-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-06-405		

TITLE:     Rental Housing Assistance: Policy Decisions and Market 
Factors Explain Changes in the Costs of the Section 8 Programs

DATE:   04/28/2006 
				                                                                         
----------------------------------------------------------------- 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-06-405

     

     * Report to the Chairman, Subcommittee on Housing and Community
       Opportunity, Committee on Financial Services, House of Representatives
          * April 2006
     * RENTAL HOUSING ASSISTANCE
          * Policy Decisions and Market Factors Explain Changes in the Costs
            of the Section 8 Programs
     * Contents
          * Results in Brief
          * Background
          * Increases in the Number of Vouchers Drove Growth in the Size of
            Section 8 from 1998 through 2004
               * The Number of Vouchers Increased by 31 Percent
               * The Number of Project-Based Units Declined by 5 Percent
          * Section 8 New Budget Authority Increased at a Faster Rate than
            Outlays from 1998 through 2004
               * Renewal of an Increasing Number of Expiring Contracts
                 Contributed to Much of the Growth in Budget Authority
               * Total Section 8 Outlays Grew by 50 Percent
               * Estimated Outlays for Vouchers Rose by 93 Percent, while
                 Those for the Project-Based Program Remained Relatively
                 Stable
          * Policy Decisions and Market Factors Drove Increases in Section 8
            Outlays, but HUD and Congress Have Acted to Limit Further Growth
               * Policies to Expand the Number of Assisted Households
                 Resulted in an Increase in Voucher Outlays, while the
                 Declining Number of Units Limited Growth in Project-Based
                 Outlays
               * Increases in the Average Rental Subsidy per Household
                 Contributed to Higher Outlays for Both Programs
               * Administrative and Special Program Costs Contributed
                 Modestly to the Increase in Outlays for Both Programs
               * HUD and Congress Have Taken Steps to Limit Growth in the
                 Cost of Both Programs
          * Policy Changes and Trends in Market Rents and Household Incomes
            Increased the Average Subsidy per Voucher Household
               * Growth in Market Rents
               * PHAs' Exercise of Flexibility in Setting Payment Standards
               * Slower Growth in the Average Income for Voucher Households
               * Household and Neighborhood Characteristics of Voucher
                 Holders Did Not Have a Significant Effect on the Growth in
                 the Average Rental Subsidy per Household
          * Observations
          * Agency Comments
     * Objectives, Scope, and Methodology
     * Information on the Section 8 Moderate Rehabilitation Program
     * Description of the Section 8 Rental Housing Assistance Programs
     * Data on Budgetary Costs for the Voucher and Project-Based Programs
     * Description of the Econometric Analysis of Rental Subsidy Costs per
       Household for the Voucher Program
          * Model and Data
          * Results
     * Trends in Rents and Household Payments for the Voucher and
       Project-Based Programs
          * Annual Increases in Voucher Rents and Fair Market Rents Were
            Similar
     * Comments from the Department of Housing and Urban Development
     * GAO Contact and Staff Acknowledgments

Report to the Chairman, Subcommittee on Housing and Community Opportunity,
Committee on Financial Services, House of Representatives

April 2006

RENTAL HOUSING ASSISTANCE

Policy Decisions and Market Factors Explain Changes in the Costs of the
Section 8 Programs

Contents

Tables

Figures

April 28, 2006Letter

The Honorable Robert W. Ney Chairman Subcommittee on Housing   and
Community Opportunity Committee on Financial Services House of
Representatives

Dear Mr. Chairman:

Annual appropriations for the Department of Housing and Urban
Development's (HUD) Section 8 rental housing programs have grown
significantly in recent years, climbing to nearly $21 billion in 2006.1
Authorized by the Housing and Community Development Act of 1974 (P.L.
93-383), as amended, Section 8 currently comprises the Housing Choice
Voucher (voucher) and project-based programs. While these programs are the
federal government's primary tool for making rental housing units
affordable to low-income households, they are not entitlements-that is,
not all eligible households receive assistance. HUD and Congress have
recently supported changes to the voucher program designed to limit
further growth in appropriations without reducing the number of households
that receive assistance. However, analysis of factors driving the growth
in the budgetary costs of Section 8-which could assist in such efforts-has
been limited, in part because HUD has not separately reported budget
authority or outlays for each program.2

The voucher and project-based programs share the common mission of making
housing affordable to low-income households by paying subsidies that make
up the difference between a unit's rent and the household's payment, which
is generally 30 percent of monthly income after adjustments. Despite their
shared goals, the two programs operate differently. The voucher program,
administered by state and local public housing agencies (PHA), provides
vouchers that eligible households can use to rent houses or apartments in
the private market. The subsidies in the voucher program are tied to the
household, rather than to any particular dwelling unit (i.e., they are
tenant-based). In contrast to vouchers, the project-based program is
administered by property owners who have agreed to rent specific dwelling
units to eligible households. The subsidies in the project-based program
are tied to the rental unit rather than to any specific tenant. Under both
programs, the program administrators enter into contracts with HUD that
specify the number of vouchers or units authorized to receive rental
subsidies and the duration of the subsidy payments. For the voucher
program, PHAs have flexibility to determine the maximum amount of rental
subsidy they can pay for assisted households within limits set by HUD. For
example, HUD establishes "fair market rents" for each geographic area,
based on actual market rents for standard-quality rental units, but PHAs
may choose a "payment standard" that is up to 10 percent lower or higher.

Each year, Congress appropriates budget authority to cover the cost of
Section 8 contracts. HUD uses its budget authority to incur obligations
(e.g., enter into contracts) that result in expenditures, or outlays, of
federal funds. Unlike budget authority, outlays occur when payments are
made and thus reflect the programs' actual annual cost of providing rental
assistance. Originally, Section 8 contracts were written with 5- to
40-year terms, and Congress appropriated all of the budget authority
needed to cover the projected costs of these contracts up front. Under
this approach, further appropriations were generally not needed; HUD could
make outlays to provide the subsidy in each year, using the budget
authority already appropriated, until the contract expired. However, for
expiring contracts that are renewed, appropriations of new budget
authority are needed to cover the renewed contract term. Today, contracts
are renewed and funded in 1-year increments for vouchers and either 1- or
5-year increments for the project-based program.

In response to your request, this report provides information on trends in
the size and the cost of the Section 8 programs from 1998 through 2004
(the last year for which data were available at the time of our analysis).
Specifically, the report discusses (1) the annual numbers of vouchers in
the voucher program and units in the project-based program, (2) the annual
new budget authority and outlays for each program, and (3) the factors
that

have affected outlays.3 In addition, for the voucher program-the larger of
the Section 8 programs and the focus of recent efforts to limit costs-the
report discusses factors, such as changes in market rents, that have
affected the annual average rental subsidy cost per household.

To determine the annual number of vouchers and project-based units and the
annual amounts of new budget authority and outlays for each program, we
obtained and analyzed data from HUD's budget office, annual budget
requests and other budget documents, and audited financial statements.
HUD's budget office was not able to report data on budget authority and
outlays for the voucher and project-based programs separately. We
estimated outlays using data from HUD's accounting systems, which showed
the rental assistance payments paid to PHAs and property owners under each
program from 1998 through 2004. Our report generally presents budget
authority and outlay amounts in nominal dollars; amounts adjusted for
inflation are shown in appendix IV. To identify the factors that have
affected outlays, we analyzed data on rents and household incomes from
HUD's administrative databases and reviewed reports by HUD, Congressional
Budget Office (CBO), and Congressional Research Service (CRS). To assess
the impact of different factors on the average rental subsidy cost per
household for the voucher program, we developed a statistical model using
data from HUD and the Census Bureau. Our model allowed us to estimate the
effect of each of several variables-market rents, household incomes,
household and neighborhood characteristics, and a measure of the
relationship between program policies and market rents-on the average
rental subsidy per voucher household while controlling for the effects of
other variables. Because data on certain variables used in our statistical
model were not available for 1998, our analysis of average rental subsidy
costs in the voucher program covers the period from 1999 through 2004. We
assessed the reliability of the data from HUD's accounting systems and
administrative databases by reviewing related documentation and
interviewing agency officials who work with the systems. In addition, we
performed internal checks to determine the extent to which the data fields
were populated and the reasonableness of the values contained in the
fields. We concluded that the data were sufficiently reliable for the
purposes of this report. To address all of the objectives, we interviewed
officials from HUD's Offices of the Chief Financial Officer, Public and
Indian Housing, Housing, and Policy Development and Research. We also met
with CBO and CRS officials and representatives of various industry and
research groups. We conducted our work in Washington, D.C., and Chicago,
Illinois, from April 2005 through March 2006 in accordance with generally
accepted government auditing standards. Appendix I provides additional
details on our scope and methodology.

Results in Brief

Overall, the size of the Section 8 program-that is, the combined number of
authorized vouchers and project-based units-increased annually, from 2.93
million in 1998 to 3.36 million in 2004, or 15 percent during the period.
This growth resulted exclusively from the authorization of additional
vouchers; the number of project-based units actually declined.
Specifically, the number of authorized vouchers rose from about 1.60
million in 1998 to 2.09 million in 2004, a 31 percent increase. At the
same time, the number of authorized project-based units fell from 1.33
million to 1.27 million, a 5 percent decrease, primarily because property
owners and HUD decided not to renew some project-based contracts. HUD
generally provided vouchers to households in project-based units for which
contracts were not renewed so that these households could continue
receiving rental assistance. These "tenant protection" vouchers, including
those provided to households no longer receiving assistance under other
HUD programs, accounted for 42 percent of the increase in the number of
vouchers from 1998 through 2004.

Both annual new budget authority and outlays for the Section 8 programs
increased significantly from 1998 through 2004. Appropriations of new
budget authority grew from $9.4 billion to $19.3 billion (from $10.6
billion to $19.3 billion after adjusting for inflation). This new budget
authority was primarily needed to renew expiring contracts covering
818,095 vouchers and 373,310 project-based units. Annual outlays rose from
$14.8 billion to $22.2 billion (from $16.8 billion to $22.4 billion after
adjusting for inflation). On an annual basis, total Section 8 outlays
typically exceeded new budget authority appropriated during this period
because HUD was making rental assistance payments under long-term
contracts for which Congress had appropriated budget authority in previous
years. Although HUD did not separately track outlays for the voucher and
project-based programs during this period, we estimate that outlays for
vouchers increased from $7.5 billion in 1998 to $14.5 billion in 2004, and
accounted for nearly all of the growth in total Section 8 outlays. Outlays
for the project-based program rose from an estimated $7.3 billion to $7.7
billion.

A number of policy decisions and market factors contributed to the growth
in outlays for the voucher and project-based programs from 1998 through
2004. These include increases in the total number of assisted households,
the average rental subsidy cost per household, and other program expenses.
Specifically, about 43 percent of the growth in voucher outlays was the
result of policy decisions that increased the number of assisted
households-for example, the decision to offer tenant protection
vouchers-while over half was the result of an increase in the average
annual rental subsidy cost per voucher household from $4,420 to $6,262 (a
42 percent increase, or 25 percent after adjusting for inflation). In
contrast, most of the growth in project-based outlays resulted from a
12-percent increase in the average annual rental subsidy per household
($5,305 to $5,948)-an increase that was partially offset by the declining
number of project-based units. Congress and HUD have taken steps to limit
further growth in the budgetary costs of the Section 8 programs. For
example, in 2003, Congress authorized changes to the method for
calculating the amounts of voucher funding in order to slow the growth in
both new budget authority and outlays. In addition, for the project-based
program, Congress and HUD continued steps begun in 1997 to reduce
above-market rents at some properties and to limit annual rent increases.

On the basis of our statistical analysis of cost data for the voucher
program, we found that three factors primarily accounted for the growth in
the average annual rental subsidy per voucher household from 1999 through
2004:4

o Over one-half of the total increase was explained by changes in market
rents.

o About one-quarter of the total increase was due to overall higher
payment standards-that is, by PHAs' exercising their flexibility to
increase the maximum amount of rental subsidy they can pay for assisted
households.

o About 16 percent of the increase was due to the fact that the incomes of
assisted households grew modestly during this period and did not keep up
with increases in market rents.

We also found that although household and neighborhood characteristics
were important determinants of per household rental subsidies, on average,
they did not vary enough from 1999 through 2004 to cause a substantial
change in the average per household rental subsidy.

In written comments on a draft of this report, HUD suggested that the
report include a description of its proposed legislation for reforming the
voucher program and provide additional explanations of the differences
between the average per household costs for the project-based and voucher
programs. We revised our final report in response to HUD's comments where
appropriate.

Background

Prior to the 1970s, the federal government made housing affordable to low-
and moderate-income households by subsidizing the production of privately
and government-owned properties with below-market interest rate mortgages,
direct loans, and other development subsidies. Under these production
programs, the rent subsidies were project based, and tenants received
assistance only while living in the subsidized units. In the early 1970s,
concerns were raised about the effectiveness of these programs: Many
moderate-income tenants benefited from federal assistance, while
lower-income families did not; federal costs for producing the housing
exceeded the private sector costs to produce the same services; and
allegations of waste surfaced. Interest in a more cost-effective approach
led Congress to explore options for using existing housing to shelter
low-income tenants. Section 8 of the Housing and Community Development Act
of 1974, as amended, authorized programs that reflected both approaches-a
tenant-based rental certificate program (now called the voucher program)
for use in existing housing and a project-based program. The project-based
program comprises multiple subprograms, including Section 8 New
Construction/Substantial Rehabilitation, Loan Management Set-Aside, and
Property Disposition. Appendix III contains detailed descriptions of these
subprograms.

The voucher program provides vouchers to eligible households to rent
houses or apartments in the private market from landlords who are willing
to accept the vouchers. Voucher holders are responsible for finding
suitable housing that complies with HUD's housing quality standards. The
voucher program pays the difference between the lesser of the unit's gross
rent or a local "payment standard," and the household's payment, which is
generally 30 percent of monthly income, after certain adjustments.5 To be
eligible to apply for assistance, households must have very low
incomes-less than or equal to 50 percent of area median income (AMI) as
determined by HUD. Under the provisions of the Quality Housing and Work
Responsibility Act of 1998 (P.L. 105-276), at least 75 percent of new
participants in the voucher program must be households with extremely low
incomes-at or below 30 percent of AMI.6 Households already participating
in the voucher program remain eligible for assistance as long as their
incomes do not rise above 80 percent of AMI. The voucher program is
administered by over 2,500 state and local PHAs that are responsible for
inspecting dwelling units, ensuring that rents are reasonable, determining
households' eligibility, calculating households' payments, and making
payments to landlords. HUD provides funding to PHAs for administrative
expenses as well as rental subsidies.

The project-based program subsidizes rents at properties whose owners have
entered into contracts with HUD to make rents affordable to low-income
households. Often these properties were financed with mortgages insured or
subsidized by HUD or with bonds issued by state and local housing finance
agencies. Property owners and managers are responsible for administering
the program at about 22,000 properties nationwide. The project-based
program operates much like the voucher program, paying the difference
between a HUD-approved unit rent and the household's payment, which is
generally equal to 30 percent of adjusted monthly income.7 In general,
only households with low incomes (i.e., at or below 80 percent of AMI) are
eligible for assistance, and since 1998 at least 40 percent of new
residents must have extremely low incomes. Private property owners and
managers have requirements similar to those for PHAs for administering the
project-based program-they must ensure that households meet program
eligibility requirements and must calculate households' payments. HUD pays
rent subsidies directly to the property owners but does not pay them a
separate administrative fee, as the owners' administrative costs are
reflected in the HUD-approved rents. However, because of limited staff
resources and the large number of project-based Section 8 contracts, HUD
pays contract administrators (state and local PHAs) administrative fees to
oversee most of the contracts, a task that requires processing monthly
payment vouchers, reviewing property owners' tenant information files, and
addressing health and safety issues.

Each year, Congress appropriates budget authority to cover the costs of
new Section 8 contracts, renewals of expiring contracts, amendments to
existing project-based contracts, and administrative fees.8 For the period
covered by our review (1998 through 2004), Congress appropriated funds for
the Section 8 programs in HUD's Housing Certificate Fund account.9 Over
time, Congress has changed the way it funds the Section 8 programs. From
1974 to 1983, Congress made large up-front appropriations to cover the
projected costs of multiyear Section 8 contracts. Initially, voucher
contracts were written for 5 years and were renewable, at HUD's
discretion, for up to 15 years, while the terms for project-based
contracts ranged from 15 to 40 years. When these initial contracts began
to expire in 1989, HUD required new budget authority to renew them. Owing
to budget constraints, Congress funded Section 8 contracts with amounts
that led to shorter contract terms. HUD initially renewed expiring
contracts generally for 5-year terms but starting in the mid-1990s
switched to 1-year terms for the voucher program and either 1- or 5-year
terms for the project-based program.

The Section 8 programs are not entitlements, and as a result, the amount
of budget authority HUD requests and Congress provides through the annual
appropriations process limits the number of households that Section 8 can
assist. Historically, appropriations for the Section 8 programs (as well
as for other federal housing programs) have not been sufficient to assist
all households that HUD has identified as having housing needs-that is,
households with very low incomes that pay more than 30 percent of their
income for housing, live in substandard housing, or both. According to HUD
data for calendar year 2003, Section 8 and other federal housing programs
assisted an estimated 4.3 million households, or 27 percent of all renter
households with very low incomes (see fig. 1).10 HUD estimated that over 9
million very low income households (about 59 percent) did not receive
assistance and had housing needs. Of these 9 million households with
housing needs, over 5 million had what HUD terms "worst case" needs-that
is, they paid over half of their income in rent, lived in severely
substandard housing, or both.

Figure 1: About 59 Percent of Very-Low Income Renter Households Did Not
Receive Housing Assistance and Had Housing Needs in 2003

Note: Total = 15.7 million households that rented and had very low
incomes.

Increases in the Number of Vouchers Drove Growth in the Size of Section 8
from 1998 through 2004

The combined number of authorized vouchers and project-based units grew
from about 2.93 million to 3.36 million from 1998 through 2004-an overall
increase of about 15 percent and an average annual increase of about 2
percent (see fig. 2). Most of this increase occurred from 1998 to 2001,
when about 327,000 vouchers were added. However, as figure 2 shows, this
overall trend masked a difference in the trends for the individual
programs: The number of vouchers grew by 31 percent during this period,
while the number of project-based units declined by 5 percent.

Figure 2: The Total Number of Authorized Section 8 Vouchers and Units
Increased from 1998 through 2004

Note: Totals may not add because of rounding.

It is important to note that at any given time the actual number of
households assisted with Section 8 programs is likely to be less than the
number of authorized vouchers and project-based units, because some
authorized vouchers and units may not be in use. For example, vouchers may
go unused because households may not be able to find units that meet the
program's affordability requirements and quality standards. (As discussed
subsequently in this report, the extent to which authorized vouchers are
actually used to rent units-and thus incur subsidy costs-is called the
voucher utilization rate.) Project-based units may not be in use during
the period when landlords are seeking new occupants for units that have
been vacated.

The Number of Vouchers Increased by 31 Percent

From 1998 through 2004, the number of authorized vouchers grew from about
1.60 million to almost 2.09 million, an increase of 490,944 vouchers (see
fig. 2). This increase represents an average annual growth rate of almost
5 percent. The new vouchers were composed of both "incremental vouchers"
and tenant protection vouchers. Incremental vouchers are those that
resulted from Congress' decision to expand the program to serve more
households. Notices published in the Federal Register and HUD data
indicate that the agency awarded 276,981 incremental vouchers and 205,853
tenant protection vouchers from 1998 through 2004 (see table 1).

Incremental vouchers consist of three major types: fair share,
welfare-to-work, and special purpose. Fair share vouchers are those that
HUD allocates to PHAs on a competitive basis using a formula that accounts
for poverty rates, renter populations, vacancies, overcrowding, and other
measures, in each county and independent city throughout the country.
Welfare-to-work vouchers are designated for households for which a lack of
stable, affordable housing is a barrier to employment and that are making
the transition to economic self-sufficiency.11 Finally, special purpose
vouchers include those designated for a variety of special needs
populations, such as persons with disabilities. Fair share vouchers
accounted for about 56 percent of the total, while welfare-to-work and
special purpose vouchers represented 18 percent and 26 percent,
respectively.

From 1998 through 2002, Congress provided new funding each year for a
large number of incremental vouchers to help address the unmet housing
needs of very low-income households, and fair share vouchers were the key
type of incremental vouchers used to increase the number of assisted
households. Starting in 2003, Congress provided no new funding for fair
share vouchers, but did provide new funding for a smaller number of
special purpose vouchers. By 2004, however, no new funding was provided
for any type of incremental voucher.

Unlike incremental vouchers, tenant protection vouchers do not add to the
total number of authorized units under Section 8 (and other HUD programs
for which they are used) because they replace one form of HUD assistance
with another. Tenant protection vouchers are offered to eligible
households that had received housing assistance under various HUD programs
(including the project-based program, certain HUD mortgage insurance
programs, and public housing) before the assistance was terminated. As
part of its annual budget request, HUD estimates the number of tenant
protection vouchers it will need and the amount of funding required for
these vouchers. As table 1 shows, the number of tenant protection vouchers
awarded from 1998 through 2004 remained relatively stable, from a low of
22,839 in 2002 to a high of 36,000 in 2001.

Table 1: Number of New Vouchers Awarded by Major Voucher Type, 1998-2004

                                        

        Year       
Voucher type      1998   1999    2000    2001   2002   2003   2004   Total 
Incremental     32,358 52,540  76,934  90,493 22,856  1,800      0 276,981 
Fair share           0      0  60,801  78,475 16,460      0      0 155,736 
Welfare-to-work      0 50,000       0       0      0      0      0  50,000 
Special         32,358  2,540  16,133  12,018  6,396  1,800      0  71,245 
purposea                                                           
Tenant          27,736 29,158  29,333  36,000 22,839 26,787 34,000 205,853 
protection                                                         
Subtotal                                                           482,834 
Unknownb                                                             8,110 
Total           60,094 81,698 106,267 126,493 45,695 28,587 34,000 490,944 

Source: GAO analysis of HUD data.

aSpecial purpose vouchers include Mainstream vouchers, which are targeted
to persons with disabilities, and other smaller voucher subprograms.

bFor 8,110 vouchers, or about 2 percent of the total, data indicating
voucher type were not available.

The Number of Project-Based Units Declined by 5 Percent

The number of authorized project-based units fell from 1.33 million to
1.27 million, a decline of approximately 62,000 units (see fig. 2). This
represented an average annual decrease of less than 1 percent. The number
of project-based Section 8 units declined primarily because either
property owners or HUD decided not to renew Section 8 contracts. Owners
may choose not to renew their contracts and to opt out of the program for
a variety of reasons, including plans to convert the properties to
market-rate rental units. HUD may decide not to renew some contracts if
property owners have not complied with program requirements, such as
maintaining the property in decent, safe, and sanitary condition.

If a property owner or HUD decides not to renew a project-based Section 8
contract, the property is no longer required to comply with program rules,
including affordability requirements. To protect Section 8 households from
rent increases that may result when owners opt out of their contracts, HUD
provides a special type of tenant protection voucher known as an enhanced
voucher. Enhanced vouchers are designed to ensure that tenants can afford
to remain in the properties that are no longer receiving project-based
Section 8 assistance-even if the rents for these units exceed those for
the regular voucher program (such vouchers are considered enhanced because
they allow these higher subsidies). If HUD terminates a project-based
Section 8 contract, the agency usually provides affected families with
regular vouchers to allow them to find other housing. The substitution of
tenant protection vouchers for subsidies previously paid for project-based
units has helped minimize the net loss of Section 8 units.

Section 8 New Budget Authority Increased at a Faster Rate than Outlays
from 1998 through 2004

Although both budget authority and outlays for the Section 8 programs
increased significantly from 1998 through 2004, the rates of growth
differed. Appropriations of new budget authority grew more than twofold
during this period (105 percent), partly because HUD needed more budget
authority to cover the cost of renewing long-term contracts that began to
expire in 1989. In comparison, from 1998 through 2004 total Section 8
outlays rose at a slower rate (50 percent). However, this increase masks
substantial differences in the rates of growth for the individual Section
8 programs. Although HUD did not separately track outlays for the voucher
and project-based programs during this period, we estimate that outlays
increased by 93 percent for the voucher program and by 6 percent for the
project-based program.

Renewal of an Increasing Number of Expiring Contracts Contributed to Much
of the Growth in Budget Authority

Appropriations of new budget authority for Section 8 grew from $9.4
billion in 1998 to $19.3 billion in 2004, an overall increase of about 105
percent and an average annual rate of 13 percent (see fig. 3). During
2001, new budget authority grew by 22 percent, the largest single-year
increase during this period.12 For the other years, the annual increase in
new budget authority ranged from 10 to 17 percent. Over the same period,
new budget authority for Section 8 accounted for an increasing share of
HUD's total annual appropriations, growing from 41 percent in 1998 to 54
percent in 2004.13 Part of the growth reflects the effects of inflation.
After adjusting for inflation, new budget authority rose from $10.6
billion in 1998 to $19.3 billion in 2004 (82 percent).14 Appendix IV
contains detailed information on budgetary costs in nominal and
inflation-adjusted dollars.

Figure 3: New Budget Authority for Section 8 More than Doubled from 1998
through 2004, while Outlays Grew by 50 Percent

Notes:

Because HUD does not report budget authority separately by subprograms, we
were unable to exclude the Moderate Rehabilitation program from budget
authority.

New budget authority dropped in 2000 because it did not include a $4.2
billion advance appropriation that was contained in the 2000
appropriations but was not available for obligation until 2001.

New budget authority reflects across-the-board reductions by Congress in
2001 (0.22 percent), 2003 (0.65 percent), and 2004 (0.59 percent).

Outlays for the Section 8 programs are based on our estimate of outlays
for rental assistance payments and certain administrative expenses under
the voucher and project-based programs only. The Moderate Rehabilitation
program, for example, is not included in our estimate of outlays.

We adjusted outlays for 1999 to include an advance payment of $680 million
that was made in 1998 for the 1999 voucher program and reduced outlays for
1998 by the same amount.

HUD did not separately track budget authority for the voucher and
project-based programs for the period covered by our analysis. HUD budget
officials told us they had no need to do so because Congress funded both
programs under a single budget account, the Housing Certificate Fund.
However, to provide better transparency and strengthen oversight of the
programs, Congress directed HUD to create two new budget
accounts-Tenant-Based Rental Assistance and Project-Based Rental
Assistance-for all new Section 8 appropriations.15 Beginning with its 2006
budget, HUD has provided separate information for each program.

The substantial growth in new budget authority stemmed primarily from
decisions to renew expiring long-term Section 8 contracts. From 1974 to
1983, Congress made large up-front appropriations to cover the projected
costs of multiyear Section 8 contracts that were written in those years.
Because Congress and HUD funded these long-term contracts up front, they
generally did not require new budget authority during the years specified
in the contracts.16 During the early to mid-1990s, large numbers of these
long-term contracts reached the end of their terms. Decisions to renew the
contracts created the need for new budget authority. As figure 4 shows,
the trend in the numbers of expiring contracts continued from 1998 through
2004. Specifically, the number of project-based units with expiring
contracts that were renewed grew significantly-by 373,310 units from 1998
through 2004. (As noted previously, because some project-based contracts
were not renewed, the total number of authorized project-based units
declined during this period-even as the number needing new budget
authority grew.) Additional new budget authority was required each year to
cover the renewal of 818,095 vouchers from 1998 through 2004.17

Figure 4: Renewals of Expiring Vouchers and Project-Based Units Have Grown
Significantly since 1998

A factor also contributing to the need for new budget authority was a
declining amount of "carryover" budget authority. Carryover consists of
unobligated budget authority (not yet committed to specific contracts),
including funds that have been "recovered" (de-obligated from expired
contracts that did not need all of the budget authority that had been
obligated for them). Congress may rescind any portion of such unused
budget authority and in fact enacted rescissions in the Section 8 program
during each of the years we examined.18 Total budget authority available
to renew Section 8 contracts in any year thus consists of both the
carryover, net of rescissions, as well as new budget authority, and
represents all of the funds available to HUD for future obligations and
outlays.

Typically, HUD has had large amounts of carryover funds in the Section 8
programs, and these carryover funds have helped offset the need for new
budget authority. However, as shown in figure 5, the carryover amounts
generally declined during the period we examined. For example, about $7.5
billion in carryover funds in 1998 lessened the need for new
appropriations of budget authority in that year, whereas the decline in
carryover funds in later years increased the need for new appropriations.
Partly because of declining carryover amounts during this period, total
available budget authority grew at a slower rate than new budget
authority. More specifically, total available budget authority grew from
$14.0 billion to $20.9 billion over this period (fig. 5), an average
annual rate of about 7 percent. Congress rescinded between $1.6 billion
and $2.9 billion each year during the period.

Figure 5: Total Available Budget Authority for Section 8 Grew, although
Carryover Declined from 1998 through 2004

Notes:

Total available budget authority may not add because of rounding.

New budget authority dropped in 2000 because it did not include a $4.2
billion advance appropriation that was contained in the 2000
appropriations but was not available for obligation until 2001.

Appendix IV provides detailed information on total available budget
authority in inflation-adjusted dollars.

Total Section 8 Outlays Grew by 50 Percent

As figure 3 shows, annual outlays for Section 8 programs grew from $14.8
billion in 1998 to $22.2 billion in 2004, an overall increase of about 50
percent and an average annual increase of 7 percent.19 About 78 percent of
this growth occurred from 2002 to 2004, with 2003 representing the largest
annual increase ($2.5 billion). Despite this growth, total Section 8
outlays accounted for a relatively stable share of HUD's total outlays
over this period, ranging from 45 percent in 1998 to 52 percent in 2004.20

Outlays for Section 8 generally exceeded new budget authority for the
program each year from 1998 through 2004 (see fig. 3). This pattern
resulted primarily from the way the program was originally funded. As
noted previously, initial Section 8 contracts generally had long terms and
received large up-front appropriations of budget authority to cover their
projected costs. As a result, HUD has for many years-including the 1998
through 2004 period-made outlays for contracts that have not required new
budget authority. During this period, the gap between outlays and new
budget authority narrowed as the number of expiring vouchers and
project-based units that required new budget authority grew and were
renewed on an annual basis. If all Section 8 contracts had reached the end
of their multiyear terms and were renewed annually, new budget authority
requirements would more closely approximate the expected annual outlays.

Estimated Outlays for Vouchers Rose by 93 Percent, while Those for the
Project-Based Program Remained Relatively Stable

Since HUD did not separately track outlays for the voucher and
project-based programs (for the same reasons it did not do so for budget
authority), we developed our own estimates of outlays for both programs
based on data from the accounting systems HUD uses to record Section 8
rental subsidy payments.21 On the basis of these data, we estimated that
from 1998 though 2004:

o Outlays for the voucher program rose from $7.5 billion to $14.5 billion
(fig. 6)-an overall increase of 93 percent and an average annual rate of
increase of 12 percent. The largest annual increases-approximately 20
percent-occurred both in 2002 and 2003. About 56 percent of the total
increase in outlays also occurred in these 2 years. As discussed in more
detail in a subsequent section of this report, the growth in voucher
program outlays resulted in large part from increases in the average
rental subsidy per household and decisions by Congress to expand the
number of vouchers.

o In contrast, outlays for the project-based program remained relatively
stable, rising from $7.3 billion to $7.7 billion, or about 6 percent from
1998 through 2004-an average annual rate of about 1 percent.

Figure 6: Estimated Outlays Grew Faster for Vouchers than for the
Project-Based Program from 1998 through 2004

Note: We adjusted voucher outlays for 1999 to include an advance payment
of $680 million that was made in 1998 and reduced voucher outlays for 1998
by the same amount.

Because of its much faster rate of growth, the voucher program accounted
for nearly all of the growth in total Section 8 outlays from 1998 through
2004. Specifically, the program accounted for about $7.0 billion (94
percent) of the $7.4 billion increase in total Section 8 outlays during
this period. In contrast, the project-based program accounted for only
$419 million (6 percent) of the overall increase in total Section 8
outlays. In 1998, the voucher and project-based programs each represented
about half of the total outlays for the Section 8 programs. In a
relatively short time span, voucher outlays surpassed those for the
project-based program by a significant margin, and by 2004 the voucher
program was responsible for about 65 percent of total Section 8 outlays.

Outlays for the project-based program increased at a rate slower than
inflation from 1998 through 2004. Specifically, after adjusting for
inflation, outlays dropped from $8.3 billion to $7.8 billion, a decrease
of 6 percent. The growth in voucher outlays, however, significantly
outpaced the rate of inflation, increasing from $8.5 billion to $14.6
billion (71 percent) in inflation-adjusted dollars.22 Additional
information on outlays in nominal and inflation-adjusted dollars appears
in appendix IV.

Policy Decisions and Market Factors Drove Increases in Section 8 Outlays,
but HUD and Congress Have Acted to Limit Further Growth

A number of policy decisions and market factors contributed to the growth
in total Section 8 outlays from 1998 through 2004, including decisions to
expand the number of households receiving vouchers, increases in the
average rental subsidy per household, and other program costs. Figure 7
shows the general relationship between these policy decisions and market
factors and Section 8 outlays. Although these factors also affected budget
authority, our analysis focuses on outlays because, unlike budget
authority, outlays occur when payments are made and thus reflect the
actual annual cost of providing rental assistance. Congress and HUD have
taken steps to limit further growth in Section 8 program costs-for
example, by changing the program's funding formula for vouchers.

Figure 7: Factors Affecting Section 8 Outlays

Policies to Expand the Number of Assisted Households Resulted in an
Increase in Voucher Outlays, while the Declining Number of Units Limited
Growth in Project-Based Outlays

Decisions to increase the number of households receiving vouchers were a
significant driver of growth in voucher outlays from 1998 through 2004. As
noted previously, between 1998 and 2004 Congress authorized funding for a
total of 490,944 incremental and tenant protection vouchers. This trend,
coupled with a rise in the percentage of authorized vouchers in use (the
utilization rate) that started in 2001, increased the number of assisted
households and, in turn, the amount of outlays for vouchers. We estimate
that about $3.0 billion (43 percent) of the increase in voucher outlays
from 1998 through 2004 was attributable to the additional assisted
households resulting from the authorization of new vouchers and higher
utilization rates (table 2).23

Certain policy changes were designed to increase average voucher
utilization rates. For example, starting in 2002, PHAs that applied for
fair share vouchers had to maintain utilization rates of at least 97
percent to be eligible to receive them. Also, according to HUD, Congress'
decision in 2003 to limit the funding basis for voucher contracts to only
vouchers that were actually in use effectively encouraged PHAs to increase
their utilization rates in order to receive more funding.

Table 2: Estimated Impact of Policy Decisions and Other Factors on the
Change in Outlays for the Voucher and Project-Based Programs from 1998
through 2004

                                        

             Dollars in millions                                
                                            Estimated change in 
                                                  outlays       
                   Factors                             Vouchers Project-based 
Change in the number of assisted                      $3,028        -$367a 
households/units                                             
Change in rental subsidy per householdb                3,569           616 
Change in administrative costsc                          368           170 
Total                                                 $6,966          $419 

Source: GAO analysis of HUD data.

aThis amount represents the estimated outlays that did not occur in 2004
because the number of project-based units declined. This amount partially
offset the increase in outlays caused by growth in the average rental
subsidy per household. Taken together, these two factors produced a net
increase of $250 million (-$367 million plus $616 million) in outlays, or
about 60 percent of the total change in outlays for the project-based
program.

bFor each program, these amounts were derived by taking the difference
between the (1) change in total program outlays for this period and (2)
individual changes in program outlays due to the other two factors.

cThese amounts comprise actual increases in the administrative fee for
vouchers and the cost of Performance-Based Contract Administrators for the
project-based program. Limitations in the data did not allow us to
identify other program costs.

Using the average annual household subsidy in 2004 for the project-based
program ($5,948), we estimate that the decline of about 62,000 units
reduced project-based outlays by roughly $367 million (see table 2).
However, this decrease was more than offset by the other factors, leading
to an overall increase of $419 million.

Although the decline in the number of project-based units caused outlays
for the project-based program to be less than they would have been
otherwise, its effect on total Section 8 outlays was offset to a large
degree by the issuance of tenant protection vouchers to households
displaced from their project-based units. As noted previously, under the
project-based program (or other HUD programs), tenants in units receiving
assistance that is terminated (e.g., because the unit owner decides not to
renew an expiring contract) may face higher rental payments. To protect
these tenants from potentially unaffordable rent increases and continue
providing assistance, Congress made tenant protection vouchers available.
In effect, outlays from the project-based program were shifted to the
voucher program, although not on a one-for-one basis because the per
household subsidy costs were different for project-based units and
vouchers.24

Increases in the Average Rental Subsidy per Household Contributed to
Higher Outlays for Both Programs

Increases in the average rental subsidy per household also contributed to
the growth in outlays for the voucher and project-based programs, although
the average subsidy increased more for vouchers than for project-based
programs.25 As figure 8 shows, the average subsidy for vouchers grew from
$4,420 to $6,262 from 1998 through 2004, an overall increase of 42
percent.26 The annual rate of increase in the average per household
subsidy for vouchers was 6 percent during the period, ranging from a low
of 1 percent in 1999 to a high of 11 percent in both 2002 and 2003. The
high rate of growth in 2002 and 2003 coincided with the largest yearly
increases in voucher outlays (see fig. 6). For 2004, the annual rate of
increase slowed to over 2 percent after several years of substantial
growth. The growth in the number of enhanced vouchers, which, as
previously noted, allows for higher subsidies, may have contributed to the
overall increase. As described in table 2, an estimated $3.6 billion (51
percent) of the increase in voucher outlays was due to growth in the
average rental subsidy per household.27

Figure 8: The Average Annual Rental Subsidy per Household Grew by 42
Percent for Vouchers and 12 Percent for the Project-Based Program from
1998 through 2004

Note: These averages do not include an administrative fee for vouchers or
the cost of Section 8 contract oversight for the project-based program.
However, the costs to property owners for administering the project-based
program, which are reflected in the units' rents, are accounted for in
these averages.

In comparison, the average rental subsidy per household for the
project-based program grew more modestly during the period-from $5,305 to
$5,948, an overall increase of 12 percent and an average annual increase
of 2 percent.28 The annual rate of increase in average per household
subsidy did not exceed 1 percent from 1998 through 2001 and remained at
less than 4 percent from 2002 through 2004. As described in table 2, we
estimate that this raised outlays for the project-based program by about
$616 million.29 The decline in the number of project-based units partially
offset this increase in program outlays, however.

As figure 8 shows, during the period we examined, the per household
subsidy in the voucher program was initially less than the project-based
per household subsidy but then became greater. However, this trend does
not mean that the project-based program has become more cost-effective.
Any comparison of the cost-effectiveness of these programs should account
for all subsidies received during the properties' life cycles, adjusted
for any differences in unit and household characteristics, such as the
number of bedrooms and family size.30 For example, the average
project-based subsidy per household during the period we examined did not
account for the effects of past subsidies or for potential future
subsidies that may be needed to maintain properties in the program.
Similarly, it is important to note that the nationwide trends we present
do not reflect the considerable variation that exists across local rental
housing markets. That is, even during the period we examined, in some
markets the per household subsidy for vouchers may have remained below
that for the project-based program.

For both the voucher and project-based programs, many policy decisions and
market factors influenced the average per household rental subsidy, such
as HUD's fair market rent (FMR) determinations, housing market conditions,
household incomes, and policies for limiting the cost of rental
assistance. More detailed information on the trend in the average rental
subsidy per household and the specific impact of these factors on per
household rental subsidies for vouchers are discussed in a subsequent
section of this report.

Administrative and Special Program Costs Contributed Modestly to the
Increase in Outlays for Both Programs

Other costs for program administration and special programs contributed to
the change in outlays for the voucher and project-based program, although
to a lesser extent than the other factors (see table 2). More
specifically, according to data from HUD's accounting systems,
administrative costs for vouchers increased by about $368 million from
1998 through 2004. Although complete data on administrative costs for the
project-based program were not available, a major administrative expense
was HUD's Performance-Based Contract Administrator initiative, which
started in 2000. This initiative, intended to augment HUD's oversight of
project-based Section 8 contracts, added $170 million in outlays from 1998
through 2004.

According to HUD, outlays for special programs increased but were
relatively small during the period covered by our analysis. There have
been multiple special programs, including the Family Self-Sufficiency
program, which paid for service coordinators to help participating
families achieve economic independence. The Family Self-Sufficiency
program accounted for about $50 million in outlays in 2004. Since detailed
data on the outlays for special programs were not readily available for
this period, we were unable to comprehensively estimate their impact on
outlays.

HUD and Congress Have Taken Steps to Limit Growth in the Cost of Both
Programs

HUD has implemented measures to limit increases in the cost of the Section
8 programs. For example, as noted previously, in 2003 Congress authorized
changes to HUD's policies for funding vouchers to slow the growth in new
budget authority and, in turn, outlays. Before 2003, Congress appropriated
budget authority using a unit-based approach that covered all vouchers
authorized in each contract, whether or not all of the vouchers had been
utilized. Concerned that appropriations were exceeding actual program
needs, Congress changed the formula for funding voucher contracts to a
dollar-based approach, basing it on actual expenditures from the previous
year plus an inflation factor. In addition, Congress authorized a
contingency fund to cover increases in rental costs in excess of the
inflation factor.

In HUD's 2004 budget, Congress authorized the creation of a Quality
Assurance Division within HUD to provide more oversight of the
administration and cost of the voucher program. A key part of this effort
involves monitoring and verifying program costs reported by PHAs. The
division also audits PHAs' program records to ensure that voucher costs
were reported accurately and monitors local rental market trends to
determine whether HUD's FMRs were set too high or too low. In addition,
quality assurance staff review PHAs' compliance with HUD's requirement
that rents for voucher units be reasonable-that is, comparable to rents
for similar unassisted units in the market.

Congress and HUD have taken further steps since the period of our analysis
to limit cost growth. For example, Congress made further changes to the
voucher's dollar-based formula in 2005 that eliminated all contingency
funding, so that PHAs were expected to absorb all additional cost
increases during the year. To help PHAs keep their costs within their
funding levels, HUD issued guidance in 2005 concerning options PHAs could
exercise to limit costs.31 These options included the following.

o Reduce payments standards: Because PHAs may set their own payment
standards-that is, the maximum rent that can be used to calculate rental
subsidies-anywhere between 90 and 110 percent of the FMR for their area,
reducing payment standards allow PHAs to limit growth in rental subsidy
payments.

o Ensure reasonable rents: Statute and HUD regulations require PHAs to
compare rents for voucher units to those for comparable unassisted units
and reduce rents for voucher units if warranted.32 To ensure that rents
are reasonable, PHAs can conduct more frequent reviews of rents charged by
landlords. Any rent reductions would reduce the rental subsidy payments
that PHAs make.

o Deny moves within and outside PHA jurisdiction: The voucher program
allows households to move anywhere within and outside of a PHA's
jurisdiction. However, if a PHA has insufficient funding, it can deny a
voucher household's move to an area that would result in higher subsidy
costs-for example, an area with a higher payment standard.

o Not reissue vouchers or terminate assistance: Vouchers can become
available to new households when assisted households leave the program
(turnover). To limit costs, PHAs can choose not to reissue turnover
vouchers or pull back outstanding vouchers for other unassisted households
searching for housing. PHAs can also terminate assistance if they
determine that the funding provided by HUD is insufficient, although
according to HUD, the department is not aware of any instance in which a
PHA has terminated voucher assistance.

o Set higher minimum rents: HUD policy allows PHAs to set a minimum rent
for households that can range from as low as $0 to as high as $50. Some
PHAs currently allow certain households with very little income to pay
rents that are below the minimum rent ceiling (i.e., less than $50). To
reduce their costs, these PHAs can raise the minimum rent to $50.

Furthermore, HUD supports proposed legislation-the State and Local Housing
Flexibility Act of 2005-that would replace the existing voucher program
with the "flexible voucher program."33 This proposed program would, among
other things, allow individual PHAs to set (within broad federal
guidelines) eligibility requirements, the maximum period that a household
could receive assistance, and households' contributions toward rents.
According to HUD, this proposed program, which would initially continue to
fund vouchers using the dollar-based approach, would create incentives and
provide flexibilities for PHAs to manage their funds in a cost-effective
manner.34

For the project-based program, Congress has taken steps to control the
cost of rental subsidies, and as our analysis shows, these steps have
limited growth in the program's average rental subsidy per household and
thus in outlays. In 1997, Congress passed the Multifamily Assisted Housing
Reform and Affordability Act, which established the Mark-to-Market
program. When properties entered the project-based program in the late
1970s through the mid-1980s, HUD often subsidized rents that were above
local market levels to compensate for high construction costs and
program-related administrative expenses. Thereafter, these rents were
adjusted annually using an operating cost factor determined by HUD. In the
early 1990s, HUD concluded that the continued growth in subsidy levels
would be unsupportable within HUD's budget limitations. The Mark-to-Market
program, which began in 1998, authorized HUD to reduce rents to market
levels on project-based properties with HUD-insured

mortgages.35 According to HUD, the program has reduced project-based
rental subsidy costs at over 2,700 properties by an estimated $216 million
per year since 2000.36

Policy Changes and Trends in Market Rents and Household Incomes Increased
the Average Subsidy per Voucher Household

We developed a statistical model to assess the impact that certain
variables-specifically, market rents, payment standards, household
incomes, and household and neighborhood characteristics-had on the change
in the average rental subsidy per household for the voucher program.37
Changes in market rents explained a significant part of the increase in
the average rental subsidy per household. Specifically, we estimate that
from 1999 through 2004, over one-half of the increase in the average per
household subsidy was explained by higher market rents, all other things
being equal.38 Higher payment standards and the relatively slow growth in
household incomes also contributed to the increase. Although we found that
household and neighborhood variables were important determinants of per
household rental subsidies, their average values did not vary enough from
1999 through 2004 to cause a significant change in the average per
household rental subsidy over this period.

Growth in Market Rents

Because voucher households rent units in the private market, trends in
market rents have a major effect on per household rental subsidies. To
assess the impact of market rents on per household rental subsidies, we

used HUD's FMRs as indicators of local market rents.39 Our model estimated
the average per household subsidy that HUD paid in each year (baseline
estimate).40 We then used the model to estimate the average per household
subsidy HUD would have paid in each year, had the average market rents
remained at the 1999 level, adjusted for overall price level changes.
Comparing this figure with the baseline estimate indicates the influence
of changes in rents. We estimate that from 1999 through 2004 the average
annual rental subsidy per household would have grown from $5,225 to $5,800
(an increase of 11 percent), if the average market rents had remained at
1999 levels, compared with the 24 percent growth, from $5,225 to $6,478,
in the baseline estimate (fig. 9).41 Expressed differently, the effect of
market rents accounted for over half of the increase in the average per
household subsidy, all other things being equal.42

Figure 9: Growth in Market Rents Had a Significant Impact on the Estimated
Average Rental Subsidy per Household

PHAs' Exercise of Flexibility in Setting Payment Standards

In 1998, the Quality Housing and Work Responsibility Act (P.L. 105-276)
authorized PHAs to set local payment standards anywhere between 90 to 110
percent of the FMR without the need for prior HUD approval. This
flexibility was intended to make it easier for voucher households to find
housing successfully, reduce concentrations of poverty by helping voucher
households find housing in neighborhoods with higher incomes, and allow
PHAs to respond to local market conditions. The result of this policy was
that the average payment standard, as a percentage of the FMR, increased
from about 96 percent in 1999 to 103 percent in 2004. The average voucher
rent as a percentage of the FMR also increased, rising from about 94
percent in 1999 to 97 percent in 2004 (see app. VI for detailed discussion
of the trends in voucher rents).

To assess the impact of higher payment standards on the change in per
household rental subsidies, we compared our baseline estimate with the
average per household subsidy that our model predicted HUD would have paid
in each year had the average payment standard, as a percentage of the FMR,
remained at its 1999 value. As shown in figure 10, we estimate that over
this period the average per household subsidy would have grown from $5,225
to $6,169 (an 18 percent increase) if the average payment standard as a
percentage of the FMR had remained at the 1999 level, compared with the 24
percent growth, from $5,225 to $6,478, in the baseline estimate. Further,
we estimate that the impact of higher payment standards accounted for
about one-quarter of the increase in the average per household subsidy
from 1999 through 2004, all other things being equal.43

Figure 10: Estimated Growth in the Average Rental Subsidy per Household
Would Have Been Less Had the Average Payment Standard Remained at the 1999
Level

Slower Growth in the Average Income for Voucher Households

Slow growth in household incomes, which did not keep pace with the
increases in market rents, also contributed to higher per household rental
subsidies. Specifically, from 1999 through 2004, the average income of
voucher households grew from $8,779 to $10,086, an overall increase of 15
percent and an average annual rate of about 3 percent. However, market
rents, as measured by FMRs, increased by about 23 percent over this
period, or an average annual rate of over 4 percent. To determine the
impact of household income on the change in per household rental
subsidies, we compared the baseline estimate with the estimated amount
that our model predicted HUD would have paid had the average household
income grown at the same rate as the average market rent. As shown in
figure 11, we estimate that over this period the average per household
subsidy would have grown from $5,225 to $6,279 (an increase of 20 percent)
if the average income had grown as fast as the average market rent,
compared with the 24 percent growth, from $5,225 to $6,478, in the
baseline estimate. Further, we estimate that the effect of relatively slow
growth in the average household income accounted for about 16 percent of
the increase in the average per household subsidy, all other things being
equal.44

Figure 11: Had Household Income Grown as Fast as Market Rents, Growth in
the Estimated Average Rental Subsidy per Household Would Have Been Less

Household and Neighborhood Characteristics of Voucher Holders Did Not Have
a Significant Effect on the Growth in the Average Rental Subsidy per
Household

We analyzed certain household characteristics, such as family size, family
types (for example, whether the household was headed by an elderly person
or a person with a disability), and others, and found that, while they
were major determinants of per household rental subsidies, they did not
vary enough over this period to effect significant change in the average
per household rental subsidy. Stated differently, these factors exhibited
about the same influence on per household voucher subsidies throughout the
period, and thus do not help explain the overall trend of increased rental
subsidy.

In addition, we analyzed the characteristics of the neighborhoods-also
important determinants of per household subsidies-where voucher

holders live.45 Specifically, given the significant increases in voucher
rents and payment standards, we explored the extent to which the increase
in the average per household subsidy was the result of voucher households
moving to neighborhoods with less poverty and other favorable
characteristics. However, just as with the household characteristics, the
average values of these variables did not vary enough from 1999 through
2004 to cause a substantial change in the average per household rental
subsidy over this period. Because we did not have comprehensive data on
the quality of rental units in the voucher program, we could not explore
whether the trends in higher voucher rents and payment standards were also
accompanied by changes in the quality of units occupied by voucher
holders.

Observations

The cost of providing rental assistance has been a long-standing issue for
policymakers and has led Congress, on different occasions, to reform
various housing programs. Recent proposals for reform have focused on the
voucher program, which experienced a significant growth in outlays and
constituted nearly all of the increase in total Section 8 outlays from
1998 through 2004. We found that the growth both in the number of assisted
households-driven largely by policy decisions to expand this
nonentitlement program-and in the average rental subsidy per household
explain much of the increase in voucher outlays over this period. In turn,
the average per household subsidy rose in large part because of changes in
the rental market, use of higher payment standards by PHAs, and household
incomes that grew more slowly than rents.

To the extent that policymakers wish to stem the rising cost of the
voucher program, our analysis suggests that future increases could be
mitigated by reducing the number of assisted households, lowering payment
standards, requiring households to pay a larger share of their incomes
toward rent, subsidizing households with higher incomes, or a combination
thereof. However, these actions require making difficult trade-offs
between limiting program costs and achieving long-standing policy
objectives, such as serving more needy households, having assisted
households pay a relatively small share of their incomes in rents, making
it easier for voucher holders to find housing (especially in tight rental
markets), reducing the concentration of poverty, and giving PHAs the
flexibility to respond to local rental market conditions.

Congress and HUD have already responded to the increasing cost of vouchers
by changing the way the program is funded. Specifically, HUD no longer
provides funding to PHAs based on the number of authorized vouchers, but
rather based on the prior year's level of voucher expenditures, adjusted
by an inflation factor. While this approach allows HUD to limit the annual
rate of increase in the program's cost, it does not directly address the
policy decisions and market factors that we identified as contributing to
the increase in program costs. Instead, it will be up to PHAs to exercise
their flexibilities and make decisions regarding how to use the voucher
funding that they receive from HUD. For example, some PHAs may choose to
reduce their local payment standard, a course that, as our analysis
suggests, would likely limit growth in voucher costs. The decisions that
PHAs make will eventually influence trends in outlays, per household
subsidies, and unit rents, and these trends will become more apparent in
the years following the period covered by our analysis.

Agency Comments

We provided HUD with a draft of this report for review and comment. In a
letter from the Acting Deputy Chief Financial Officer (see app. VII), HUD
suggested technical clarifications, which we incorporated where
appropriate, and made the following comments:

o HUD noted that the draft report's discussion of efforts to limit growth
in program costs did not cite the department's recent legislative
proposal-the State and Local Housing Flexibility Act-to reform the voucher
program. The proposal's primary mechanism for limiting cost growth is the
continued implementation of a dollar-based approach for funding the
voucher program. Our draft report discussed the dollar-based approach and
its intended impact on program costs. However, in response to HUD's
comment, we added language to the final report describing the
legislation's key provisions and objectives.

o HUD indicated that the draft report was incorrect in stating that to be
eligible for assistance under the voucher program, households must have
very low incomes-less than or equal to 50 percent of AMI. HUD said that
households must have low incomes-less than or equal to 80 percent of
AMI-to be eligible. The income limit that HUD referred to generally
applies to households already participating in the voucher program. The
income limit cited in our draft report referred to the eligibility
criteria for new applicants. We revised the final report to make this
distinction clearer.

o HUD said that our draft report's discussion of the growth in
appropriations from 1998 through 2004 that was due to expiring Section 8
contracts may have inadvertently cited 1989 (rather than 1998) as the year
in which contracts began to expire. Based on our analysis of prior studies
on this issue, 1989 is generally regarded as the year in which Section 8
contracts started to expire. Contracts that expired, and were renewed with
shorter terms in 1989 and afterwards, required new appropriations for
renewals in subsequent years, including the years covered by our analysis.
Accordingly, we made no changes to the final report.

o Finally, HUD stated that the draft report did not mention a critical
reason that the lower cost per unit in project-based programs did not
imply greater cost effectiveness-specifically, that vouchers are used for
units that, on average, have more bedrooms and serve larger households
than project-based units. In response to HUD's comments, we revised the
final report to reflect the fact that determining the cost-effectiveness
of HUD's housing programs must account for not only all subsidies received
over time but also unit and household characteristics.

As agreed with your office, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 30 days from
the report date. At that time, we will send copies to the Secretary of
Housing and Urban Development and other interested congressional
committees. We will also make copies available to others upon request. In
addition, this report will be available at no charge on the GAO Web site
at http://www.gao.gov .

If you have any questions about this report, please contact me at (202)
512-8678 or [email protected] . Contact points for our Office of Congressional
Relations and Public Affairs may be found on the last page of this report.
GAO staff who made major contributions to this report are listed in
appendix VIII.

Sincerely yours,

David G. Wood Director, Financial Markets and Community Investment

Objectives, Scope, and MethodologyAppendix I

This report provides information on trends in the size and cost of the
Department of Housing and Urban Development's (HUD) Section 8 program from
1998 through 2004. Specifically, our report objectives were to determine
(1) the annual numbers of vouchers in the voucher program and units in the
project-based programs, (2) the annual new budget authority and outlays
for each program, (3) the factors that have affected outlays, and (4) the
impact of factors on the average rental subsidy cost per household for the
voucher program.

To determine the annual numbers of vouchers in the voucher program and
units in the project-based program, we obtained and reviewed data on the
numbers of authorized vouchers and project-based units from 1998 through
2004 from HUD's budget office. We compared the annual numbers of vouchers
and project-based units that HUD provided with information reported in the
agency's annual budget requests to ensure that they were consistent. We
obtained data on the number of units authorized under the Section 8
Moderate Rehabilitation program from HUD's program offices. We compiled
and analyzed HUD notices of funding announcements and awards published in
the Federal Register to determine the different types of new vouchers that
were added to the program.

To determine the annual amount of new budget authority and outlays for
each program, we obtained and analyzed data from HUD's budget office,
annual budget requests and other budget documents, and audited financial
statements. We also reviewed relevant prior reports from HUD, HUD's Office
of Inspector General (OIG), the Congressional Budget Office (CBO), and the
Congressional Research Service (CRS). Because HUD's budget office was not
able to report data on outlays for the voucher and project-based programs
separately, we obtained data on rental assistance payments from HUD's
accounting systems and estimated the amount of rental assistance payments
paid to public housing agencies (PHA) and property owners under each
program from fiscal years 1998 through 2004. Specifically, from the HUD
Central Accounting and Program System (HUDCAPS), we obtained information
on rental assistance payments and other expenses for the voucher and the
Section 8 Mod Rehab program, as well as for a limited number of contracts
for the project-based program. From HUD's Program Accounting System (PAS),
we obtained similar information for the remaining project-based Section 8
contracts. In total, the data we used comprised approximately 3 million
payment records. Our analysis included payment records associated with the
voucher and project-based programs only and did not include payment
records for other HUD rental assistance programs, such as the Section 202
Supportive Housing for the Elderly and Section 811 Supportive Housing for
Persons with Disabilities programs. We included payment records for
certain administrative expenses, such as fees paid to PHAs for the voucher
program and to Performance-Based Contract Administrators for the
project-based program.

We compared our estimate of outlays for the voucher, project-based, and
Mod Rehab programs and other related expenses (total outlays) with
published totals in HUD's annual budget requests. Our estimates using
HUDCAPS and PAS were, on average, 0.7 percent less than the totals in
HUD's annual budget requests. For 1998 and 1999, our estimate of total
outlays varied from the published totals by -1.2 percent and -4.2 percent,
respectively. For 2000 through 2004, our estimates of total outlays were
within 0.4 percent. One reason for the variation between our estimates and
the published totals is that our analysis did not include certain
nonrental assistance activities paid for with Section 8 funds.

In order to assess the reliability of the data from HUDCAPS and PAS, we
reviewed related documentation and interviewed agency officials who work
with these databases. In addition, we performed internal checks to
determine the extent to which the data fields were populated and the
reasonableness of the values contained in the fields. We concluded that
the data were sufficiently reliable for the purposes of this report.

To identify the factors that have affected outlays, we analyzed our
reports and reports by HUD, CBO, CRS, transcripts of congressional
committee hearings, and congressional committee reports. We also obtained
and analyzed data on rental subsidies per household, a key factor
affecting outlays, from two HUD databases-the Public and Indian Housing
Information Center (PIC) for the voucher program and the Tenant Rental
Assistance Certification System (TRACS) for the project-based program.
Using these data, we analyzed trends in unit rents, household incomes, and
household rental payments. In order to assess the reliability of the data
from PIC and TRACS, we reviewed related documentation and interviewed
agency officials who work with these databases. In addition, we performed
internal checks to determine the extent to which the data fields were
populated and the reasonableness of the values contained in the fields. We
concluded that the data were sufficiently reliable for the purposes of
this report.

To assess the impact of different factors on the average rental subsidy
cost per household for the voucher program, we developed a statistical
model using data from HUD and the Census Bureau. Specifically, we obtained
household-level data from PIC on the rental subsidies per household, unit
rents, household incomes, various demographic characteristics, and
geographic information about where households were located. We also
incorporated information from the 2000 Decennial Census and HUD on
neighborhood characteristics at the census tract level. Our model allowed
us to estimate the effect of each variable-market rents, household
incomes, household and neighborhood characteristics, and a measure of the
relationship between the payment standard and HUD's fair market rent-on
the average rental subsidy per voucher household, while controlling for
other variables. The PIC data for 1998 did not have complete information
for certain fields (such as the fair market rent associated with an
individual household), and consequently, we did not include data for 1998
in our model. Appendix V contains further information on the results of
our statistical analysis.

To address all of the objectives, we interviewed officials from HUD's
Offices of the Chief Financial Officer, Public and Indian Housing,
Housing, and Policy Development and Research. We also met with CBO and CRS
officials and representatives of various industry and research groups: the
Center for Budget and Policy Priorities, the Council of Large Public
Housing Authorities, the National Leased Housing Association, and the
National Low Income Housing Coalition. We conducted our work in
Washington, D.C., and Chicago, Illinois, from April 2005 through March
2006 in accordance with generally accepted government auditing standards.

Information on the Section 8 Moderate Rehabilitation ProgramAppendix II

This appendix provides information on the Section 8 Moderate
Rehabilitation (Mod Rehab) program. The Mod Rehab program was created in
1978 to add to the existing stock of assisted housing. It did this by
providing funding to upgrade a portion of the estimated 2.7 million
then-unassisted rental housing units with deficiencies that required a
moderate level of repair, and rental subsidies for low-income households
to live in them. Congress funded no new contracts for the Mod Rehab
program after 1989 and repealed the program in 1991.

Under annual contracts with public housing agencies (PHA) that administer
the Mod Rehab program, HUD provides the funding for rental subsidies as
well as an administrative fee to the agencies. The administering agencies,
in turn, enter into contracts with property owners. Under these contracts,
property owners rehabilitate their housing units to meet HUD's standards
for housing quality by completing repairs costing at least $1,000 and make
the rehabilitated units available to eligible households. In exchange,
PHAs screen applicants for eligibility and pay the difference between the
approved contract rent and the household's portion of the rent. The Mod
Rehab has features that are common to both the project-based and voucher
programs. For example, similar to the voucher program, the Mod Rehab
program is administered by PHAs and was intended to utilize the existing
stock of privately owned rental housing. However, Mod Rehab is
fundamentally a project-based program because the rental subsidy is tied
to a specific unit, not the household.

During the 11 years that Congress funded new contracts under the Mod Rehab
program, the term for the Section 8 contracts was 15 years. When the
oldest of these contracts began to expire in 1995 and 1996, HUD instructed
PHAs to replace them with vouchers. Since fiscal year 1997, however, HUD
has renewed expiring contracts on an annual basis if the owners opt to do
so and the properties consist of more than four rental units.

As shown in table 3, the Mod Rehab program has undergone significant
reductions in the number of units-from 71,659 in 1998 to 34,141 in 2004, a
decline of about 52 percent. As with project-based Section 8, owners of
Mod Rehab properties can choose to leave the program upon contract
expiration, and in these cases, eligible households can receive enhanced
vouchers.

Table 3: Number of Authorized Units under the Section 8 Moderate
Rehabilitation Program, 1998-2004

                                        

                Year                                         Authorized units 
1998                                                                71,659 
1999                                                                64,463 
2000                                                                57,777 
2001                                                                52,342 
2002                                                                49,013 
2003                                                                42,504 
2004                                                                34,141 

Source: HUD.

Data on budget authority for the Mod Rehab program were not available
separately. From 1998 through 2004, HUD received budget authority for the
Mod Rehab program as part of the overall appropriations for Section 8 in
the Housing Certificate Fund account. Starting in its 2006 budget request,
HUD included renewal funding for the Mod Rehab program in its
Project-Based Rental Assistance budgetary account. Similarly, data on Mod
Rehab outlays were not available. However, as we did for the voucher and
project-based programs, we estimated Mod Rehab outlays using data from
HUD's accounting systems. As table 4 shows, from 1998 through 2004,
estimated Mod Rehab outlays decreased by over 50 percent, from $472
million to $246 million. The decrease in outlays was due to significant
reductions in the number of units assisted under the program.

Table 4: Estimated Outlays for the Section 8 Moderate Rehabilitation
Program, 1998-2004

                                        

     Dollars in millions   
                            Estimated outlays   
Year                         Nominal dollars    Inflation-adjusted dollars 
1998                                    $472                          $536 
1999                                     364                           408 
2000                                     351                           386 
2001                                     332                           358 
2002                                     305                           322 
2003                                     280                           289 
2004                                     246                           248 

Source: GAO analysis of HUD Central Accounting and Program System.

Description of the Section 8 Rental Housing Assistance Programs Appendix
III

Federal rental housing assistance, which began with the enactment of the
U.S. Housing Act of 1937, includes subsidies to construct new affordable
housing and to make rents affordable in existing rental housing. From 1937
through 1974, the emphasis was almost exclusively on new construction.
Questions about the cost-effectiveness of new construction led Congress to
explore options for using existing housing to shelter low-income families.
In 1974, it added Section 8 to the U.S. Housing Act of 1937 and created
the Existing Housing Certificate program, the first major program to rely
on existing privately owned rental housing and to provide tenant-based,
rather than project-based, assistance. Another type of Section 8
assistance, the voucher program, started as a demonstration program in
1983, was made permanent in 1988, and operated simultaneously with the
certificate program until 1998. At that time, the two programs were
consolidated into the Housing Choice Voucher program, which combined
features of both earlier programs. This program is now the largest federal
housing assistance program. Table 5 summarizes the Section 8 rental
housing assistance programs, including their authorization date and
current status.

Table 5: HUD Programs with Section 8 Rental Assistance, in Order of Year
Authorized

                                        

Program         Type of subsidy Year       Status        Description       
                                   authorized               
Section 202     Project-based:  1959       No new        Provides direct   
Elderly and                                commitments   loans at          
Disabled        direct loan                since 1991    below-market      
Housing Direct  with                                     rates for up to   
Loan Program    below-market                             40 years to       
                   interest rates,                          finance the       
                                                            construction of   
                   rental                                   rental housing    
                   assistance                               for the elderly   
                   payments                                 and disabled. All 
                   generally                                projects built    
                   through Section                          since 1974 also   
                   8                                        receive Section 8 
                                                            rent subsidies.   
Section 8 New   Project-based:  1974       No new        Provides rent     
Construction                               commitments   subsidies in new  
and Substantial rental                     since 1983,   or substantially  
Rehabilitation  assistance                 except for    rehabilitated     
                   payment,                   Section 202   projects. Subsidy 
                                              program (see  initially covered 
                   below-market               above)        the difference    
                   interest rate                            between tenants'  
                   loansa                                   payment and fair  
                                                            market rent,      
                                                            determined by     
                                                            HUD. Subsidy      
                                                            contracts were    
                                                            for 20 to 40      
                                                            years. Tax        
                                                            incentives and    
                                                            financing         
                                                            arrangements also 
                                                            may reduce        
                                                            owners' effective 
                                                            mortgage interest 
                                                            rates and project 
                                                            rents. Current    
                                                            restructuring of  
                                                            ongoing contracts 
                                                            will result in    
                                                            realignment of    
                                                            subsidy payments. 
Section 8 Loan  Project-based:  1974       No new        Provides          
Management                                 commitments   subsidies to      
Set-aside and   rental                                   units in          
Property        assistance                               financially       
Disposition     payment                                  troubled projects 
                                                            in the            
                                                            FHA-insured       
                                                            inventory and on  
                                                            the sale of       
                                                            HUD-owned         
                                                            projects.         
                                                            Subsidies ensure  
                                                            improved cash     
                                                            flows and         
                                                            preserve projects 
                                                            for lower-income  
                                                            tenants.          
                                                            Subsidies cover   
                                                            the difference    
                                                            between tenant    
                                                            payments and unit 
                                                            rents, which      
                                                            often are below   
                                                            market rates      
                                                            because of other  
                                                            federal           
                                                            subsidies.        
Section 8       Tenant-based:   1974       Merged in     Aids low-income   
Existing                                   1998 with the households to     
Housing         rental                     voucher       rent housing      
Certificates    assistance                 program to    units in the      
                   payment                    become the    market. Rent      
                                              Housing       cannot exceed the 
                                              Choice        HUD-established   
                                              Voucher       fair market rent  
                                              program       for the           
                                                            geographical      
                                                            area. HUD pays    
                                                            the difference    
                                                            between the       
                                                            actual unit rent  
                                                            and the tenant    
                                                            payment.          
                                                            Administered by   
                                                            local public      
                                                            housing           
                                                            authorities,      
                                                            which enter into  
                                                            contracts with    
                                                            landlords.        
Section 8       Project-based:  1978       No new        Provides rental   
Moderate                                   commitments   subsidies to      
Rehabilitation  rental                     since 1989    units in          
                   assistance                               privately owned   
                   payment                                  properties where  
                                                            the owners agreed 
                                                            to make up to     
                                                            $1,000 per unit   
                                                            in repairs in     
                                                            order to receive  
                                                            rental            
                                                            assistance.       
                                                            Although the      
                                                            program was       
                                                            repealed in 1991, 
                                                            property owners   
                                                            may request       
                                                            1-year renewals   
                                                            of existing       
                                                            contracts. Unlike 
                                                            project-based     
                                                            Section 8, Mod    
                                                            Rehab relied on   
                                                            existing private  
                                                            housing and was   
                                                            administered by   
                                                            public housing    
                                                            authorities.      
                                                            However, like     
                                                            project-based     
                                                            Section 8, rental 
                                                            assistance under  
                                                            Mod Rehab is tied 
                                                            to the unit, and  
                                                            a household can   
                                                            benefit from the  
                                                            subsidy only if   
                                                            it remains in the 
                                                            unit.             
Section 8       Tenant-based:   1983       Merged in     Similar to the    
Vouchers                                   1998 with     Section 8         
                   rental                     Existing      Certificate       
                   assistance                 Housing       program in that   
                   payment                    Certificates  assisted          
                                              to become the households could  
                                              Housing       live in privately 
                                              Choice        owned units, and  
                                              Voucher       public housing    
                                              program       authorities       
                                                            administered the  
                                                            program. Unlike   
                                                            the certificate   
                                                            program in that   
                                                            recipients could  
                                                            occupy units      
                                                            whose rents       
                                                            exceeded the      
                                                            voucher payment   
                                                            standard-roughly  
                                                            equivalent to the 
                                                            fair market       
                                                            rent-if they paid 
                                                            the difference.   
                                                            If rents were     
                                                            below the payment 
                                                            standard,         
                                                            households could  
                                                            keep the          
                                                            difference (also  
                                                            known as the      
                                                            shopper's         
                                                            incentive).       
Housing Choice  Tenant-based:   1998       Active        Aids low-income   
Voucher Program                                          households to     
                   rental                                   rent housing      
                   assistance                               units in the      
                   payment                                  market. Public    
                                                            housing           
                                                            authorities have  
                                                            discretion to set 
                                                            voucher payment   
                                                            standard anywhere 
                                                            between 90 and    
                                                            110 percent of    
                                                            the fair market   
                                                            rent. HUD pays    
                                                            the difference    
                                                            between the       
                                                            payment standard  
                                                            (or, if less, the 
                                                            unit's rent) and  
                                                            the total tenant  
                                                            payment, which is 
                                                            usually at least  
                                                            30 percent of     
                                                            adjusted          
                                                            household income. 
                                                            If the unit's     
                                                            rent exceeds the  
                                                            payment standard, 
                                                            the tenant can    
                                                            pay the           
                                                            difference,       
                                                            provided that     
                                                            household initial 
                                                            rent burden does  
                                                            not exceed 40     
                                                            percent of        
                                                            adjusted income). 

Source: GAO.

aThe subsidy is provided by another housing program.

Data on Budgetary Costs for the Voucher and Project-Based ProgramsAppendix
IV

This appendix provides detailed data on total available budget authority
and outlays for the Section 8 programs. Since we are evaluating budget
trends over a 7-year period, we present the budgetary data in both nominal
(current) and inflation-adjusted dollars. We use the gross domestic
product (GDP) index to adjust for inflation and 2004 as the reference
year.

Table 6: Total Available Budget Authority in Nominal Dollars, 1998-2004

                                        

Dollars in  
    millions   
Fiscal year New budget Carryover Rescission         Total available budget 
                authority                                           authority 
1998            $9,373    $7,542    -$2,897                        $14,018 
1999            10,327     6,366     -2,000                         14,692 
2000             7,177     7,721     -2,243                         12,655 
2001            18,110     5,375     -1,947                         21,538 
2002            16,281     4,093     -1,589                         18,786 
2003            17,112     3,549     -1,600                         19,060 
2004            19,257     4,439     -2,844                         20,852 

Source: HUD.

Note: Totals may not add because of rounding.

Table 7: Total Available Budget Authority in Inflation-Adjusted Dollars,
1998-2004

                                        

Dollars in  
    millions   
Fiscal year New budget Carryover Rescission         Total available budget 
                authority                                           authority 
1998           $10,552    $8,491    -$3,262                        $15,782 
1999            11,476     7,075     -2,223                         16,328 
2000             7,817     8,410     -2,443                         13,784 
2001            19,272     5,720     -2,072                         22,919 
2002            17,000     4,274     -1,659                         19,616 
2003            17,521     3,633     -1,638                         19,516 
2004            19,257     4,439     -2,844                         20,852 

Source: HUD.

Note: Totals may not add because of rounding.

Table 8: Outlays for the Section 8 Programs, 1998-2004

                                        

     Dollars in millions   
Fiscal year                 Nominal dollars     Inflation-adjusted dollars 
1998                                $14,773                        $16,794 
1999                                 15,306                         17,148 
2000                                 15,681                         17,267 
2001                                 16,488                         17,734 
2002                                 18,235                         19,225 
2003                                 20,715                         21,414 
2004                                 22,159                         22,372 

Source: GAO analysis of data from the HUD Central Accounting and Program
System (HUDCAPS) and Program Accounting System (PAS).

Table 9: Outlays for the Voucher and Project-Based Programs, 1998-2004

                                        

Dollars  
      in    
millions 
Fiscal   Voucher                    Project-based 
year     Nominal Inflation-adjusted               Nominal Inflation-adjusted 
            dollars            dollars               dollars            dollars 
1998      $7,513             $8,540                $7,261             $8,254 
1999       8,109              9,101                 7,197              8,047 
2000       8,641              9,516                 7,040              7,750 
2001       9,328             10,037                 7,160              7,697 
2002      11,083             11,684                 7,153              7,541 
2003      13,247             13,694                 7,468              7,720 
2004      14,479             14,617                 7,680              7,755 

Source: GAO analysis of data from HUDCAPS and PAS.

Table 10: Estimated Impact of Policy Decisions and Other Factors on the
Change in Outlays for the Voucher and Project-Based Programs in
Inflation-Adjusted Dollars from 1998 through 2004

                                        

             Dollars in millions                                
                                            Estimated change in 
                                                  outlays       
Factors                                             Vouchers Project-based 
Change in the number of assisted                      $3,053         -$369 
households/units                                             
Change in rental subsidy per householda                2,742          -302 
Change in other program costsb                           282           172 
Total                                                 $6,077         -$499 

Source: GAO analysis of HUD data.

aThis amount represents the difference between the (1) total outlays for
the period and (2) individual changes in outlays due to the other two
factors.

bThis amount comprises increases in the administrative fee for vouchers
and the cost of Performance-Based Contract Administrators for the
project-based program. Limitations in the data did not allow us to
identify other program costs.

Description of the Econometric Analysis of Rental Subsidy Costs per
Household for the Voucher ProgramAppendix V

This appendix provides an overview of the econometric analysis we used to
investigate trends in Section 8 rental subsidies per household (housing
assistance payments, or HAP) between 1999 and 2004 for the voucher
program. These subsidies, which make up the difference between households'
payments (usually 30 percent of adjusted income) and the actual unit rent,
are limited by the payment standards set by local public housing agencies.
PHAs set these payment standards based in part on fair market rents (FMR)
that the Department of Housing and Urban Development (HUD) establishes for
individual housing markets, generally at the 40th percentile (in some
cases 50th percentile) of the distribution of rents. Raising the payment
standard relative to the FMR can provide assisted households with a wider
choice of housing, but renting more expensive units raises both the cost
of the subsidies and thus of the Section 8 programs.

Because of the potential influence on program costs, we wanted to
investigate the role of HUD and PHA policies in setting payment standards.
Since 1998, PHAs have had more leeway than they did previously to increase
(or decrease) payment standards relative to the FMR. According to HUD,
this authority has been exercised too generously and is a major cause of
the recent increase in HAPs.

Model and Data

We developed a pooled cross-section time-series model explaining monthly
HAPs as depending on a variety of housing market, program, and household
characteristics. The results and descriptive statistics are based on a 10
percent sample of voucher (and certificate) household records obtained
from HUD's Public and Indian Housing Information Center files. These files
provide snapshots of the program as of the end of each calendar year from
1998 through 2004 and provide information on HAPs, gross rents, FMRs, and
payment standards as well as household income and other characteristics.
The information in a file on a particular assisted household is current as
of a point in time-for instance, the date of a program action, usually the
date of an annual recertification for program eligibility. HUD's Office of
Policy Development and Research worked with the underlying administrative
files to (1) correct various coding errors and inconsistencies, (2)
identify the census tract of each household based on tenants' addresses,
and (3) add information of analytical interest that was

not necessarily required for program administration.1 The date of
admission to the program and the date of the program action were used to
measure how long the household had been in the program, and other fields
were used to indicate any change in the household's rental unit and
whether the household left the program. We used information on the
household's census tract to identify neighborhood characteristics. We also
used the census tract information to develop an indicator of neighborhood
quality by determining whether the voucher household's census tract was a
HUD-designated qualifying census tract (QCT).2

We excluded observations with extreme or missing values for key variables,
and we excluded duplicate observations in the latest record and the record
from the previous year. We also excluded households that appeared to have
entered or left the program more than once.

We placed each household in one of four categories, based on demographic
and labor market characteristics: single female-headed households with
children (nonelderly, nondisabled), elderly (including elderly disabled),
nonelderly disabled, and all other.3 Because groups could face different
housing and labor market conditions and the variables in our model could
have different effects on the level of HAP in each group, we estimated the
same model separately for each of the four categories. For instance,
disabled households are typically smaller than other households but may
require housing with features not commonly available in the general rental
stock. Families with children may be larger than other families, and thus
require larger units, and may also experience changes in labor market
incomes.

The purpose of the model is to explain monthly HAPs using an estimating
equation that is based on a variety of household, housing, neighborhood,
and policy factors. HAPs range from close to zero to the thousands of
dollars, with variations in each cross section and over time. In the
model, HAPs are explained by the general level of market rents, tenant
incomes, a measure of neighborhood quality, time period, and a measure of
PHA payment standard policy. We also included in our model a series of
explanatory dummy variables for household size, duration in program,
termination and moves, and metropolitan areas. All dollar amounts (e.g.,
HAP, market rents, adjusted income) are expressed in 2004:Q4 terms using
the price index for Personal Consumption Expenditures from the Bureau of
Economic Analysis.

Because gross rents are important in defining the level of HAPs, we
control for the general level of market rents in order to examine the
effects of other variables. We use the FMR for this purpose because it
provides considerable variation within cross sections and across time.4
The level of income is also important in determining the level of HAPs,
and we used adjusted income as reported in the file. This choice is
potentially problematic, as the level of HAPs may influence income by
encouraging program participants to seek work or not. However, this
problem is somewhat mitigated by the fact that adjusted income is a
predetermined rather than an actual amount. Specifically, the adjusted
income reported in the file is the PHA's projection of a household's
income in the upcoming year based on income information from the previous
year, taking into account expected changes in hours, wages, and labor
force status. Finally, the file did not include information concerning
household characteristics, such as occupation, education, and experience,
that would help explain variations in assistance payments at the
individual household level.

The policy variable of interest relates to the way PHAs set payment
standards (relative to the FMR). We define a ratio variable to measure
this policy by calculating the average payment standard and average FMR by
year and bedroom size for each PHA and then calculate the ratio (1 = 100)
of the year-specific, PHA-specific payment standard to the FMR. Missing
payment standard information were set equal to the FMR for a value of
100.5 (To limit the effects of outliers, we excluded from the analysis
those households with payment standard ratios of less than 75 and more
than 120.) The baseline specification uses the year-specific, PHA-specific
payment standard to FMR ratio as a continuous variable (also truncated at
75 and 120).

Neighborhood quality is measured in two ways, both of them based on the
household's census tract. Our base specifications use HUD-designated QCTs,
which are in less desirable neighborhoods than other tracts. Thus rents
and HAPs should be lower in those neighborhoods, given that the market
rent variable distinguishes higher-rent markets from lower-rent markets.

Because the same households are in the data set for many years, up to as
many as six times, the error terms are not likely to be independent from
each other to the extent that unobserved characteristics may make the
error terms for each household correlated with each other. However, to the
extent that this presents a problem with the confidence intervals around a
coefficient estimate (rather than a point) estimate, we believe that this
is mitigated to a large extent by the large sample sizes used in the
estimation. Table 11 shows the mean values of the variables included in
our statistical model for the whole period from 1999 through 2004.

Table 11: Average Values of Voucher Household Groups from 1999 through
2004

                                        

                             Household group   
           Variable                     Single Nonelderly  Elderly All others 
                            female-headed with   disabled          
                                      children                     
Monthly HAP                         552.871    430.302  400.898    497.301 
Qualified census tract                0.270      0.248    0.219      0.268 
Local market rent                   857.948    700.366  690.228    822.026 
Adjusted income                    9980.614   9360.595 9690.530  10819.378 
Payment standard ratio               99.109     99.385   99.168     99.324 
Duration dummy variables                                        
1 year                                0.218      0.190    0.122      0.258 
2 years                               0.191      0.167    0.116      0.189 
3 years                               0.144      0.135    0.106      0.128 
4 years                               0.103      0.102    0.086      0.083 
5 years                               0.073      0.076    0.070      0.058 
6 years                               0.056      0.059    0.062      0.047 
7 years                               0.043      0.046    0.052      0.039 
8 years                               0.033      0.038    0.044      0.031 
9 years                               0.028      0.031    0.040      0.026 
10 years                              0.023      0.027    0.038      0.023 
11 years                              0.019      0.022    0.034      0.020 
12 or more years                      0.015      0.019    0.032      0.017 
Participation status                                            
dummy variables                                                 
Mover household                       0.059      0.058    0.033      0.045 
Household assistance                  0.097      0.076    0.075      0.115 
terminated                                                      
Household size dummy                                            
variables                                                       
1 person                                N/A      0.558    0.751      0.415 
2 persons                             0.299      0.209    0.187      0.068 
3 persons                             0.336      0.113    0.036      0.111 
4 persons                             0.218      0.065    0.015      0.149 
5 or more persons                     0.095      0.032    0.007      0.123 
Year and quarter dummy                                          
variables                                                       
1999:2                                0.031      0.029    0.034      0.031 
1999:3                                0.042      0.035    0.038      0.039 
1999:4                                0.040      0.033    0.035      0.037 
2000:1                                0.026      0.025    0.030      0.025 
2000:2                                0.034      0.033    0.040      0.033 
2000:3                                0.046      0.042    0.044      0.042 
2000:4                                0.057      0.048    0.047      0.053 
2001:1                                0.041      0.038    0.042      0.041 
2001:2                                0.041      0.038    0.042      0.040 
2001:3                                0.030      0.031    0.031      0.028 
2001:4                                0.038      0.037    0.033      0.036 
2002:1                                0.020      0.021    0.021      0.021 
2002:2                                0.037      0.042    0.043      0.039 
2002:3                                0.044      0.047    0.045      0.045 
2002:4                                0.061      0.058    0.052      0.060 
2003:1                                0.029      0.034    0.034      0.031 
2003:2                                0.044      0.051    0.051      0.049 
2003:3                                0.058      0.061    0.056      0.060 
2003:4                                0.069      0.065    0.059      0.071 
2004:1                                0.029      0.034    0.033      0.030 
2004:2                                0.043      0.051    0.051      0.046 
2004:3                                0.055      0.061    0.056      0.059 
2004:4                                0.066      0.067    0.059      0.068 

Source: GAO analysis of data from PIC.

Results

The results of our regressions are reported in table 12. Unless reported
separately in parentheses, all P-values are less than 0.0001.

Table 12: Regression Results

                                        

                              Coefficient     
                            Household group              
          Variable                     Single Nonelderly  Elderly All others 
                           female-headed with   disabled                     
                                     children                                
Intercept                         -240.552   -168.373 -115.147   -292.440 
Qualified census tract             -28.283    -18.869  -25.689    -34.603 
Local market rent                    0.649      0.691    0.633      0.660 
Adjusted household                  -0.022     -0.022   -0.021     -0.022 
income                                                                    
Payment standard ratio               3.866      3.398    3.418      4.237 
Duration dummy                                                            
variables                                                                 
1 year                              46.034     23.026   33.999     53.647 
2 years                             35.756     18.196   24.746     44.206 
3 years                             29.083     13.634   18.374     33.276 
4 years                             25.382      8.878   15.256     29.547 
5 years                             23.069      7.110   17.204     28.888 
6 years                             21.209      6.446   15.695     29.878 
7 years                             20.167      5.282   13.495     21.174 
8 years                             19.276      6.233   13.291     20.160 
9 years                             13.259      6.013   13.024     19.785 
10 years                             9.483      4.646   14.813     16.343 
11 years                            11.401      6.443   16.090     17.874 
12 or more years                     7.772      4.026   10.311     19.456 
Participation status                                                      
dummy variables                                                           
Mover household                     41.479     27.219   24.174     31.057 
Household assistance               -27.264    -22.529  -20.597    -26.525 
terminated                                                                
Household size dummy                                                      
variables                                                                 
1 person                               N/A   -100.261 -138.022    -95.574 
2 persons                          -71.769    -67.421 -105.279    -71.650 
3 persons                          -53.702    -47.023  -73.181    -58.171 
4 persons                          -37.119    -32.854  -55.639    -44.040 
5 or more persons                  -22.759    -19.995  -20.483    -27.020 
Year and quarter dummy                                                    
variables                                                                 
1999:2                              -0.257      1.863    2.465      2.120 
                                                                             
                                        (.89)      (.42)    (.40)      (.61) 
1999:3                               2.704      5.671    5.199      3.722 
                                                                             
                                        (.11)      (.01)    (.07)      (.35) 
1999:4                              -1.691     -3.055    0.294      4.068 
                                                                             
                                        (.33)      (.17)    (.92)      (.31) 
2000:1                              18.328     13.973   12.736     15.781 
2000:2                              25.116     20.466   16.514     25.087 
2000:3                              32.275     27.919   25.471     29.245 
2000:4                              13.081      9.142   10.418     14.786 
2001:1                              20.595     18.043   23.518     21.154 
2001:2                              29.780     22.617   26.271     32.555 
2001:3                              29.524     26.093   16.319     28.751 
2001:4                              15.548      9.251    2.708     19.327 
                                                                             
                                                            (.35)            
2002:1                               9.472      7.623    2.221     10.239 
                                                                             
                                                            (.49)      (.02) 
2002:2                              17.449     13.194    8.509     14.673 
2002:3                              16.950     11.318    5.394     16.654 
                                                                             
                                                            (.05)            
2002:4                              12.098      5.315   -5.070     12.949 
                                                                             
                                                   (.01)    (.06)            
2003:1                              36.467      8.982    2.196     28.221 
                                                                             
                                                            (.46)            
2003:2                              42.401     14.587    2.684     27.116 
                                                                             
                                                            (.33)            
2003:3                              46.316     17.019    8.263     32.605 
2003:4                              44.151     14.244    4.514     36.258 
                                                                             
                                                            (.09)            
2004:1                              57.448     21.379    7.499     43.155 
                                                                             
                                                            (.06)            
2004:2                              55.581     19.587    2.651     32.373 
                                                                             
                                                            (.34)            
2004:3                              53.379     16.720    9.975     42.051 
2004:4                              52.209     13.051    6.428     39.735 
                                                                             
                                                            (.02)            
Sample size                        375,753    187,216  131,693     93,536 
R2                                    0.83       0.81     0.71       0.81 

Source: GAO analysis of data from PIC.

In general, the results are consistent with our general expectations. For
example,

o HAPs increase with market rent levels and decrease with adjusted
incomes.

o Households in less desirable neighborhoods, as measured by the QCT
variable, are about $20 to $30 per month less ($240 to $360 annually),
depending on the group.

o Smaller households receive smaller HAPs, and those in the program longer
receive smaller HAPs.

o Those that ultimately leave the program receive smaller HAPs, in some
cases because incomes may have increased to the point that the households
are no longer eligible.

o Households that move to a new unit tend to receive higher HAPs.

o HAPs increase as the payment standard increases relative to the FMR.

The time period dummy variables used in our model suggest that, at least
for households that are neither elderly nor disabled, HAPs were
approximately $40 to $50 per month ($480 to $600 annually) higher in 2004
than in 1999, even after controlling for changes in market rent levels and
payment standards.

To present the results in terms of trends, we focused on those variables
for which the average values changed significantly over the time period.
Table 13 presents averages of selected variables-HAP, market rents,
adjusted income, and payment standard ratio-for the largest group (single
female-headed households with children).

Table 13: Averages for Selected Variables, 1998-2004

                                        

                            Year   
Variable                   1999     2000     2001     2002    2003    2004 
Monthly HAP              $487.0   $492.5   $509.7   $558.7  $612.9  $615.8 
Market rents             $803.4   $805.0   $824.4   $880.1  $907.1  $896.9 
Adjusted income          $9,619  $10,064  $10,136  $10,002  $9,970 $10,031 
Payment standard ratio     96.1     94.3     95.9    100.4   102.5   103.1 

Source: GAO analysis of data from PIC.

Trends in Rents and Household Payments for the Voucher and Project-Based
ProgramsAppendix VI

The rental subsidy per household of both Section 8 programs is the
difference between a household's payment and the lesser of either the
payment standard or the unit's gross rent. Trends in rents and household
payments, therefore, drive changes in the rental subsidy per household.
For the voucher program, average rents grew by 35 percent from 1998
through 2004 (fig. 12). The average annual increase in voucher rents was 5
percent during this period, ranging from a low of 3 percent in 2004 to 8
percent in 2002. Average project-based rents grew by 12 percent over this
period, an average annual rate of 2 percent. Rents in the voucher program
grew almost three times faster than those in the project-based program (35
percent versus 12 percent) over this period. A major reason for this
difference is that voucher rents are determined by the private market,
while project-based rents are adjusted annually using a HUD-determined
operating cost factor.

Figure 12: Average Annual Rents for the Voucher and Project-Based
Programs, 1998-2004

Annual increases in household payments did not keep pace with the
increases in voucher rents. Specifically, the average household payment by
voucher households rose by 24 percent over this period and grew at an
average annual rate of 4 percent (fig. 13). The disparity in the rates of
increase between rents and household payments accelerated the growth in
the average per household subsidy for vouchers. In contrast, the annual
rate of increase in the average project-based rent was similar to that of
household payments. As a result, growth in the average per household
subsidy kept pace with rents and household payments in the project-based
program.

Figure 13: Average Annual Household Payment for the Voucher and
Project-Based Programs, 1998-2004

Annual Increases in Voucher Rents and Fair Market Rents Were Similar

Although the average voucher rent grew dramatically from 1998 through
2004, our analysis found that this increase was consistent with the growth
in the average fair market rent. Fair market rents, which HUD sets for
each locality, reflect the cost of modest, standard-quality housing.1 We
created a fair market rent index, weighted by the proportion of voucher
households in each locality, and compared it to the average rent for
vouchers, which was similarly weighted, in order to assess the change in
the average rent for vouchers over time. From 1999 through 2004 (the only
years for which complete data on fair market rents and voucher holders
were available), the average rent in the voucher program grew by 27
percent, while the average fair market rent grew by 23 percent (fig. 14).
Starting in 2003, the average voucher rent increased at a faster rate than
the average fair market rent-5 percent versus 4 percent, respectively, in
2003, and 3 percent versus 1 percent, respectively, in 2004-thus narrowing
the gap between them.

Figure 14: Average Voucher Rents and Fair Market Rents, 1999-2004

Note: Data on the fair market rent associated with the individual
household were not available for 1998.

A major reason for the trend in the growth in the average voucher rent was
PHAs' authority to set their payment standard above the applicable fair
market rent. As previously noted, each PHA sets a local payment standard
up to 110 percent of the fair market rent for their area. The average
payment standard as a percentage of the fair market rent has steadily
increased, from about 96 percent in 1999 to 103 percent in 2004.
Accordingly, the average voucher rent as a percentage of the fair market
rent also increased, from about 94 percent in 1999 to 97 percent in 2004.

Comments from the Department of Housing and Urban DevelopmentAppendix VII

GAO Contact and Staff AcknowledgmentsAppendix VIII

David G. Wood, (202) 512-8678 or [email protected]

In addition to the contact named above, Steve Westley, Assistant Director;
Stephen Brown; Emily Chalmers; Mark Egger; Daniel Garcia-Diaz; John T.
McGrail; Marc W. Molino; Rose Schuville; and William Sparling made key
contributions to this report.

(250243)

www.gao.gov/cgi-bin/getrpt? GAO-06-405 .

To view the full product, including the scope

and methodology, click on the link above.

For more information, contact David G. Wood at (202) 512-8678 or
[email protected].

Highlights of GAO-06-405 , a report to the Chairman, Subcommittee on
Housing and Community Opportunity, Committee on Financial Services, House
of Representatives

April 2006

RENTALHOUSING ASSISTANCE

Policy Decisions and Market Factors Explain Changes in the Costs of the
Section 8 Programs

From 1998 through 2004, annual budget authority for Section 8 grew from
$9.4 billion to $19.3 billion (105 percent, or 82 percent after adjusting
for inflation), while outlays grew from $14.8 billion to $22.2 billion (50
percent, or 33 percent after inflation adjustment). The steep rise in
budget authority was partly due to the additional funding needed to cover
the cost of renewing long-term contracts. GAO estimates that voucher
outlays grew by 93 percent from 1998 through 2004 (71 percent after
inflation adjustment), accounting for almost all of the growth in total
Section 8 outlays. Estimated project-based outlays grew by 6 percent (and
actually declined after inflation adjustment) over this period.

GAO estimates that about 43 percent of the growth in voucher outlays from
1998 through 2004 stemmed from policy decisions that increased the number
(from 1.6 million to 2.1 million) and use of vouchers, while over half of
this growth was due to an increase in the average rental subsidy per
household. For the project-based program, a modest increase in the average
rental subsidy per household drove the growth in outlays but was partly
offset by a reduction of 62,000 in the number of units.

On the basis of statistical analysis of cost data, GAO estimates that
growth in the average annual rental subsidy per voucher household from
1999 through 2004 is primarily explained by changes in market rents (about
one-half of the growth), PHAs' decisions to increase the maximum
subsidized rents (about one-quarter), and lagging growth in assisted
household incomes (about 16 percent.) Household and neighborhood
characteristics, while important cost determinants, did not vary enough to
cause a substantial change in the average rental subsidy per household.

Annual Budget Authority and Outlays for Section 8, 1998-2004

Annual appropriations for the Department of Housing and Urban
Development's (HUD) Section 8 programs-a key federal tool for subsidizing
rents of low-income households-have increased sharply in recent years,
raising concerns about their cost. Section 8 pays the difference between a
unit's rent and the household's payment (generally 30 percent of adjusted
income). Section 8 includes a voucher program administered by public
housing agencies (PHA) that allows eligible households to use vouchers to
rent units in the private market and a project-based program administered
by property owners who receive subsidies to rent specific units to
eligible households. In both programs, contracts between HUD and the
administrators specify the duration and amount of the subsidy.

GAO assessed Section 8 trends from fiscal years 1998 through 2004 and
examined (1) annual budget authority and outlays for each program; (2)
factors that have affected outlays; and (3) the estimated impact of
factors, such as market rents, on the average rental subsidy per voucher
household.
*** End of document. ***