Content uploaded by Daniel H. Weinberg
Author content
All content in this area was uploaded by Daniel H. Weinberg on Apr 02, 2016
Content may be subject to copyright.
The Section
8
housing program is the largest
U.
S.
housing assistanceprogram.
This
article
presents the
first systematic analysis
of
the benefits participating households receive and
compares those benefits with federal government costs. Section
8
New Construction
projects provide acceptable housing at an affordable price
to
those low
-
income
households fortunate enough
to
be accepted by the developers. However, about 20%
of
the
total project rents goes
to
program
-
related costs rather than
to
tenant beneJits.
A
major
advantage
of
the Section
8
Existing Housing program is that tenant benefits are provided
without excessive rent increases.
But,
most units
in
that program do not meet Section
8
Acceptability Criteria.
A
shft in emphasis from the New Construction
to
the Existing
Housing program can save thegovernment significant amounts of money yet maintain the
number of households assisted.
DANIEL
H.
WEINBERG
U.S.
Department
of
Health and Human Services
I.
INTRODUCTION
In the early stages of the Reagan administration, tremendous
attention has been focused on how to reduce the size
of
the federal
budget. One target has been housing assistance programs, which serve
only a fraction of those in need while providing some assistance to those
not obviously poor, at a cost many perceive to be excessive. The political
debate about changes in the federal government's housing assistance
strategy can be informed by the results
of
a
recently completed national
evaluation
of
the housing benefits received by participants in the major
federal housing program
-
the Lower Income Housing Assistance
Program, more popularly known as the Section
8
housing program.
AUTHOR'S
NOTE:
The research reported here wasperformedunder contract H
-
2902
to
Abt Associotes
Inc.
from the
U.S.
Department of Housing and Urban Development. This
article makes use
of
the work of several people: Mireille Leger and Sally Merrill
of
Abt
Associates developed the hedonic index from the Section
8
Housing Measurement
EVALUATION REVIEW, Val.
6
No.
1,
February
1982
5
-
24
0
1982
Sage
Publications,
Inc.
5
6
EVALUATION REVIEW
I
FEBRUARY
1982
The Section 8 program was authorized by the Housing and
Community Development Act of 1974 and is operated by the US.
Department of Housing and Urban Development (HUD). It is designed
to provide housing subsidies to low
-
and moderate
-
income families. The
program uses three major methods of housing assistance: (1) allowing
eligible households to find housing units in the private market and then
subsidizing their rent (the Existing Housing program), (2) guaranteeing
private developers of new multifamily projects that HUD will provide
subsidies for eligible households (the New Construction program), and
(3) guaranteeing developers undertaking major renovations of existing
structures similar subsidies (the Substantial Rehabilitation program).
As of September
30,
1980, 661,838, 238,783, and 28,516 households
were served by the three programs, respectively, for
a
total of 929,137
households.
Providing acceptable housing at an affordable price to low
-
income
renters is a primary objective of the Section
8
program. The main tools
used to achieve this end in the New Construction (NC) program are
requiring the developer to have the newly constructed units meet certain
housing standards and making direct payments
to
the project manager
to cover the difference between the agreed
-
upon
gross
rent of the unit
and the households’ rent payments
(25%
of the recipient households’.
incomes). Similarly, in the Existing Housing
(EH)
program, units are
evaluated according to “Section 8 Acceptability Criteria” by Public
Housing Agency (PHA) inspectors, and landlords are required
to
have
deficiencies remedied before payments are made to the landlord.
Program regulations place an upper_limit on both program costs and
housing quality by limiting rents to be no more than “Fair Market
Rents.” Because of the small size of the Substantial Rehabilitation
program relative
to
the others, it was not afocus of the evaluation effort.
Prior to this research, only preliminary and limited evaluations of the
benefits to households from the program have been possible (see Drury
Surveys, Stephen Malpezzi and Larry Ozanne of the Urban Instituie developed hedonic
indices from Annual Housing Survey data, and George Hill and John Sneed of the Sorites
Group developed the normal rent equations (also from Annual Housing Survey data).
Thanks must also go
to
Richard Zwetchkenbaum for catching some of my computer
errors and for general research assistance. and
to
James Wallace, projeci director of the
overall analysis, Stephen Mayo, projeci reviewer, Howard Bloom, Sheldon Danziger, and
two anonymous referees
for
their comments andsuggesiions.
7%
article reflects ihe views
of ihe auihor and not those of the Departmenis of Housing and Urban Development or
Health and Human Services or of Abt Associates Inc.
Weinberg
/
SECTION
8
HOUSING BENEFITS
7
et al., 1978; Olsen and Rasmussen, 1979). Other portions of this research
project examined the demographic characteristics of participants and
the economic and racial/ ethnic integration implications of the program
(see Wallace et al., 1981, for complete results of the evaluation). The
research reported here focuses on the housing improvements of
households entering the
EH
and NC programs in mid
-
1979.
Two viewpoints should be kept in mind in reviewing the analysis: (1) a
participant focus
-
Did the Section
8
program help participating low
-
income households become better off than they were in their preprogram
dwellings?, and
(2)
a
bureaucratic focus
-
Has the Section 8 program
provided acceptable housing at affordable prices and reasonable costs?
Both market
-
oriented and regulation
-
based measures of housing qual
-
ity are used to evaluate both housing change and program performance.
Twenty Standard Metropolitan Statistical Areas (SMSAs) were
selected with probability proportional to the size of the Section 8
program in place. In the selected SMSAs, the NC panel was established
for projects in the process of renting up at least a substantial proportion
of units in mid
-
1979. In these projects, managers were asked to identify
applicants that had been selected to move into the project. Application
data were collected for these households and a preprogram interview
addressed primarily to housing conditions was administered. For
households completing that interview,
a
Housing Measurement Survey
(HMS) was conducted by
a
member of the evaluation staff; this form
obtained objective measures of a wide range of housing conditions. The
households in this respondent group that actually moved into the
project were then given another interview and an HMS was conducted
on their
or
another (identical) unit in the project.
For the EH panel, PHAs were asked to identify applicants that had
been recently provided a “Certificate of Family Participation,” permit
-
ting them
to
look for housing that could be subsidized. Application data
about these households were collected and
a
preprogram interview
administered whenever possible. As was true for the NC panel, an HMS
was conducted on the housing units of households completing this
interview. Certificate holders that later qualified for
a
program rent
subsidy, either in their preprogram unit or another, were administered a
recipient interview and an HMS was conducted on their dwelling unit.
Data on the NC program were collected in 13 SMSAs at 32 projects
and on the EH program in 15 SMSAs for 83 PHAs.1 The results
reported below are based solely on those interviewed households who
became recipients. The sample was weighted to adjust for differing
I
8 EVALUATION REVIEW
/
FEBRUARY
1982
,
sampling fractions in order to make it representative of the national
Section 8 population entering the programs in mid
-
1979. (Details on the
weighting scheme are presented in Wallace et al., 1981.) Since the
sample has been subject to both self
-
selection by the recipients via the
decisions to apply and meet the program requirements and selection by
the PHA or the developer, it would be inappropriate to extend the
results reported below without modification to
a
similar but open-
enrollment national housing program.
The plan of the article is as follows. The next section will deal with the
impact of program regulations on housing quality by examining both
the official measure of housing adequacy (the Section 8 Acceptability
Criteria) and rent burden (the proportion of income spent on rent).
Section
I11
will then examine two more market
-
oriented measures of
housing
-
housing expenditures and a hedonic measure of housing
quality (a measure of the market value of
a
unit). The final section will
draw some conclusions about the costs
of
program units relative to the
benefits households receive and present some concluding remarks.
II.
PROGRAM REGULATIONS AND THEIR
IMPACT
ON
HOUSING DEPRIVATION
The focus of this section is on measures of the housing change
experienced by recipients
-
a comparison of the housing provided by the
program with that of the households’ preprogram units.
..
~~
-
THE SECTION
8
ACCEPTABILITY CRITERIA
To provide a policy
-
oriented measure consistent across the NC and
EH
programs and across sites, the Section 8 regulations
(24
Code of
Federal Regulations 882.109) have been used as the main housing
quality standard.* The acceptability criteria cover sanitary facilities (a
working toilet, sink, and bath), food preparation and refuse disposal (a
working stove, refrigerator, and kitchen sink), space and security
(sufficient rooms and lockable doors and windows), thermal environ
-
ment (safe heating), illumination and electricity (adequate fixtures and
outlets,
no
hazards), structure and materials (satisfactory floors, walls,
ceilings, and steps), interior air quality (adequate ventilation), access
Weinberg
/
SECTION
8
HOUSING
BENEFITS
9
(direct access, fire exits), site and neighborhood (no serious adverse
environmental conditions), and rodent infestation (no rats).
While explicit in many instances, the acceptability criteria leave open
the interpretation of such terms as
“appropriate,” “proper,” “adequate,”
“approved,” and
so
forth. Appendix A, available from the author on
request, presents an operationalization of the criteria, using character
-
istics of the unit as measured on the HMSs, as developed in conjunction
with researchers at HUD. The presence of a prohibited condition was
sufficient to “fail” a unit; missing information was not. Consequently,
the measure is likely to err on the side of rating some units
as
acceptable
that may be unacceptable.
Table 1 presents the evaluation of the Section
8
acceptability of
recipient households’ preprogram and program units. Almost two-
thirds of all preprogram units failed the criteria. This failure rate is
significantly reduced by both Section
8
programs: from a preprogram
rate of 64% to only 11% in the NC program, and from
66%
to 55%in the
the EH program. Note that
a
portion, albeit small, of the newly
constructed units failed the criteria; note also that well over half of the
units being subsidized by the
EH
program in these sites (and therefore
rated acceptable by the PHAs) were nevertheless rated unacceptable by
our measure. Failures of this federally imposed regulation are due to
two factors
-
differences in interpretation among our research defini
-
tion, HUD intentions, and PHA implementation and the inaccurate
application of existing guidelines by PHA inspectors.
Stayers in this program showed basically no change in their housing
(of the stayers who indicated that some repairs had been made to their
units by the landlord, over three
-
quarters nevertheless lived in a
program unit which failed the acceptability criteria). In contrast, movers
showed a significant improvement, but were in preprogram housing that
failed the criteria at a much higher rate.
In evaluating this finding of program unit failure, it is important to
analyze specific reasons for failure. Investigation revealed
a
wide range
of reasons for failure of the preprogram units. Problems with illumina
-
tion and electricity (mainly the presence of electrical hazards) and
structure and materials (usually serious structural or surface problems
in multiple rooms) occurred most frequently. Other important problems
were with space and security (primarily nonlockable doors or windows),
thermal environment (unsafe heat sources), interior air quality (insuffi
-
cient bathroom ventilation), and site and neighborhood (various
10
EVALUATION REVIEW
/
FEBRUARY
1982
TABLE
1
Section
8
Acceptability (percentage unacceptable)
Household Preprogram Program Sample
Group
Unit
Unit
Changea
Size
New
Construction
64%
11%
-53%** 37
1
(2)
(2)
Existing Housing
66 55
-11**
392
(2)
(3)
Movers
79 54 -25** 149
Stayers
-
2
24 3
NOTE:
Standard errors in parentheses.
a.
Mean difference as
a
percentage
of
mean normal housing services.
**t-statistic significant
at
the
0.01
level.
problems). The least often occurring problems were improper access
and rodent infestation (rats). More than one
-
third of the preprogram
units failed more than one criterion. An even larger proportion failed
several subcomponents within a single criterion.
In the
NC
program, program unit failures were concentrated in two
components
-
the same two criteria failed most often by the preprogram
units (illumination/ electricity and structure/ materials). Most units that
failed did not have acceptable illumination and electricity, attributable
entirely to the presence
of
electrical hazards. Many
of
these may have
been
a
result of tenant installation of improper extension cords, but an
entire project in San Diego failed due to a minor electrical
-
hazard in
-
its
mechanical system. Other failures were attributable to loose steps
or
railings (improper installation), major trash accumulation (a vacant lot
next door), and
a
door
or
window not lockable. These failures may be
due either to faulty construction
or
to tenant behavior. Regardless of
who is responsible, though, most of these deficiencies seem relatively
inexpensive to repair, though there is currently
no
incentive for the
landlord
or
tenant to do
so.
Clear conclusions emerge from this analysis. While most NC
program units meet the Section
8
Acceptability Criteria (at least when
first constructed), it is surprising that a few do not.
For
those units, it
appears that in many cases modest and relatively inexpensive repairs
could be made that would overcome the failure to meet the measure
used. This is in contrast to the preprogram units that failed to meet the
Weinberg
/
SECTION
8
HOUSING BENEFITS
I1
criteria
-
those units often had multiple and serious problems much
more difficult to repair. Thus
NC
program units represent a consider
-
able improvement in housing conditions for their occupants, a result of
the large (and clearly significant) differences in the rates at which
preprogram and program units meet the acceptability criteria. The
magnitude of the change is
so
large that minor changes in the criteria
would not alter this conclusion.
On the other hand, it appears that the PHAs administering the
EH
program are not holding program units strictly to the requirements of
the applicable Section
8
regulations in deciding whether dwelling units
should be subsidized. More than half
of
the recipients are living in units
which do not meet the criteria. The discussion in Section
111,
however,
will show that the program rents are quite close to the estimated market
value of these units (the program rents of the
NC
program units are quite
a bit above their estimated market value). One implication therefore is
that more stringent application of the Section
8
housing requirements
might also require that
HUD
allow higher program rents in order to
allow households wanting to participate but in inadequate housing to
find acceptable housing.
RENT BURDEN
The fraction of income spent on rent is the second major criterion of
housing need in most housing policy discussions. Policy makers have
(as illustrated by the Brooke amendment and the Section
8
regulations),
yet this number has been questioned (see, for example, Lane, 1977) and
is,
in the final analysis, essentially arbitrary. Program regulations,
however, required expenditure of
25%
of “income after allowance” for
total family contribution, except for large very
-
low
-
income or very large
low
-
income households (where the percentage is 15%). (Recently,
Reagan administration officials have proposed that this fraction be
raised to
30%
or even higher.)
Whether the “correct” percentage is
25,
30,
or even up to
35,
it is clear
that recipient households were spending a good deal on rent in their
proprogram unit, Table
2
presents the proportionspent on rent by these
households (the ratio of net rent to Section
8
eligibility income) in their
preprogram and program units. (Net rent is the amount paid by the
household to the landlord plus any utility payments it makes. Section
8
argued that
25%
of income is an appropriate fraction to spend on rent
-
12
EVALUATION REVIEW
I
FEBRUARY
1982
TABLE
2
Rent
Burden
New
Construction Existing
Housing
Program Program
Preprogram Program Preprogram Program
Unit Unit
Unit
Unit
Median
a
rent burden 36% 25%
54%
26%
(3) (2)
(6)
(3)
Percentage
paying
more than 25 percent 67
’
50
92
56
more than 35 percent
45
10
78 26
Sample
Size
245 299
NOTE: Rent burden calculated as the fraction
of
Section
8
eligibility income spent
on
net
rent. Standard errors
of
the median in parentheses.
a. Unweighted median.
eligibility income is used instead of income after allowances because the
interviews did not collect the expense information required
to
compute
allowable deductions.) The Section
8
program reduced rent burdens
substantially, though more than
50%
of households were still paying
more than
25%
of their eligibility income for rent.3
III.
MARKET-ORIENTED MEASURES
OF
HOUSING
..
-
This section focuses
on
quantifying the housingchange in market
terms, in contrast to the previous section in which simpler policy-
oriented measures
of
housing quality were used. The discussion thus
shifts from concern with getting households out of substandard housing
and away from “excessive” rent burdens to focus on how much housing
change they make. The market
-
oriented measures used are housing
expenditures (rent) and a hedonic index of housing services (explained
further below).
HOUSING EXPENDITURES
Rent is usually considered a good measure of the “amount of
housing” provided by a unit. Increased expenditures
on
housing,
however, do
not
always imply increased housing quality, because rent is
Weinberg
/
SECTION
8
HOUSING BENEFITS
13
also determined by a variety of other factors not related to housing
quality, such as inflation and the substantial discount known to be
associated with long residency. In addition, the lease negotiations
between HUD and the project developer or between the
PHA
and the
landlord may lead
to
rents higher or lower than they would be in the
unassisted private market. Indeed, it has been argued that landlords see
the Fair Market Rent as a target rather than as a maximum rent.
It
is
therefore also important to look at a measure of housing quality (a
hedonic index of housing services) that estimates the market value of a
unit.
It is also worth noting that it is possible that programcosts will cause
total expenditures to exceed the market rental value of the units.
Reasons for such a disparity include special costs associated with the
development and operation of subsidized housing, possibly inefficient
construction practices, high factor costs (e.g., Davis
-
Bacon wages), or
inefficient operation. For a detailed discussion, see Mayo et al. (1980b:
Chapter
5)
or Sumka and Stegman (1978).
Gross rent is defined as the contract rent plus utilities, if utilities are
paid separately by the household, excluding the cost of furnishings.
However, households may get help with their rent from friends,
relatives, or, in particular, the government (through earmarked welfare
payments or other subsidized housing programs). In addition, some of
the (potential) Section
8
households that were interviewed were part of a
larger household unit. In either case,
gross
rent would not be a true
representation of the preprogram housing costs of that household.
Accordingly, a second measure of housing expenditures has been
examined
-
a net rent (rent actually paid) that indicates the Section
8
household’s share of the gross rent. The net rent measure reflects the
effective housing cost of the household before and during the program.
Table
3
presents the mean monthly preprogram and program levels
of gross and net rent, along with the change and percentage change in
rent. The Section 8 program clearly had an effect
on
the housing
of
recipients. If gross rent can be interpreted directly as a measure
of
housing quality, NC recipients more than doubled their housing
consumption and
EH
recipients increased their housing consumption
by 19%. (Among
EH
participants, movers experienced a Iarger change
in
gross
rent,
38%,
as compared to
8%
for stayers.)
The benefit to the household of increased housing consumption is
augmented by
a
significant reduction in net rent (the out
-
of
-
pocket costs
to the household) of
26%
for NC households and
46%
for
EH
households. By these measures, Section
8
program recipients appear to
-
14
EVALUATION REVIEW
I
FEBRUARY
1982
TABLE
3
Gross
and
Net
Rent
Mean Mean Mean
Preprogram Program Mean Percentage Sample
Rent Rent Change Change
a
Size
New Construction Program
Gross Rent for
housing unit
Net rent
of
household
Existing Housing Program
Gross
rent for
housing unit
Net rent of
household
$171
(4)
155
(4)
204
(3)
208
(4)
$364
(3)
115
(3)
24
3
(3)
112
(3)
$192** 112% 322
(6)
-40**
-
26 269
(5
1
38** 19 369
(3)
-96**
-
46 299
(4
1
NOTE: Standard error of mean
in
parentheses.
a. Mean change as a percentage
of
mean preprogram rent.
**t-statistic significant at the 0.01 level.
be getting more housing and paying less for it. (Bear in mind, though,.
that these gross rents may not reflect actual housing value.) EH
households seem to get more rent reduction than housing improvement
in contrast to
NC
recipients, who appear to be getting the reverse.
The changes in rent cannot beinterpreted directly as Section
8
program effects because no information was collectedThat would
indicate the changes in rent that would have occurred naturally in the
absence
of
the program. To obtain an estimate of the “normal” rent of
Section
8
households, cross
-
section demand functions were estimated
on metropolitan area samples of the unsubsidized but income
-
eligible
population, as collected by the Annual Housing Surveys (see
US.
Department of Housing and Urban Development). These equations
related gross rent to income and a wide variety
of
household character
-
istics including age, sex, race, household size, household composition,
and education.4
The predicted average gross rent using the coefficients estimated
from the Annual Housing Survey(AHS) data and applied to the Section
8
household sample was therefore used as an indicator of a household‘s
normal
gross
housing expenditures in the absence of subsidized housing
programs. Recent movers were used to predict rents in the
NC
program
Weinberg
/
SECTION
8
HOUSING BENEFITS
15
because recipients in that program must
all
be recent movers and movers
typically increase their rent more than do stayers. In the EH program,
recent movers were used to predict rents for movers, and all households
were used to predict rents for stayers (as corrected for length of tenure).
To the extent that movers were induced to move by the program, this
procedure overestimates their normal expenditures;
to
the extent that
stayers were induced to stay by the program, this procedure underesti
-
mates their normal expenditures. Evidence in Weinberg et al. (1981)
indicates that the incentives created by the EH program to move are not
likely to be large and therefore the procedure used here is not likely to
lead to significant misestimation. Normal rent has been corrected for
inflation between the year of the AHS and 1979 using the Consumer
Price Index rent subindex, as computed by metropolitan area.
Table 4 presents a comparison of normal and program gross rents. It
shows that the gross rent of recipients’ units was considerably more than
households normally would have been paying
(85%
more in the NC
program and 30% more in the
EH
program). Again, it is worth noting
that the gross rents may overestimate the “housing value” of the unit.
A household’s benefit is thus in two parts
-
the amount of housing
that it consumes beyond what it would normally consume and the
reduction in out
-
of
-
pocket rent payments (an increase in disposable
income). Assuming, for the moment, that gross rents are a good measure
of the amount of housing that recipients consume, the housing benefit
was an average of $168 per month per household in the NC program and
$56
per month per household in the EH program. The income benefit is
then the difference between normal and net rent
-
an average of $74 per
household per month in the NC program and $76 in the EH program.
The total household benefit
of
the NC program was then $242 per
household per month or $2904 per year,
31%
as income and
69%
as
housing.5 Similarly, the total household benefit of the EH program was
$132 per month or $1584 per year,
58%
as income and 42% as housing.
This naive approach to estimating program effects may overstate the
program benefits in several ways. First, the gross rent of an
NC
program
unit, as agreed to by the project developer and HUD, may overstate the
market value of the unit. Similarly, there is evidence that landlords in
the EH program increase rents by an average of $9 per month on units in
the program beyond what similar unsubsidized households would pay
for those units. In either case, the housing services actually provided by
the unit would be less than that indicated by the gross rent.
Second, it is well
-
known that in
-
kind benefits (such as housing) are
not valued as highly by the household as would an equivalent dollar cash
16
EVALUATION REVIEW
/
FEBRUARY 1982
1
TABLE
4
Normal
and
Actual Program Gross
Rent
Mean Mean Percentage Sample
Pronram Normal Program Difference Difference
a
Size
New Construction
$197
$365 $168** 85% 389
(5)
(3) (6)
Existing Housing 187 243 56** 30 457
(3)
(2)
(3)
NOTE: Standard error
of
mean
in
parentheses.
a. Mean difference as a percentage
of
mean normal rent.
**t-statistic significant at the
0.01
level.
payment (see, for example, Thurow, 1979). To the extent that house
-
holds discount housing benefits, the simple measure would overstate the
value of household benefits. Third, any household benefits should be
evaluated against the cost to the government of providing those benefits.
Mayo et al. (1980b) have shown that total costs can easily be twice as
much as the gross rental value of units provided in Public Housing and
other construction programs. These issues are better evaluated by
examining
a
hedonic index of housiog services.
-
HEDONIC INDICES
Housing standards are useful in assessing the adequacy of housing
obtained by Section
8
recipients, since adequacy must be defined in
terms of a set
of
standards which can be applied consistentIy- across
program units. The usefulness of these standards in measuring housing
quality is very limited, however. Because the standards omit most
neighborhood characteristics and many dwelling unit attributes, they
are quite inadequate as descriptors of overall quality.
In contrast to standardness measures which focus specifically on the
adequacy or deficiency of units, a more comprehensive approach to
housing quality measurement is provided by hedonic indices. (For
a
detailed description of the hedonic approach, see Griliches, 1971, or
Merrill, 1980.) The hedonic approach defines the dwelling unit and
neighborhood in terms of numerous attributes. Overall quality is
summarized by
a
weighted sum of these attributes. Since the weights
cannot be observed directly, they are estimated by regressing rent on the
various housing attributes, controlling for conditions
of
tenure (such as
length of residency or relationship to the landlord). The estimated
Weinberg
/
SECTION
8
HOUSING
BENEFITS
17
coefficients are then used to aggregate housing attributes into an overall
index of housing services. The index is particularly useful for assessing
the change in quality over time, for example, from preprogram to
program status. This
is
because, unlike rent, the index is not subject to
variations from nonhousing factors, such as inflation.
Two hedonic indices of housing services were available for this
analysis. The indices differ in fundamental ways, the most important of
which
is
probably the use of objective versus subjective data. The first
index was based on data collected on preprogram units for the Section
8
sample by housing evaluators using the Housing Measurement Surveys.
The other was based on recipient data from household interviews. The
questions concerning the dwelling unit and neighborhood were drawn
from the AHS (recall that the AHS is not
a
housing evaluation but
rather an interview with the households residing in a sample of dwelling
units in an SMSA).
Another crucial difference concerns the scope of data collected. The
AHS asked a limited number of questions concerning the dwelling unit
and neighborhood and tended to focus on deficiencies rather than on
quality and amenities. In contrast, the HMS collected quality data for
each room in the unit, common areas in multifamily units, mechanical
facilities, the building exterior and grounds, the blockface, and the four
surrounding parcels. Finally, AHS data were collected in either
1974,
1975,
or
1976;
thus the AHS
-
based index had to be inflated to
1979
dollars (the year of the HMS data). Utilities not included in rent were
added to make the resultant measures more comparable to program
gross rents.
Because the evaluator
-
based HMS index is more objective and
complete in coverage of elements contributing to housing value, it was
likely to provide a more accurate and coxisistent estimate of housing
services than the recipient response
-
based AHS index. In fact, although
the results were somewhat different for the two indices in some cases, the
similarities are marked. The major patterns that emerged with regard to
program costs and benefits in the Section
8
program and the differences
between the
NC
and
EH
programs were basically similar for both
indices.
The value of housing services as measured
by
the evaluator (HMS)
and recipient (AHS) indices and the change in these services for
participating households are shown in Table
5.
Both indices showed a
significant increase for
NC
recipients
(18%
using the
HMS
index). On
the other hand, the HMS index indicated a small but significant increase
in housing quality for the
EH
sample whereas the AHS index suggested
18
EVALUATION REVIEW
I
FEBRUARY 1982
TABLE
5
Hedonic Indices
of
Housing
Services
Mean
Mean Housing Services
Mean
percentage
Sample
Preprogram Program Change Change
a
Size
New
Constpction Program
HMS Index $246 $289
$44**
18% 214
AHS
Index
211
3
04
87**
41
21
1
(4)
(3)
(5
1
(4 (4)
(5
1
Existing Housing Program
HMS Index 25 6 26
1
5.1*
2 265
(3) (3)
(2.5)
(4)
(3) (2.8
AHS Index
24
3
238
-
5.5
*
-
2
3
25
NOTE: HMS Index based on Housing Measurement Surveys.
AHS
Index based
on
Annual Housing Survey interview data. Standard errors in parentheses.
a. Mean change
as
a percentage
of
mean preprogram housing services.
*t
-
statistic significant at the
0.05
level.
**t-statistic significant at the
0.01
level.
that housing services fell slightly. These results are more suggestive,
however, if
EH
movers and stayers are considered separately. Movers
obtained significant increases in housing quality according to both
indices
(8%
using the
HMS
index). Households that stayed, on the other
hand, experienced little or no increase in quality, though there was
a
slightly larger increase for households with repairs done to their units
than those without repairs. This increase for movers reflected their
preprogram housing conditions
-
their preprogram housing was more
frequently inadequate, had more deficiencies, and was more frequently
crowded than that of stayers.
The
NC
program provides an interesting reference for the housing
change of
EH
movers. These sample groups began with nearly the same
level of housing services
($245
and
$249,
respectively). On average,
EH
movers increased their housing quality by
$19
to
$268,
whereas housing
quality for
NC
households (all movers) rose by
$44
to
$289.
To obtain an estimate of normal housing services, the amount of
housing services that recipient households would be consuming in the.
absence of the program, cross
-
section demand functions were again
estimated on the AHS metropolitan area samples for eligible house
-
holds. The equations related AHS hedonic index values to the same
-
Weinberg
/
SECTION
8
HOUSING BENEFITS
19
TABLE
6
Normal
and
Actual
Housing Services
(AHS
Index)
~
Mean Housing
Mean Percentage Sample
Services
Proaram Normal Program Dijyerence
Difference
Size
New Construction
$193
$305 $112**
58%
218
(2)
(3)
(4)
Existing Housing 180
236
56**
31
405
(2)
(3)
(3)
NOTE: Standard errors in parentheses.
a. Mean difference as
a
percentage of mean normal housing services.
**t-statistic significant at the
0.01
level.
independent variables used to estimate normal rent. Predicted normal
housing services were obtained for the Section
8
sample using these
estimated coefficients and the characteristics of the sample households.
Normal housing services were then compared with the
AHS
index of
program housing services from the Section
8
interviews to get an
estimate of program effect on housing services.
As
is shown in Table
6,
the estimated increase in housing services above normal was large and
significant
(58%
in the NC program and 31% in the
EH
program).
Section
8
households obtained
a
great deal more housing services than
similar, unsubsidized households would normally have obtained.
IV.
SECTION
8
COSTS AND
HOUSEHOLD
BENEFITS
This section brings together many of the measures of benefits and
costs that have been discussed in previous sections and examines them
from the perspective
of
both
HUD
and the tenant. Figure
1
illustrates
the following costs and benefits:
(1) the average cost to the tenant (program net rent)
(2)
the total tenant benefit (the value
of
housing services less the cost to the tenant)
consisting
of
an income benefit (normal expenditure less net rent) and a housing
benefit (housing services less normal expenditure)
(3)
the total cost to
HUD
(gross rent less cost to the tenant) consisting
of
program-
related costs
(gross
rent less the value
of
housing services) and the household
sub
-
sidy (the value
of
housing services less the tenant contribution).
Omitted from the figure is the induced increase in gross rents for
households in the
EH
program (an average of
$9
per unit).
i
Ah
II
9
lE
n
(3
Z
H
f/
2
0
I
c3
z
H
I
-
f/
-
H
X
w
z
0
H
I
-
u
3
cy.
I
-
f/
z
13
20
Weinberg
/
SECTION
8
HOUSING
BENEFITS
21
As can be seen in the figure, the average cost to the tenant was $1 12 in
the NC program and $1 10 in the EH program. The total tenant benefit is
defined as the value of housing services received less tenant costs,
$179
per household per month in the New Construction program and $155
per household per month in the Existing Housing program. The total
tenant benefit may be viewed as having two components: the income
benefit and the housing benefit. The income benefit represents the
difference between what such households would normally spend for
housing (normal rent) and what they contribute as recipients (net rent).
The housing benefit, on the other hand, is measured by the difference
between the value
of
the housing services tenants actually received and
what they would normally have obtained if they were not in the program
(normal housing services, equated to normal expenditures in the
absence of any tenure discount).
Finally, the total cost to HUD is the difference between gross rent and
tenant contribution. HUD purchased housing services valued at an
average of $291 in the NC projects in the sample for $362; 20% of the
total cost ($71) represents an excess cost. This finding confirms the
analysis of Mayo et al. (1980b) that the costs of construction
-
related
housing programs exceed the benefits, though the ratio of costs to value
in the Section
8
program does not seem to be as high as that found for
Public Housing or Section 236 housing. However, the gross rent figure
reported for Section
8
units is likely to be an
underestimate
of total costs
to the federal government, because project gross rents do not include
sueh costs as program administration at HUD area or central offices,
Government National Mortgage Association (GNMA) tandem plan,
foregone tax revenue attributable to tax
-
free bond financing or
accelerated depreciation allowances, potential losses attributable to
foreclosures, or foregone local property taxes.6 The cost to HUD of an
average EH unit is the difference between program
gross
rent and the
tenant contribution, or $130. Only about 4% of gross rent (the $9
induced increase in gross rent) represents a cost to HUD beyond the rent
that would be charged for the unit in the absence of the program. This
compares favorably with the 20% excess cost found for the NC program.
The EH tenant benefit also compares favorably to the benefit
received under theNC program
-
$155 for the EH program and $179 for
the NC program
-
particularly since the former was obtained at
substantially less cost to HUD. Program gross rents averaged $240 in
the EH sample and $362 in the NC sample; thus the ratio of tenant
benefits to that estimate of gross cost was .65 for Existing Housing and
.49 for New Construction.
(To
the extent that households value their
housing benefit at less than its full cost, the ratio would of course be
lower.)
22
EVALUATION REVIEW
/
FEBRUARY
1982
In sum, it seems clear that Section
8
New Construction projects are
successful in providing acceptable housing at an affordable price to
those low
-
income households fortunate enough to be accepted by the
developers. This is confirmed both by
an
analysis of several measures of
hodsing standards and by examination of changes in more market-
oriented measures of housing such as rent burden, housing expendi
-
tures, and hedonic indices of housing services. Further, households
receive
a
large benefit
-
both in
-
kind as housing and as increases in
disposable income. Tempering these findings, however, is the indication
that
a
good deal of the total project rents goes to program
-
related costs
(about
20%).
Furthermore, this is likely to be an underestimate of the
total additional cost.
The Section
8
Existing Housing program also provides affordable
housing to low
-
income recipients and significantly increases the housing
services received by households that move. One possible drawback is
that
a
majority of program units do not meet the Section
8
Acceptability
Criteria. Although the inference is that these units are not acceptable
from a policy perspective, the reasons for failure are not always
indicative of serious deficiencies; reexamination of the criteria is already
under way at HUD. In any event, excess preprogram rent burdens are
greatly reduced and total tenant benefits are substantial.
A major advantage of the
EH
program is that tenant benefitrare
provided at reasonable total cost (as measured by program gross rent)
and without an excessive increase in those rents. Program gross rents are
only about
4%
above what would normally be charged for the units; in
NC the apparent markup is
20%.
It is beyond the scope ofthis evalu-
ation to consider whether this markup (plus HUD administrative and
other uncounted expenses) can be justified as a public cost, for example,
on the basis of additions to the housing stock or through externalities in
neighborhood development or revitalization. Yet it is worth noting that
a shift in emphasis away from the
NC
program (and by implication, the
Substantial Rehabilitation program as well) toward the EH program
seems called
for.
Using the estimates here as
a
lower bound estimate of
program
-
related costs, for
each
new unit shifted from the
NC
program
to the EH program, the government can cut excess costs (in 1979
dollars) by over
$60
per month. A shift ofjust
683
units (less than.
1%
of
the current total) from the NC program to the EH program would save
the government about $1 million per year (and reduce excess costs by at
least
$.
5
million per year). Alternatively, an additional
347
households
could be given housing assistance by the
EH
program without changing
total costs.
~
Weinberg
/
SECTION 8 HOUSING BENEFITS 23
NOTES
I.
The SMSAs selected with both programs active in mid
-
1979 were Atlanta,
Baltimore, Chicago, Cleveland, Houston, Milwaukee, New
York,
Philadelphia, Provi
-
dence, Rochester, St. Louis, San Diego, and Seattle. In addition, the EH sample included
Los
Angeles and Raleigh.
2. Other measures of
substandardness
-
including
a housing deficiency measure
developed by the Congressional Budget Office (1978) and a housing adequacy measure
developed by Budding (1980)
-
indicated basically similar results to the acceptability
criteria.
3.
These calculations used interview data to compute rent, utilities, and income.
Consequently, program net rent burden can differ from 25% for three reasons: (1) actual
utility payments that differ from utility allowances, (2) use of interview
-
reported eligibility
income instead of income after allowances computed by the PHA
or
the project, and
(3)
errors in reporting rent
or
income
or
errors by the agency
or
project staff in computing the
family contribution. (Utility allowances are computed on a geographic basis and are
intended to reduce the family payment for rent sufficiently that the household can pay for
needed utilities not already included in rent and still pay only 25% of income for rent plus
utilities.) Another possibility would be payments of only 15% of gross income for rent by
certain very large low
-
income and large very
-
low
-
income families, not present in this
sample.
4. The demand functions describing the normal housing consumption
of
low
-
income
households were estimated from the sample of income
-
eligible households separately in
each of the I5 SMSAs in the samples. Following Friedman and Weinberg (1981), a log-
linear functional form with complete demographic
-
income interaction was chosen. The
lack of price data may have led to biased parameter estimates (see Polinsky, 1977) but this
problem would not have affected the equations’ predictive power. There might, however,
be
a bias in predicted rent if the
unrneasuredcharacteristics
of
Section 8 applicants differed
systematically (due, say, to self
-
selection) from those of the income
-
eligible population as
a whole.
5. This is in contrast to Mayo et al.’s (1980a) finding that, for Public Housing, from
roughly two
-
thirds to three
-
quarters of the total economic benefits appear
to
be in the
form
of
increased income rather than increased housing. Mayo’s calculation is based,
however, on a sample of Public Housing units built between 1952 and 1974, some of which
represent small estimated housing changes for the recipients (the older units) and some
relatively large estimated housing changes (more recently built units).
6. The omitted costs for the
NC
program are likely to be in the range of 20% to 30%
of
direct costs, that is, between $72 and $109: “The indirect costs associated with Section 236
such
as
the GNMA tandem plan, federal tax losses from accelerated depreciation, and
so
on.
.
.
for units in the sample examined [the Demand Experiment Program Comparisons
sample], average about
20
percent
of
the direct costs
.
. .
in Pittsburgh, and
30
percent of
those direct costs in Phoenix,” (Mayo et al., 1980b: 142).
REFERENCES
BUDDING, D. (1980) Housing Deprivation Among Enrollees in the Housing Allowance
Demand Experiment. Cambridge,
MA.
Abt Associates Inc.
24 EVALUATION REVIEW
/
FEBRUARY 1982
Congressional Budget Office (1978) “Measures of housing need: findings from the Annual
Housing Survey.” U.S.
House
of Representatives, Committee
on
Banking and Urban
Affairs, Subcommittee on Housing and Community Development. Task Force on
Assisted Housing. Washington, DC: Government Printing Office.
DRURY, M.,
0.
LEE, M. SPRINGER, and L. YAP. (1978) Nationwide Evaluation of
the Existing Housing Program. Washington, DC: Department of Housing and Urban
Development.
FRIEDMAN,
J.
and D. H. WEINBERG (1981) “The demand for rental housing: evi
-
dence from the Housing Allowance Demand Experiment.”
J.
of Urban Economics
9:
GRILICHES, Z. (1971) “Hedonic price indices revisited,” pp. 3
-
15 in Price Indexes and
Quality Change. Cambridge, MA: Harvard Univ. Press.
LANE, T.
S.
(1977) Origin and Uses
of
the Conventional Rules of Thumb. Cambridge,
MA: Abt Associates Inc.
MAYO,
S.
K.,
S.
MANSFIELD, W. D. WARNER, and R. ZWETCHKENBAUM
(1980a)
Housing Allowances and Other Rental Assistance Programs
-
A Comparison Based
on the Housing Allowance Demand Experiment, Part
1:
Participation, Housing Con
-
sumption, Location, and Satisfaction. Cambridge, MA: Abt Associates Inc.
__-
(1980b) Housing Allowances and Other Rental Assistance Programs
-
A Compar-
ison Based on the Housing Allowance Demand Experiment, Part 2: Costs and Effi
-
ciency. Cambridge, MA: Abt Associates Inc.
MERRILL,
S.
(1980) Hedonic Indices as a Measure of Housing Quality. Cambridge,
MA: Abt Associates Inc.
OLSEN, E.
0.
and D. W. RASMUSSEN. (1979) “Section 8 existing: a program evalua
-
tion.” Occasional Papers in Housing and Community Affairs 6 (December): 1
-
32.
POLINSKY,
A.
M. (1977) “The demand for housing: a study in specification and group
-
ing.” Econometrica 45: 447-461.
SUMKA, H.
J.
and M. A. STEGMAN (1978) “An economic analysis of public housing in
small cities.”
J.
of Regional Sci.: 395
-
410.
THUROW, L. C. (1974) “Cash versus in
-
kind transfers.” Amer. Economic Rev.
64:
190-
195.
US.
Department
of
Housing and Urban Development (various years) Annual Housing
Survey. Washington, DC: Government Printing Office.
WALLACE,
J.
E.,
S.
P.
BLOOM, W. L. HOLSHOUSER,
S.
MANSFIELD, and D. H.
WEINBERG (1981) Participation and Benefits in the Urban Section 8 Program: New
Construction and Existing Housing. Cambridge, MA: Abt Associates Inc.
WEINBERG, D. H.,
J.
FRIEDMAN, and S. K. MAYO (1981) “Intraurban household
mobility: the role of transactions costs, market imperfections, and household dis
-
equilibrium.”
J.
of
Urban Economics
9:
332-348.
31 1
-
331.
-
Daniel H. Weinberg is an economisr with the Office
of
Income Security Policy
in
the
U.S.
Department
of
Health and Human Services. He received his Ph. D. from Yale University
and has worked as a senior economist
for
Abt Associates Inc., where he participated in
analysis
of
the Housing Allowance Demand Experiment. His current interests include
investigation
of
the impacts
of
multiple transfer program participation
on
income distri
-
bution.
