,779,715 3,916,155 3,847,985 3,806,345
burdened
Moderately 2,279,320 2,092,580 2,176,500 2,301,640 2,476,640 2,596,670 2,713,945 2,857,355 2,933,585 3,035,445 3,113,240 3,125,150 3,352,110 3,624,275 3,603,005 3,624,205 3,670,005
rent burdened
Severely rent 312,710 271,600 295,020 323,410 378,200 408,805 453,745 477,255 470,315 480,070 504,450 502,430 581,610 693,725 698,010 743,725 745,840
burdened
Moderate Not rent 5,954,840 5,725,065 5,045,615 5,044,295 4,999,840 5,025,585 4,750,395 5,035,430 5,067,695 4,999,560 5,029,370 5,328,910 5,383,140 5,444,110 5,640,250 5,590,200 5,564,180
income burdened
Moderately 589,440 520,205 484,955 505,785 603,125 712,810 734,265 788,710 766,955 795,885 812,685 838,810 984,605 1,190,280 1,178,305 1,240,660 1,225,170
rent burdened
Severely rent 62,080 49,665 54,210 65,035 61,195 74,540 78,880 81,575 74,135 79,560 78,295 89,335 104,610 110,965 119,765 119,110 109,925
burdened
igher income Not rent 6,068,595 5,583,790 5,460,365 5,213,615 5,436,700 5,599,000 6,445,895 6,514,650 6,114,775 6,024,475 6,348,540 6,952,625 7,819,335 8,500,245 8,782,500 9,425,795 9,346,970
burdened
Moderately 146,020 135,595 153,765 145,800 195,625 194,440 280,120 279,035 253,110 261,485 279,840 299,855 369,820 445,210 465,800 512,715 509,680
rent burdened
Severely rent 15,430 12,510 17,890 20,625 26,860 23,395 33,435 28,805 28,825 31,255 32,230 30,090 43,325 44,135 48,135 56,710 46,050
burdened

Source: GAO analysis of American Community Survey data. | GAO-20-427

Notes: Estimated numbers of renter households in this table have a relative margin of error of about +16 percent of the estimated number, at the 95 percent confidence level unless otherwise noted. Estimated numbers of renter households in the higher income category have a relative margin of error of +32 percent
of the estimated number, at the 95 percent confidence level due to small sample size.

We defined extremely low income as up to 30 percent of Department of Housing and Urban Development area median family income (HAMFI); very low income as more than 30, up to 50 percent of HAMFI; low income as more than 50, up to 80 percent of HAMFI; moderate income as more than 80, up to 120
percent of HAMFI; and higher income as greater than 120 percent of HAMFI. Rent burden is the share of household income used to pay gross rent, which includes rent and utility costs. We defined rent burden as spending more than 30 percent of household income on rent, moderate rent burden as spending more
than 30 and up to 50 percent of household income on rent, and severe rent burden as spending more than 50 percent of household income on rent.

Page 62 GAO-20-427 Rental Housing
Appendix III: Additional Information on Rentership and Affordability Trends

Table 16: Estimated Percentage of Households with Rent Burdens by Income, 2001-2017

Household Rent burden 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

income

All households Not rent burdened 58 57 55 53 52 51 52 51 50 48 48 49 50 49 51 51 52
Moderately rent 22 22 23 23 24 24 24 24 24 25 25 24 24 25 24 24 24
burdened
Severely rent 20 21 23 23 25 25 24 25 26 27 28 27 26 26 25 25 24
burdened

Extremely low Not rent burdened 13 14 13 13 12 12 12 11 11 10 10 10 10 10 10 11 11

Income Moderately rent 18 18 18 18 17 16 16 15 15 15 15 15 14 17 17 17 17
burdened
Severely rent 69 68 69 69 71 72 72 74 74 75 75 75 76 73 72 72 72
burdened

Very low income Not rent burdened 27 27 27 26 23 22 21 20 20 20 19 19 17 16 17 17 17
Moderately rent 50 51 50 50 51 50 49 49 49 50 49 50 49 47 47 46 46
burdened
Severely rent 23 22 23 24 26 28 30 31 31 31 32 31 34 36 36 37 37
burdened

Low income Not rent burdened 65 67 66 64 61 59 57 56 56 55 54 54 51 47 48 47 46
Moderately rent 31 29 30 31 33 35 37 38 38 39 40 39 42 45 44 44 45
burdened
Severely rent 4 4 4 4 5 6 6 6 6 6 6 6 7 9 8 9 9
burdened

Moderate income Not rent burdened 90 91 90 90 88 86 85 85 86 85 85 85 83 81 81 80 81
Moderately rent 9 8 9 9 11 12 13 13 13 14 14 13 15 18 17 18 18
burdened
Severely rent 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2
burdened

Higher income Not rent burdened 97 97 97 97 96 96 95 95 96 95 95 95 95 95 94 94 94
Moderately rent 2 2 3 3 3 3 4 4 4 4 4 4 4 5 5 5 5
burdened
Severely rent 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 0
burdened

Source: GAO analysis of American Community Survey data. | GAO-20-427
Notes: Estimates in this table have a margin of error of +3 percentage points or fewer, at the 95 percent confidence level.

We defined extremely low income as up to 30 percent of Department of Housing and Urban Development area median family income (HAMFI); very low income as more than 30, up to 50 percent of HAMFI; low income as more than 50, up to 80 percent of HAMFI; moderate income as more than 80, up to 120
percent of HAMFI; and higher income as greater than 120 percent of HAMFI. Rent burden is the share of household income used to pay gross rent, which includes rent and utility costs. We defined rent burden as spending more than 30 percent of household income on rent, moderate rent burden as spending more
than 30 and up to 50 percent of household income on rent, and severe rent burden as spending more than 50 percent of household income on rent.

Page 63 GAO-20-427 Rental Housing
Appendix III: Additional Information on Rentership and Affordability Trends

Table 17: Estimated Rent-to-Income Ratio by Demographic Group and Locality Type, 2001-2017

Category 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
20-34 years 0.26 0.26 0.28 028 0.29 029 0.29 029 O30 0.31 O31 O30 030 0.30 029 0.29 0.28
35—49 years 0.25 0.25 0.26 027 0.27 O28 0.28 O28 0.29 0.29 O30 0.29 O29 0.29 028 0.28 0.28
50-64 years 0.26 0.26 0.27 O28 0.28 O28 0.28 O29 029 0.30 031 O30 030 0.30 030 0.29 0.29
65+ years 0.33 033 034 034 0.34 O35 034 O35 O35 035 O36 035 035 035 035 035 0.35
White 0.25 0.25 0.26 027 0.28 O28 0.28 O28 0.28 0.29 O29 0.28 O28 0.28 O28 0.28 0.27
Black 0.30 030 031 032 0.33 033 033 033 034 035 036 035 034 0.34 034 0.33 0.33
Hispanic 0.29 0.29 030 031 0.32 032 032 032 033 034 034 034 O33 0.33 033 032 0.32
Asian 0.26 0.26 0.27 O27 0.28 O28 0.28 O28 O27 0.28 O28 0.28 O27 0.28 O27 0.27 0.27
Other 0.27 0.27 0.29 O30 0.30 030 0.30 030 030 032 032 032 031 0.31 O30 0.30 0.30
High-growth metro areas 0.26 0.27 0.29 029 0.29 029 0.29 029 O31 O31 O31 O30 030 0.30 029 0.29 0.29

Moderate-growth/high-density 0.27 0.28 029 O30 030 030 030 031 O31 O32 032 0.32 0.32 0.32 0.31 0.31 0.31
metro areas

Moderate-growth/moderate- 0.27 0.27 O28 028 029 0.29 0.29 0.29 030 £0.31 0.31 0.30 0.30 0.30 0.29 0.29 0.29
density metro areas

Negative-growth metro areas 0.26 026 0.27 O28 0.29 029 0.29 030 030 031 O31 O30 030 0.30 029 0.29 0.29

Nonmetro areas 0.26 0.25 0.26 026 0.27 O27 0.27 0.28 0.29 0.29 O29 0.28 O29 0.28 O28 0.28 0.27
Extremely low income 0.70 0.70 0.70 0.72 0.74 O75 0.75 0.78 O.77 0.78 0.79 0.79 O80 0.76 0.75 0.76 0.76
Very low income 0.38 038 038 038 039 040 041 042 041 042 042 042 043 044 044 044 0.44
Low income 0.27 0.26 0.26 027 0.27 O28 0.28 029 0.29 0.29 O29 0.29 O30 0.31 O31 O31 0.31
Moderate income 0.20 019 0.19 020 0.20 021 O21 O21 O21 O21 O21 O21 O22 0.23 022 0.23 0.23
Higher income 0.14 013 O14 O14 O14 O14 O15 O15 O15 O15 O15 O15 O16 O16 016 0.16 0.16

Source: GAO analysis of American Community Survey data. | GAO-20-427
Notes: Estimates in this table have a margin of error of +2 percentage points or fewer, at the 95 percent confidence level.

We defined four head-of-household age categories as younger (20-34 years old), early middle age (35—49 years old), late middle age (50-64 years old), and older (65 and older). We reported on five race/ethnicity categories, combining some Census categories for ease of analysis: White, Black, Hispanic (an
ethnicity that applies to individuals of any racial background), Asian (includes Asian, Native Hawaiian, and Other Pacific Islander), and Other (includes American Indian, Alaska Native, two or more races, and some other race). We identified five locality types based on population growth and density: high-growth
metro areas, moderate-growth/high-density metro areas, moderate-growth/moderate-density metro areas, negative-growth metro areas, and nonmetro areas. Rent burden is the share of household income used to pay gross rent, which includes rent and utility costs. We defined rent burden as spending more than 30
percent of household income on rent, moderate rent burden as spending more than 30 and up to 50 percent of household income on rent, and severe rent burden as spending more than 50 percent of household income on rent

Page 64 GAO-20-427 Rental Housing
Appendix IV: Estimated Rent Burden in
Statewide Rural Areas

In this appendix we present state-level analysis of housing affordability for
rural renter households. While rental affordability is a challenge in both
rural and urban areas, differences in demographics, economies, housing
stock, and federal rental assistance programs make rural rental
affordability issues unique

We defined rural areas using the U.S. Department of Agriculture’s 2010
rural-urban commuting area (RUCA) codes.*’ The data on renter
households living in these areas are from the American Community
Survey’s 5-year estimates for 2013 through 2017.

While renter households lived in rural areas of all 50 states, generally the
most populous states had the largest populations of renter households in
rural areas (fig. 16). From 2013 through 2017, more than an estimated
2.2 million renter households lived in rural areas. The states with the
largest estimated populations of renter households in rural areas were
Texas (119,000), Missouri (96,000), Wisconsin (96,000), and Kentucky
(93,000).

56Rural-specific federal rental housing assistance includes the Section 515 Rural Rental
Housing Direct Loan program, Section 538 Rural Rental Housing Guaranteed Loan
program, Section 521 Rental Assistance program, and Section 514/516 Farm Labor
Housing programs. Other federal programs, while not rural-specific, also assist renters in
rural areas, including the low-income housing tax credit and Section 8 project-based and
housing choice vouchers.

5’There is no single agreed-upon definition of rural areas used by the U.S. government.
While our nonmetro group could be considered rural, it includes urban clusters with
populations up to 50,000. Further, the other four locality types may include areas of low
population densities that could be considered rural. For these reasons, we developed a
supplementary rural analysis that captures low population density areas based on census
tracts. The definition of rural areas for purposes of this report combines census tracts with
secondary RUCA codes 7, 8, 9, 10, and 10.3 from the 2010 RUCA codes revised as of
July 3, 2019.

Page 65 GAO-20-427 Rental Housing
Appendix IV: Estimated Rent Burden in
Statewide Rural Areas

Figure 16: Estimated Number of Renter Households in Rural Areas by State, 2013—2017

Rural renter households

1-30,000
30,001-60,000

Source: GAO analysis of Department of Agriculture and American Community Survey data. | GAO-20-427

Note: Estimates in this figure have a margin of error of +13 percentage points or fewer, at the 95
percent confidence level with the exception of two states, which have a relative margin of error of +51
percent (Rhode Island) and +34 percent (New Jersey) of the estimate due to small sample size.

The prevalence of rural renter households varied significantly by state.
While only about 5 percent of renter households lived in rural areas from
2013 through 2017, some states had significantly larger proportions of
renters in rural areas. States with higher estimated proportions of rural

Page 66 GAO-20-427 Rental Housing
Appendix IV: Estimated Rent Burden in
Statewide Rural Areas

renter households generally had small populations and were in northern
New England or along the Missouri, Mississippi, or Ohio Rivers (fig. 17).
The states with the largest estimated proportions of renter households in
rural areas were Vermont (39 percent) and Montana (32 percent).

Figure 17: Estimated Percentage of Renter Households in Rural Areas by State, 2013-2017

Renter households: percent rural
0-9%
10-19%
PO 20-20%
| | 30-39%
Source: GAO analysis of Department of Agriculture and American Community Survey data. | GAO-20-427

Note: Estimates in this figure have a margin of error of +1 percentage point or less, at the 95 percent
confidence level.

Page 67 GAO-20-427 Rental Housing
Appendix IV: Estimated Rent Burden in
Statewide Rural Areas

Renter households in rural areas generally had lower incomes than other
renter households. From 2013 through 2017, while the median income for
renter households overall was an estimated $36,653, nearly three in five
rural renter households had incomes lower than $35,000. For context, a
household with two full-time jobs earning the federal minimum wage in
2017 would earn approximately $30,160.58 In general, Southern states
had the highest estimated proportion of rural renter households with
incomes less than $35,000 (fig. 18). The states with the smallest
proportion of rural renter households with incomes lower than $35,000
were New Jersey (25 percent), Rhode Island (32 percent), Alaska (35
percent), Hawaii (39 percent), and Connecticut (39 percent).

58The federal minimum wage in 2017 was $7.25 per hour. We calculate full-time jobs as
working 40 hours per week for 52 weeks per year.

Page 68 GAO-20-427 Rental Housing
Appendix IV: Estimated Rent Burden in
Statewide Rural Areas

Figure 18: Estimated Percentage of Rural Renter Households with Income Less than $35,000 by State, 2013-2017

Rural renter households: percent with income <$35,000

35-44%
45-54%
| 55-64%

65-74%
RLS Insufficient data

Source: GAO analysis of Department of Agriculture and American Community Survey data. | GAO-20-427

Note: Estimates in this figure have a margin of error of +9 percentage points or fewer, at the 95
percent confidence level.

Page 69 GAO-20-427 Rental Housing
Appendix IV: Estimated Rent Burden in
Statewide Rural Areas

Rent burden was common among renter households in rural areas, but
prevalence varied by state.59 Rent burden was slightly less common
among rural renter households from 2013 through 2017 (45 percent) than
renter households in general in 2017 (48 percent). In eight of 48 states, at
least 50 percent of rural renter households were rent burdened (fig. 19).
In general, rural rent burden was most common in the Northeast, South,
and West Coast, and least common in the U.S. interior. Louisiana had the
highest estimated rate of rent burden among rural renter households (55
percent) and Wyoming had the lowest (33 percent).

59In the publicly available Census data we used for our analysis of rent burden in rural
areas, households were aggregated into groups based on the percentage of income they
spent on rent. For this analysis, we defined the group of households spending 30 percent
or more of household income on rent as rent burdened. This definition differs slightly from
the rent burden definition we used for the larger analysis of rent burden in this report. For
the larger analysis, we defined rent burden as spending more than 30 percent of
household income on rent, moderate rent burden as spending more than 30 and up to 50
percent of household income on rent, and severe rent burden as spending more than 50
percent of household income on rent. These definitions are consistent with other housing
research and Department of Housing and Urban Development policies.

Page 70 GAO-20-427 Rental Housing
Appendix IV: Estimated Rent Burden in
Statewide Rural Areas

Figure 19: Estimated Percentage of Rural Renter Households with Rent Burdens by State, 2013-2017

Rural renter households: percent rent burdened
30-39%

Qe Insufficient data

Source: GAO analysis of Department of Agriculture and American Community Survey data. | GAO-20-427

Note: Estimates in this figure have a margin of error of +7 percentage points or fewer, at the 95
percent confidence level.

Rent burdens were more common among rural households with incomes
below $35,000. From 2013 through 2017, an estimated 70 percent of
these households were rent burdened, and in no individual state were
less than 50 percent of these households rent burdened (fig. 20). The five

Page 71 GAO-20-427 Rental Housing
Appendix IV: Estimated Rent Burden in
Statewide Rural Areas

states with the highest proportion of lower-income rural renter households
that were rent burdened were Alaska (81 percent), Massachusetts (83
percent), Hawaii (83 percent), California (83 percent), and Delaware (85
percent). As discussed previously in this report, lower-income households
with rent burdens may struggle to pay for essential needs like food,
transportation, health care, and clothing.

Figure 20: Estimated Percentage of Rural Renter Households with Income Less than $35,000 That Were Rent Burdened by
State, 2013-2017

Rural renter households with income <$35,000: percent rent burdened

51-60%
61-70%
71-80%
81-90%

SH

Insufficient data

Source: GAO analysis of Department of Agriculture and American Community Survey data. | GAO-20-427

Page 72 GAO-20-427 Rental Housing
Appendix IV: Estimated Rent Burden in
Statewide Rural Areas

Note: Estimates in this figure have a margin of error of +9 percentage points or fewer, at the 95
percent confidence level.

Rent burdens were uncommon among rural households with incomes of
$35,000 or greater. From 2013 through 2017, only an estimated 9 percent
of these households were rent burdened, and in no state were more than
30 percent of these households rent burdened (fig. 21). In 40 of 48 states,
less than an estimated 15 percent of rural renter households with
incomes of $35,000 or greater were rent burdened. The four states with
the highest proportion of rural renter households with income $35,000 or
greater that were rent burdened were Connecticut (28 percent), Hawaii
(26 percent), California (24 percent), and Massachusetts (22 percent).

Page 73 GAO-20-427 Rental Housing
Appendix IV: Estimated Rent Burden in
Statewide Rural Areas

Figure 21: Estimated Percentage of Rural Renter Households with Income $35,000 or Greater That Were Rent Burdened by
State, 2013-2017

>

Rural renter households with income >$35,000: percent rent burdened
1-10%
11-20%

RL Insufficient data

Source: GAO analysis of Department of Agriculture and American Community Survey data. | GAO-20-427

Note: Estimates in this figure have a margin of error of +11 percentage points or fewer, at the 95
percent confidence level.

Page 74 GAO-20-427 Rental Housing
Appendix V: Comparison of GAO Housing
Conditions Indexes and HUD Adequacy
Index

This appendix describes how the indexes we developed to analyze rental
housing conditions compare to an index the Department of Housing and
Urban Development (HUD) uses to measure housing adequacy. Although
our index uses many of the same American Housing Survey variables as
HUD’s adequacy index, differences in our analytic methods allowed us to
produce more detailed results on housing conditions.

HUD measures housing adequacy as part of its ongoing efforts to analyze
and report on worst case housing needs®°

The adequacy index is a measure that is based on 19 variables in the
American Housing Survey. It categorizes housing units as severely
inadequate, moderately inadequate, or adequate based on whether a
surveyed housing unit meets certain conditions or criteria.®1 Severely
inadequate housing units represented 2 to 3 percent of all rental units
from 2001 through 2017.62

We developed two indexes based on a factor analysis of 13 quality-
related variables and nine variables we identified as essential
components of a dwelling. We determined that two indexes were needed
to describe rental housing unit conditions based on American Housing
Survey data, as relevant variables fell into two categories that required
different statistical treatment and interpretation. Figure 22 provides a
detailed comparison between the variables and scoring techniques of our
indexes and HUD’s adequacy index.

S0HUD defines worst case needs as renter households that have very low incomes, and
do not receive housing assistance, and pay more than one-half their income for rent, live
in severely inadequate conditions, or both.

61The conditions and criteria for a severely inadequate score are based on a combination
of variables, including the presence of bathroom and kitchen facilities, whether a unit has
heat in the winter, and the presence of other hazards to health and well-being such as
water leaks, wall cracks, and rodents. The conditions and criteria for a moderately
inadequate score are based on a combination of 12 variables related to the same set of
hazards to health and well-being in addition to other factors such as lacking a source of
heat or having a kitchen lacking a sink, refrigerator, or cooking appliance. Any unit that
does not satisfy the conditions and criteria of a severely or moderately inadequate score is
scored as adequate.

82HUD has attributed the consistency in housing adequacy to more stringent building
codes over time and ongoing demolition of obsolete units. Department of Housing and
Urban Development, Worst Case Housing Needs: 2017 Report to Congress (Washington,
D.C.: August 2017).

Page 75 GAO-20-427 Rental Housing
Appendix V: Comparison of GAO Housing
Conditions Indexes and HUD Adequacy Index

Figure 22: Comparison of the Department of Housing and Urban Development’s (HUD) Adequacy Index and GAO’s Housing
Conditions Indexes

GAO completeness index
Units with any of the conditions below are incomplete

No electricity Y

No hot or cold water

No bathroom sink, toilet, or tub/shower

Non-exclusive use of bathrooms

No kitchen sink, refrigerator, or cooking appliance

Non-exclusive use of kitchen

Unvented heating equipment

No heating equipment

Heating breakdowns

Uncomfortably cold

Blown fuses

Exposed wires

Outlets not in all rooms

Outside leaks

Inside leaks

Floor holes

Wall cracks

Peeling paint

Rodents inside

Toilet breakdowns

Water breakdowns

Sewer breakdowns

Source: GAO analysis of Department of Housing and Urban Development (HUD) information. | GAO-20-427

Note: We developed two indexes to define the range of housing conditions based on American
Housing Survey data. The quality index, based on 13 quality-related variables, identified units as

Page 76 GAO-20-427 Rental Housing
Appendix V: Comparison of GAO Housing
Conditions Indexes and HUD Adequacy Index

having no quality issues, less substantial quality issues, and substantial quality issues. The
completeness index, based on nine variables we identified as essential components of a dwelling,
identified units as being incomplete or complete.

We compared HUD’s 2017 housing adequacy findings to the results of
our indexes and identified some notable differences. Among rental units
that HUD considered adequate in 2017, an estimated 8 percent had
substantial quality issues as measured by our quality index—affecting 2.7
million households. In addition, another estimated 9.7 million units had
less substantial quality issues. These units did not satisfy HUD’s scoring
criteria for inadequate or moderately inadequate units, but they had a
combination of issues that exceeded our statistical thresholds for
substantial and less substantial quality issues.® Figure 23 provides a
detailed comparison of how our results compare to HUD’s.

83We assigned a weight to each variable based on its estimated correlation with all other
quality-related variables, whereas HUD’s index is based on whether a given combination
of variables is present.

Page 77 GAO-20-427 Rental Housing
Appendix V: Comparison of GAO Housing
Conditions Indexes and HUD Adequacy Index

Figure 23: Comparison of Findings from the Department of Housing and Urban Development’s (HUD) Adequacy Index and
GAO’s Housing Conditions Indexes, 2017

Overall (rental housing units in millions)

Overlap by GAO quality and completeness indexes

No quality issues (GAO) QOOOOOQOOCOOOOOOOOCO0CE! 20.41
Moderately inadequate (HUD) [] 0.34
Severely inadequate (HUD) 0.00

Less substantial quality issues (GAO) BBDBPPPHHLD 9.88
Adequate (HUD) [] [IT [IIT LIL 9.66
Moderately inadequate (HUD) |] 0.22
Severely inadequate (HUD) 0.00

Substantial quality issues (GAO) K®6&d 4.34
Adequate (HUD) C1] 2.71
Moderately inadequate (HUD) [i] 1.18
Severely inadequate (HUD) §J 0.45

Incomplete (GAO) | 1.06
Adequate (HUD) []0.34
Moderately inadequate (HUD) [J 0.40
Severely inadequate (HUD) fJ0.32

HUD adequacy index GAO quality index GAO completeness index
CL] Adequate O No quality issues cor Incomplete

Da Moderately inadequate @B Less substantial quality issues

Y | Severely inadequate ® Substantial quality issues

Source: GAO analysis of American Housing Survey data from Department of Housing and Urban Development (HUD). | GAO-20-427

Page 78 GAO-20-427 Rental Housing
Appendix VI: Additional Information on
Rental Housing Conditions

This appendix provides additional information on rental housing conditions by household income, affordability (rent
burden), race/ethnicity, age, rental unit age, and structure type, based on two indexes we developed to analyze
American Housing Survey data®

The appendix also includes information on household crowding based on our analysis of American Community
Survey data by household income, rent burden, race/ethnicity, and age.

Table 18: GAO Quality and Completeness Indexes, Rental Units by Housing Condition and Income, 2017

Numbers in millions

Renter Index category
b / Incomplete housing Substantial quality Less substantial quality No quality issues All rental units
y race units issues issues

ethnicity

Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated

number percentage number percentage number percentage number percentage number percentage

All 1.06 3 4.34 12 9.88 28 20.41 57 35.69 100
households
Extremely 0.47 5 1.41 14 2.56 26 5.34 55 9.77 100
low income
Very low 0.17 3 0.68 13 1.41 26 3.09 58 5.34 100
income
Low income 0.20 3 0.83 12 2.05 29 3.93 56 7.01 100
Moderate 0.23 2 1.42 10 3.87 28 8.05 59 13.57 100
and higher
income

Source: GAO analysis of American Housing Survey data. | GAO-20-427
Notes: We developed two indexes to define the range of housing conditions based on American Housing Survey data. The quality index, based on 13 quality-

related variables, identified units as having no quality issues, less substantial quality issues, and substantial quality issues. The completeness index, based
on nine variables we identified as essential components of a dwelling, identified units as being incomplete or complete.

64App. Il includes more detailed information about the indexes we developed.

Page 79 GAO-20-427 Rental Housing
Appendix VI: Additional Information on Rental Housing Conditions

Estimates of the number of housing units in this table have a relative margin of error of +25 percent or less of the estimated number. Estimates of the percent
of housing units in this table have a margin of error of +7 percentage points or fewer, at the 95 percent confidence level.

Extremely low income is defined as up to 30 percent of HUD area median family income (HAMFI); very low income as more than 30, up to 50 percent of

HAMFI; low income as more than 50, up to 80 percent of HAMFI; moderate income as more than 80, up to 120 percent of HAMFI, and higher income as

greater than 120 percent of HAMFI.
Table 19: GAO Quality and Completeness Indexes, Rental Units by Housing Conditions and Rent Burden, 2017

Numbers in millions

Renter Index category
b t Incomplete housing Substantial quality Less substantial quality No quality issues All rental units
y ren units issues issues

burden

Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated

number percentage number percentage number percentage number percentage number percentage

All 1.06 3 4.34 12 9.88 28 20.41 57 35.69 100
households
Not rent 0.42 2 2.03 11 5.10 28 10.37 58 17.92 100
burdened
Rent 0.64 4 2.31 13 4.78 27 10.04 57 17.77 100
burdened
Moderate 0.29 3 1.07 12 2.42 27 5.08 57 8.85 100
burden
Severe 0.35 4 1.24 14 2.36 26 4.97 56 8.92 100
burden

Source: GAO analysis of American Housing Survey data. | GAO-20-427

Notes: We developed two indexes to define the range of housing conditions based on American Housing Survey data. The quality index, based on 13 quality-
related variables, identified units as having no quality issues, less substantial quality issues, and substantial quality issues. The completeness index, based
on nine variables we identified as essential components of a dwelling, identified units as being incomplete or complete.

Estimates of the number of housing units in this table have a relative margin of error of +23 percent or less of the estimated number. Estimates of the percent
of housing units in this table have a margin of error of +5 percentage points or fewer, at the 95 percent confidence level.

Households that were not rent burdened had rent costs that were 30 percent or less of household income. Rent-burdened households had rent costs more
than 30 percent of household income. Of those, moderate burden households had rent costs of more than 30 and up to 50 percent of household income, and
severe burden households had rent costs that exceeded 50 percent of household income.

Page 80 GAO-20-427 Rental Housing
Appendix VI: Additional Information on Rental Housing Conditions

Table 20: GAO Quality and Completeness Indexes, Rental Units by Condition and Race/Ethnicity, 2017

Numbers in millions

Renter Index category
households by
race/ethnicity

Incomplete housing Substantial quality Less substantial No quality issues All rental units
units issues quality issues

Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated
number percentage number percentage number percentage number percentage number percentage

All households 1.06 3 4.34 12 9.88 28 20.41 57 35.69 100
White 0.43 2 2.04 11 5.09 29 10.24 58 17.81 100
Black 0.14 2t 1.03 14 2.09 28 4.20 56 7.47 100
Hispanic 0.33 4 0.92 13 1.87 26 4.19 57 7.31 100
Asian 0.11 6 0.16 8 0.47 23 1.29 63 2.03 100
Other 0.05 4 0.18 17 0.35 33 0.49 46 1.07 100

Source: GAO analysis of American Housing Survey data. | GAO-20-427

Notes: We developed two indexes to define the range of housing conditions based on American Housing Survey data. The quality index, based on 13 quality-
related variables, identified units as having no quality issues, less substantial quality issues, and substantial quality issues. The completeness index, based
on nine variables we identified as essential components of a dwelling, identified units as being incomplete or complete.

Estimates of the number of housing units in this table have a relative margin of error of +30 percent or less of the estimated number. Estimates of the percent
of housing units in this table have a margin of error of +9 percentage points or fewer, at the 95 percent confidence level.

We reported on five race/ethnicity categories, combining some Census categories for ease of analysis: White, Black, Hispanic (an ethnicity that applies to
individuals of any racial background), Asian (includes Asian, Native Hawaiian, and Other Pacific Islander), and Other (includes American Indian, Alaska
Native, two or more races, and some other race).

Page 81 GAO-20-427 Rental Housing
Appendix VI: Additional Information on Rental Housing Conditions

Table 21: GAO Quality and Completeness Indexes, Rental Units by Condition and Age, 2017

Numbers in millions

Renter Index category
households Incomplete housing Substantial quality Less substantial No quality issues All rental units
y age units issues quality issues
Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated
number percentage number percentage number percentage number percentage number percentage
All 1.06 3 4.34 12 9.88 28 20.41 57 35.69 100
households
Younger 0.28 3 1.39 13 3.19 30 5.76 54 10.62 100
Early middle 0.30 3 1.43 13 2.99 28 5.97 56 10.69 100
age
Late middle 0.25 3 1.14 13 2.30 27 4.85 57 8.54 100
age
Older 0.24 4 0.36 6 1.36 24 3.73 66 5.68 100

Source: GAO analysis of American Housing Survey data. | GAO-20-427

Notes: We developed two indexes to define the range of housing conditions based on American Housing Survey data. The quality index, based on 13 quality-
related variables, identified units as having no quality issues, less substantial quality issues, and substantial quality issues. The completeness index, based
on nine variables we identified as essential components of a dwelling, identified units as being incomplete or complete.

Estimates of the number of housing units in this table have a relative margin of error of +13 percent or less of the estimated number. Estimates of the percent
of housing units in this table have a margin of error of +3 percentage points or fewer, at the 95 percent confidence level.

We defined four head-of-household age categories as younger (20-34 years old), early middle age (35-49 years old), late middle age (50-64 years old), and
older (65 and older).

Page 82 GAO-20-427 Rental Housing
Appendix VI: Additional Information on Rental Housing Conditions

Table 22: GAO Quality and Completeness Indexes, Rental Units by Condition and Structure Age, 2017

Numbers in millions

Index category

Incomplete housing Substantial quality Less substantial quality No quality issues All rental units

Year built units issues issues
Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated
number percentage number percentage number percentage number percentage number percentage
2000-2017 0.11 2 0.33 5 1.46 24 4.27 69 6.17 100
1980-1999 0.26 3 0.87 10 2.28 27 5.04 60 8.44 100
1960-1979 0.28 3 1.10 12 2.65 28 5.28 57 9.32 100
1940-1959 0.17 3 0.88 16 1.53 28 2.81 52 5.39 100
1920-1939 0.12 4 0.51 16 0.98 31 1.57 49 3.17 100
Before 1920 0.13 4 0.63 2 0.99 31 1.45 45 3.20 100

Source: GAO analysis of American Housing Survey data. | GAO-20-427

Note: We developed two indexes to define the range of housing conditions based on American Housing Survey data. The quality index, based on 13 quality-
related variables, identified units as having no quality issues, less substantial quality issues, and substantial quality issues. The completeness index, based
on nine variables we identified as essential components of a dwelling, identified units as being incomplete or complete.

Estimates of the number of housing units in this table have a relative margin of error of +18 percent or less of the estimated number. Estimates of the percent
of housing units in this table have a margin of error of +5 percentage points or fewer, at the 95 percent confidence level.

Page 83 GAO-20-427 Rental Housing
Appendix VI: Additional Information on Rental Housing Conditions

Table 23: GAO Quality and Completeness Indexes, Rental Units by Condition and Unit Type, 2017

Numbers in millions

Index category

Incomplete housing Substantial quality Less substantial quality No quality issues All rental units

Structure Type units issues issues

Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated Estimated
number percentage number percentage number percentage number percentage number percentage

Multifamily 0.67 3 2.42 11 5.46 26 12.76 60 21.31 100
Small (2-9 units) 0.28 3 1.21 12 2.83 2 6.11 59 10.43 100
Large (10 or more 0.39 4 1.20 11 2.63 24 6.65 61 10.87 100
units)

Single family 0.33 3 1.71 13 4.00 31 7.05 54 13.10 100
Detached 0.25 3 1.33 14 3.05 32 4.97 52 9.61 100
Attached 0.08 2 0.38 11 0.95 27 2.08 60 3.49 100

Mobile home 0.05 4 0.21 16 0.42 33 0.60 47 1.29 100

Source: GAO analysis of American Housing Survey data. | GAO-20-427

Notes: We developed two indexes to define the range of housing conditions based on American Housing Survey data. The quality index, based on 13 quality-
related variables, identified units as having no quality issues, less substantial quality issues, and substantial quality issues. The completeness index, based
on nine variables we identified as essential components of a dwelling, identified units as being incomplete or complete.

Estimates of the number of housing units in this table have a relative margin of error of +21 percent or less of the estimated number. Estimates of the percent
of housing units in this table have a margin of error of +8 percentage points or fewer, at the 95 percent confidence level.

Page 84 GAO-20-427 Rental Housing
Appendix VI: Additional Information on Rental Housing Conditions

Figure 24: Estimated Percentage of Crowded Renter Households by Race/Ethnicity, 2001-2017

Percentage of renter households
30

25

20

15 aN

10

0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Year

=mm White
Hispanic
Other

Source: GAO analysis of American Community Survey data. | GAO-20-427

Notes: We defined a crowded household as having more than two people per bedroom. A studio apartment occupied by two people did not meet our
crowding definition.

Estimates in this figure have a margin of error of +2 percentage points or fewer, at the 95 percent confidence level.

Page 85 GAO-20-427 Rental Housing
Appendix VI: Additional Information on Rental Housing Conditions

Figure 25: Estimated Percentage of Crowded Renter Households by Age, 2001—2017

Percentage of renter households
10

=

9

3
2

0

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Year

Younger (20-34 years old)
=== Early middle age (35-49 years old)
Mmmm Late middle age (50-64 years old)
====== Older (65 years and older)

Source: GAO analysis of American Community Survey data. | GAO-20-427

Notes: We defined a crowded household as having more than two people per bedroom. A studio apartment occupied by two people did not meet our
crowding definition.

Estimates in this figure have a margin of error of +1 percentage point or less, at the 95 percent confidence level.

Page 86 GAO-20-427 Rental Housing
Appendix VI: Additional Information on Rental Housing Conditions

Figure 26: Estimated Percentage of Crowded Renter Households by Income, 2001-2017

Percentage of renter households

12
10
=
8 ~
~ =
~,
ee
2
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Year

Mam =Extremely low income

Very low income
we mee Low income

=a um== Moderate income
eeeee Higher income

Source: GAO analysis of American Community Survey data. | GAO-20-427

Notes: We defined a crowded household as having more than two people per bedroom. A studio apartment occupied by two people did not meet our
crowding definition.

Estimates in this figure have a margin of error of +1 percentage point or less, at the 95 percent confidence level.

Page 87 GAO-20-427 Rental Housing
Appendix VI: Additional Information on Rental Housing Conditions

Figure 27: Estimated Percentage of Crowded Renter Households by Locality Type, 2001-2017

Percentage of renter households

12
=
~
~
10 © a oe ~
= —_
8
6
4 ees ee
2
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Year
High-growth

ma mm Moderate-growth/high-density

Moderate-growth/moderate-density

Negative-growth
Nonmetro

Source: GAO analysis of American Community Survey data. | GAO-20-427

Notes: We defined a crowded household as having more than two people per bedroom. A studio apartment occupied by two people did not meet our
crowding definition.

Estimates in this figure have a margin of error of +1 percentage point or less, at the 95 percent confidence level.

Page 88 GAO-20-427 Rental Housing
Appendix VII: GAO Contact and Staff
Acknowledgments

GAO Contact Daniel Garcia-Diaz, (202) 512-4529 or garciadiazd@gao.gov

Staff In addition to the contact named above, Heather Chartier (Analyst in
Charge), Jeremy Anthony, Daniel Benson, Abigail Brown, Stephen

Acknowledgments Brown, Nina Daoud, Davis Judson, John McGrail, Yann Panassie, Dae

Park, Lena Richenberg, Paul Schmidt, Jennifer Schwartz, Jena Sinkfield,
Farrah Stone, and Jeff Tessin made key contributions to this report.

(102611) Page 89 GAO-20-427 Rental Housing
GAO’s Mission

Obtaining Copies of
GAO Reports and
Testimony

The Government Accountability Office, the audit, evaluation, and investigative
arm of Congress, exists to support Congress in meeting its constitutional
responsibilities and to help improve the performance and accountability of the
federal government for the American people. GAO examines the use of public
funds; evaluates federal programs and policies; and provides analyses,
recommendations, and other assistance to help Congress make informed
oversight, policy, and funding decisions. GAO’s commitment to good government
is reflected in its core values of accountability, integrity, and reliability.

The fastest and easiest way to obtain copies of GAO documents at no cost is
through our website. Each weekday afternoon, GAO posts on its website newly
released reports, testimony, and correspondence. You can also subscribe to
GAO’s email updates to receive notification of newly posted products.

Order by Phone

Connect with GAO

To Report Fraud,
Waste, and Abuse in
Federal Programs

Congressional
Relations

Public Affairs

Strategic Planning and
External Liaison

The price of each GAO publication reflects GAO’s actual cost of production and
distribution and depends on the number of pages in the publication and whether
the publication is printed in color or black and white. Pricing and ordering
information is posted on GAO’s website, https://www.gao.gov/ordering.htm.

Place orders by calling (202) 512-6000, toll free (866) 801-7077, or
TDD (202) 512-2537.

Orders may be paid for using American Express, Discover Card, MasterCard,
Visa, check, or money order. Call for additional information.

Connect with GAO on Facebook, Flickr, Twitter, and YouTube.
Subscribe to our RSS Feeds or Email Updates. Listen to our Podcasts.
Visit GAO on the web at https://www.gao.gov.

Contact FraudNet:
Website: https://Awww.gao.gov/fraudnet/fraudnet.htm
Automated answering system: (800) 424-5454 or (202) 512-7700

Orice Williams Brown, Managing Director, WilliamsO@gao.gov, (202) 512-4400,
U.S. Government Accountability Office, 441 G Street NW, Room 7125,
Washington, DC 20548

Chuck Young, Managing Director, youngc1@gao.gov, (202) 512-4800
U.S. Government Accountability Office, 441 G Street NW, Room 7149
Washington, DC 20548

James-Christian Blockwood, Managing Director, spel@gao.gov, (202) 512-4707
U.S. Government Accountability Office, 441 G Street NW, Room 7814,
Washington, DC 20548

Please Print on Recycled Paper.


File and source

File

gao-20-427.pdf

Size

2,897,593 bytes

SHA-256

ac4301a1662b759274ec86e7b12a840466be7a89692e44a22bda37cd253e3452

Our copy

gao-20-427.pdf

Original

www.gao.gov