Single-Family Housing: Cost, Benefit, and Compliance Issues Raise
Questions about HUD's Discount Sales Program (30-JAN-04,	 
GAO-04-208).							 
                                                                 
In 2001, the Department of Housing and Urban Development's (HUD) 
Inspector General reported on serious problems in HUD's Discount 
Sales Program, under which nonprofit organizations purchase	 
HUD-owned properties at a discount, rehabilitate them, and resell
them to low- and moderate-income homebuyers. The objectives of	 
the program are to expand affordable housing opportunities, help 
revitalize neighborhoods, and reduce HUD's property inventory in 
a timely, efficient, and cost-effective manner. Although the	 
Inspector General recommended that the agency suspend the program
and evaluate its viability, HUD did neither. GAO was asked to	 
assess (1) the costs of the program to HUD, (2) the benefits of  
the program to homebuyers, and (3) HUD's efforts to monitor	 
participating nonprofits and enforce program requirements.	 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-04-208 					        
    ACCNO:   A09166						        
  TITLE:     Single-Family Housing: Cost, Benefit, and Compliance     
Issues Raise Questions about HUD's Discount Sales Program	 
     DATE:   01/30/2004 
  SUBJECT:   Administrative costs				 
	     Cost analysis					 
	     Excess profits					 
	     Housing programs					 
	     Low income housing 				 
	     Monitoring 					 
	     Nonprofit organizations				 
	     Program abuses					 
	     Program evaluation 				 
	     HUD Discount Sales Program 			 

******************************************************************
** 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-04-208

United States General Accounting Office

GAO	Report to the Chairman, Subcommittee on Housing and Transportation,

          Committee on Banking, Housing and Urban Affairs, U.S. Senate

January 2004

SINGLE-FAMILY HOUSING

Cost, Benefit, and Compliance Issues Raise Questions about HUD's Discount Sales
                                    Program

                                       a

GAO-04-208

Highlights of GAO-04-208, a report to the Chairman, Subcommittee on
Housing and Transportation, Committee on Banking, Housing and Urban
Affairs, U.S. Senate

In 2001, the Department of Housing and Urban Development's (HUD) Inspector
General reported on serious problems in HUD's Discount Sales Program,
under which nonprofit organizations purchase HUD-owned properties at a
discount, rehabilitate them, and resell them to low- and moderateincome
homebuyers. The objectives of the program are to expand affordable housing
opportunities, help revitalize neighborhoods, and reduce HUD's property
inventory in a timely, efficient, and cost-effective manner. Although the
Inspector General recommended that the agency suspend the program and
evaluate its viability, HUD did neither. GAO was asked to assess (1) the
costs of the program to HUD, (2) the benefits of the program to
homebuyers, and (3) HUD's efforts to monitor participating nonprofits and
enforce program requirements.

GAO recommends that HUD (1) evaluate options to improve the program's
benefit to homebuyers, the agency's monitoring of nonprofits, and
enforcement of excess profits requirements; (2) assess the extent to which
the program is meeting its objectives; and (3) terminate the program if
its current cost plus the resources needed to improve it exceed the
program's benefits. HUD agreed with GAO's recommendations to evaluate the
program but said that the report overstated the program's costs and
understated its benefits.

www.gao.gov/cgi-bin/getrpt?GAO-04-208.

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].

January 2004

SINGLE-FAMILY HOUSING

Cost, Benefit, and Compliance Issues Raise Questions about HUD's Discount Sales
Program

GAO found that the Discount Sales Program poses significant costs to HUD,
is of questionable benefit to homebuyers, and has serious monitoring and
compliance problems. GAO estimates that the program cost HUD between $18.8
and $23.9 million in calendar year 2002. Between $15.1 and $20.2 million
was a reduction in net revenue resulting from HUD's selling approximately
1,200 properties through the program instead of through its regular sales
process. Personnel expenses for administering the program accounted for
the remaining $3.7 million. GAO's analysis of 238 properties sold under
the program in 2002 suggests that most of the homebuyers did not benefit
financially. Assuming that nonprofits and homebuyers would incur the same
rehabilitation costs, GAO estimates that 76 percent of the homebuyers
would have spent less purchasing the properties through HUD's regular
process and paying for the rehabilitation work themselves. And while the
program can help homebuyers access a range of homeownership services,
these services are also available from other sources. GAO did not evaluate
the extent to which the program generated other benefits, such as
neighborhood revitalization.

While uncovering numerous program violations, HUD's monitoring efforts
have faced challenges. For example, HUD monitors nonprofits through desk
reviews of the annual reports it requires nonprofits to submit each
February. However, as of July 2003, HUD's four homeownership centers,
which administer the program, had not received reports for more than half
of the properties the agency estimates were purchased and resold under the
program in 2002. Even with this problem, the desk reviews found that 28 of
the 44 nonprofits that submitted reports violated resale limits, earning
an estimated total of $704,720 in excess profits (see figure). HUD
requires that nonprofits use their excess profits to pay down the
mortgages of the homebuyers they overcharged, but the agency's ability to
enforce this requirement is extremely limited. As of July 2003, nonprofits
had made only $62,000 in payments on mortgages.

Resale Price Violations on Discounted Properties Sold in Calendar Year
2002

Denotes number of nonprofits or properties with violations out of total.

Source: HUD.

Contents

  Letter

Results in Brief
Background
In 2002, the Discount Sales Program Cost HUD at Least

$18.8 Million

The Discount Sales Program Is Not Likely to Benefit Most Homebuyers
Financially but Can Help Them Access Homeownership Services and Assistance

Limited Monitoring Efforts Have Uncovered Numerous Violations,

but Effective Enforcement Has Been Difficult Conclusions Recommendations
for Executive Action Agency Comments and Our Evaluation

1 2 4

8

14

19 26 27 28

Appendixes

Appendix I:

Appendix II:

Appendix III:

Appendix IV:

Objectives, Scope, and Methodology

Statistical Models Used to Estimate Program Costs and Benefits

Comments From the Department of Housing and Urban Development

GAO Contacts and Staff Acknowledgments

GAO Contacts Acknowledgments 31

34

44

46 46 46

Tables	Table 1: Table 2:

Table 3:

Table 4:

Table 5:

Table 6:

HUD's Calendar Year 2002 Single-Family Property Sales 5
Estimated Reduction in Net Revenue Due to the Discount
Sales Program in Calendar Year 2002, by Discount Level 11
Estimated Reduction in Net Revenue Due to the Discount
Sales Program in Calendar Year 2002, by Homeownership
Center 12
HUD's Estimate of Personnel Costs for the Discount Sales
Program in Calendar Year 2002, by Homeownership
Center 13
Estimated Number of Homebuyers Who Did and Did Not
Benefit Financially by Purchasing Homes through HUD's
Discount Sales Program 16
Percentage of Estimated Excess Profits Used to Pay Down
Homebuyer Mortgages, as of July 2003 26

                                    Contents

Table 7: Variable Names, Descriptions, and Mean Values 39 Table 8: Coefficients
                            from Estimated Models 40

Figure Figure 1:	Resale Price Violations Disclosed by Desk Reviews of
Nonprofits' Annual Reports on Calendar Year 2002 Program Activity

Abbreviations

FHA Federal Housing Administration
GAO General Accounting Office
HOC homeownership center
HUD Department of Housing and Urban Development
SAMS Single-Family Acquired Asset Management System

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

A

United States General Accounting Office Washington, D.C. 20548

January 30, 2004

The Honorable Wayne Allard
Chairman
Subcommittee on Housing and Transportation
Committee on Banking, Housing and Urban Affairs
United States Senate

Dear Mr. Chairman:

Each year, the Department of Housing and Urban Development (HUD)
acquires tens of thousands of single-family properties through
foreclosures
when homeowners default on mortgages insured by the Federal Housing
Administration (FHA). HUD sells these properties in as-is condition
through its regular sales process and a number of smaller, specialized
programs, including the Discount Sales Program. Under the Discount Sales
Program, HUD sells properties at 10, 15, or 30 percent discounts to
nonprofit organizations1 and government entities2 (nonprofits) that then
rehabilitate (rehab) the homes as necessary and resell them to low- and
moderate-income homebuyers. However, a November 2001 report by
HUD's Inspector General concluded that low- and moderate-income
homebuyers did not benefit significantly from the program; that many
participating nonprofits were actually profit-motivated entities; and that
HUD lacked effective approval, monitoring, and enforcement procedures.3
In light of these problems, the Inspector General recommended, among
other things, that the agency suspend the program and evaluate the
program's viability. Although HUD acted on many of the report's
recommendations, it neither suspended nor evaluated the program.

1HUD requires these organizations to have 501(c)(3) status. Section
501(c)(3) of the Internal Revenue Code covers charitable organizations
that are eligible to receive tax-deductible contributions. Such
organizations must not be organized or operated for the benefit of private
interests, and no part of their net earnings may benefit any private
shareholder or individual.

2Government entities include states, counties, cities, or other units of
government such as public housing authorities. For purposes of this
report, the term "nonprofit" includes nonprofit organizations and
government entities.

3U.S. Department of Housing and Urban Development, Office of Inspector
General, Nonprofit Participation in HUD Single-Family Programs,
2002-SF-0001 (Washington, D.C.: November 5, 2001). The audit covered the
discounted properties HUD sold between January 1, 1998, and April 30,
2001.

As agreed with your office, this report assesses (1) the cost of the
program to HUD, (2) the benefits of the program to homebuyers, and (3)
HUD's efforts to monitor participating nonprofits and enforce program
requirements. To address these objectives, we reviewed the program
activities of HUD's Office of Housing and four homeownership centers (HOC)
in Atlanta, Georgia; Denver, Colorado, Philadelphia, Pennsylvania; and
Santa Ana, California. As agreed with your office, our work focused on the
properties HUD sold through the program in calendar year 2002 and the
monitoring and enforcement activities the HOCs performed in connection
with those properties. To estimate the program's cost to HUD and its
benefits to homebuyers, we performed a statistical analysis using data
from HUD's Single-Family Acquired Asset Management System (SAMS)4 and the
U.S. Census Bureau. We did not evaluate the extent to which the program
generated other benefits, such as neighborhood revitalization. Appendixes
II and III provide detailed information on our objectives, scope, and
methodology.

Results in Brief	We estimate that the Discount Sales Program cost HUD
between $18.8 and $23.9 million in calendar year 2002. Most of this cost,
between $15.1 and $20.2 million, was a reduction in net revenue resulting
from HUD's selling approximately 1,200 properties through the program
rather than through its regular sales process, in which properties' prices
are not discounted. The reduction in net revenue varied according to the
discount level, with properties sold at a 30 percent discount accounting
for the largest share of the total reductions. Personnel expenses for
administering the program accounted for the remainder of HUD's cost. HUD
estimates that its headquarters and HOCs allotted about 45 staff years5 to
the program in calendar year 2002, primarily to oversee participating
nonprofit organizations. According to HUD, these staff years equated to
approximately $3.7 million in personnel costs. HUD officials said that in
the absence of the Discount Sales Program, these staff years would be
allocated to other activities.

4SAMS contains information on the properties acquired and sold by HUD.

5We use the term "staff year" to mean the amount of time a full-time
employee would work in the course of 1 year.

Our analysis of 238 homes sold under the Discount Sales Program in
calendar 2002 indicates it is likely that most of the homebuyers did not
benefit financially from the program.6 Specifically, assuming that
nonprofits and homebuyers would incur the same costs to rehab a property,
we estimate that 76 percent of the homebuyers would have spent less if
they had purchased the properties through HUD's regular sales process and
paid for the rehab work themselves. The estimated proportion of homebuyers
who did not benefit varied by discount level, ranging from more than 90
percent at the 15 percent discount level to one-half at the 30 percent
discount level. In part, homebuyers did not benefit because some
nonprofits overcharged for some properties and some program rules
authorize nonprofits to pass on costs-such as financing as closing
costs-that homebuyers would likely not incur under HUD's regular sales
process. We also found that while the program can provide access to a
range of services and assistance that may benefit homebuyers, such as
homeownership counseling, home maintenance courses, and down payment
assistance, these services are also available from other sources. HUD has
stated that the program generates benefits and supports policy goals, such
as neighborhood revitalization and stability, that extend beyond benefits
to individual homebuyers, but the agency has not studied the extent to
which the program is serving these ends.

HUD's monitoring efforts have uncovered numerous program violations, but
implementing the monitoring process and effectively enforcing program
rules have been difficult. For example, one of the HOCs' primary
monitoring tools is the desk review of the annual reports participating
nonprofits are required to submit each February. However, as of July 2003
the centers had not received annual reports from nonprofits responsible
for more than half of the 626 discounted properties that HUD estimates
were purchased, rehabbed, and resold under the program during calendar
year 2002. Further, as a condition of program participation, nonprofits
must allow HUD staff to perform on-site reviews of their operations.
However, the HOCs' use of these on-site reviews has been uneven. The
Atlanta and Denver centers reviewed about half of their approved
nonprofits in calendar year 2002, but the Philadelphia and Santa Ana
centers reviewed few and none, respectively, citing limited staff
resources and the difficulty of traveling to nonprofits' offices. But even
with these limitations, the

6These were the only properties for which HUD could provide the rehab
costs and selling prices to homebuyers as of July 2003. HUD estimates that
626 properties were purchased and resold under the program in calendar
year 2002.

HOCs' desk and on-site reviews identified numerous violations of program
requirements. For example, the centers' desk reviews found that nonprofits
often did not comply with HUD limits on the resale price of discounted
properties and earned excess profits from the program. Specifically, the
reviews showed that 28 of the 44 nonprofits that submitted annual reports
sold one or more properties at prices exceeding those allowed under the
program, resulting in 124 homebuyers being overcharged an estimated total
of $704,720. However, the centers have had limited success enforcing the
requirement that nonprofits making excess profits pay down the mortgages
of homebuyers they overcharged. As of July 2003, nonprofits had used less
than 10 percent of the estimated excess profits they made to pay down
homebuyers' mortgages.

This report recommends that the Secretary of HUD (1) evaluate options to
improve the program's benefit to homebuyers, the agency's monitoring of
nonprofits, and enforcement of excess profits requirements; (2) assess the
extent to which the program is meeting its objectives; and (3) terminate
the program if its current cost plus the resources needed to improve it
exceed the program's benefits. In comments on a draft of the report, HUD
agreed with our recommendations to further assess the program but added
that our analysis appeared to overstate the program's cost and understate
its benefits.

Background	Established in 1993, HUD's Discount Sales Program seeks to
expand affordable housing opportunities, help revitalize neighborhoods,
and reduce the agency's inventory of single-family properties in a timely,
efficient, and cost-effective manner.7 Under the program, approved
nonprofit organizations receive a discount when purchasing HUD-owned
single-family properties and are required to rehabilitate and resell them
to low- or moderate-income homebuyers. As of August 2003, 423 nonprofits
were approved to participate in the program, down from more than 2,000
nonprofits 3 years earlier. HUD attributed the reduction to changes in the
program, such as stricter approval standards and increased reporting
requirements, and to the agency's efforts to remove nonprofits that
violate program rules.

7HUD has broad authority to dispose of single-family properties in its
inventory. See 12 U.S.C. 1710(g).

HUD acquires properties through foreclosures of homes with FHA-insured
mortgages. FHA provides federally backed mortgage insurance primarily to
low-income and first-time homebuyers who might otherwise have difficulty
obtaining a mortgage.8 In calendar year 2002, HUD acquired more than
67,000 properties through foreclosures and sold approximately 65,000
properties from its inventory. At the end of calendar year 2002, HUD had
an inventory of 32,018 single-family properties.

HUD sells the properties in its inventory in as-is condition through a
number of different programs. Table 1 shows the primary programs HUD uses
to sell properties and the number of properties sold under each during
calendar year 2002. Most HUD-owned properties are eligible for price
reductions under the Discount Sales Program. The program accounted for
approximately 2 percent of HUD's overall property sales in calendar year
2002.

Table 1: HUD's Calendar Year 2002 Single-Family Property Sales

                          Program Number of properties

                              Regular sales 60,634

                            Officer Next Doora 1,168

                             Teacher Next Doora 925

                              Discount Sales 1,226

                            Asset Control Areab 795

                               Dollar Homesc 275

                                  Total 65,023

aThe Officer Next Door and Teacher Next Door programs offer HUD-owned
properties at 50 percent discounts to law enforcement officers and
teachers willing to live in economically distressed neighborhoods.

bUnder the Asset Control Area program, participating nonprofits agree to
purchase all of the HUDowned properties located within specific geographic
areas. The nonprofits receive discounts of up to 50 percent off HUD's list
price.

cThe Dollar Homes program allows local governments to purchase eligible
HUD-owned properties for one dollar. The properties made available through
the program are those HUD is unable to sell within 6 months.

8FHA insures most of its mortgages for single-family homes under its
Mutual Mortgage Insurance Fund, which is funded by borrowers' insurance
premiums and covers lenders' claims on foreclosed properties. The revenue
HUD receives from selling these properties is deposited in the fund.

HUD offers discounts of 10, 15, and 30 percent to nonprofit organizations.
The size of the discount depends on several factors: whether a property in
as-is condition is eligible for FHA insurance, whether it is located in a
revitalization area,9 and whether it is sold individually or in a package
of five or more homes. HUD inspects and appraises all foreclosed
properties to determine whether they are again eligible for FHA mortgage
insurance. FHA will insure mortgages only on properties that meet HUD's
minimum property standards and local building codes or that need less than
$5,000 in repairs in order to meet these standards. Properties needing
more than $5,000 in repairs are considered uninsurable. For purposes of
the Discount Sales Program, HUD then differentiates properties by
location. All insurable properties receive a 10 or 15 percent discount
whether or not they are located in a revitalization zone. The 15 percent
discount is only applied if the property is part of a group of five or
more properties purchased in a single transaction. Uninsurable properties
lying outside revitalization areas also receive these discounts, but those
located within revitalization areas are eligible for the steepest
discount-30 percent.

Under the Discount Sales Program, HUD has two methods of selling
properties to nonprofits: competitive bidding and noncompetitive sales.
Both methods allow discounts for nonprofits. Under the competitive
process, HUD establishes a list price for the properties but will accept
bids that are lower. HUD posts the properties, with their list prices, on
the Internet in its general listings and accepts bids from prospective
owneroccupants and nonprofits, but not investors, for a priority period of
10 to 30 days, depending on the geographic area. HUD awards the property
to the owner-occupant or nonprofit with the highest bid. If the highest
bidder is a nonprofit, HUD grants a 10 or 15 percent discount off the bid
price when it closes on the home. For properties that fail to sell during
this priority period, HUD then accepts bids from the general public,
including investors.

The noncompetitive sales method applies only to uninsurable properties.
HUD lists these properties separately from its general listings and makes
them available to nonprofits through a HUD contractor's Web site.
Nonprofits have a priority period of 5 days to express interest in the
properties at HUD's list price. If more than one nonprofit expresses
interest in a property, HUD selects the buyer by lottery. As with
competitive sales, the discount is applied at closing. Properties that are
not sold

9Revitalization areas are HUD-designated locations characterized by high
levels of foreclosures, very low incomes, and low homeownership rates.

noncompetitively are placed in HUD's general listings and made available
for sale on a competitive basis.

Nonprofits that purchase properties under the Discount Sales Program are
responsible for rehabilitating them as needed to meet HUD's minimum
property standards and local building codes. Nonprofits are required to
limit their resale price to no more than 110 percent of their "net
development cost," or the sum of their allowable costs for acquiring,
rehabilitating, and reselling the properties. Nonprofits must also sell
the homes to buyers whose incomes do not exceed 115 percent of their
area's median income, adjusted for family size.

HUD's four HOCs administer the Discount Sales Program and oversee the
participating nonprofits. The HOCs process nonprofits' applications to
participate in the program and monitor nonprofits for compliance once they
begin purchasing, rehabilitating, and reselling homes. To help monitor the
program, HUD requires nonprofits to submit annual reports to the
appropriate HOC by February 1 of each year. The reports must provide
information on properties the nonprofits have bought, rehabilitated, and
resold, including repair costs, prices to homebuyers, and homebuyers'
incomes. HOC staff also conduct on-site visits to nonprofits and
properties to review files and inspect repairs, among other things. When a
nonprofit fails to follow the program's requirements, HOCs may remove the
nonprofit from the program.

In recent years, HUD has changed some of its program requirements to
increase its oversight of nonprofits. For example, in 2000 HUD issued
guidance establishing uniform standards for nonprofits applying to
participate in the program. The guidance outlines specific information
nonprofits must submit in applying for the program, mandates that
nonprofits recertify with HUD every 2 years, and requires that nonprofits
answer detailed questions about their ability to carry out affordable
housing programs.

To address concerns raised in the HUD Inspector General's 2001 report on
the program, HUD issued additional guidance in December 2001 designed to
strengthen the program's reporting and accountability requirements. Until
this guidance was issued, nonprofits needed to meet HUD's annual reporting
and net development cost requirements only for properties purchased at a
30 percent discount. The guidance expanded annual reporting and resale
price requirements to all properties, regardless of

discount level, and clarified HUD's net development cost calculation by
providing a detailed list of allowable and unallowable costs.

Also in response to the Inspector General's report, HUD issued guidance in
January 2002 designed to tighten eligibility requirements for nonprofits.
Among other things, the guidance described circumstances that could create
a conflict of interest between nonprofits and their business partners. In
addition, it required that nonprofits be incorporated as 501(c)(3)
organizations for at least 2 years and have a minimum of 2 consecutive
years of affordable housing experience within the last 5 years. Finally,
to accommodate HUD's on-site reviews, the guidance required nonprofits to
maintain property records in a specified format.

HUD did not implement all of the Inspector General's recommendations.
Specifically, the Inspector General's report recommended that HUD suspend
the program and evaluate it to determine whether the program was viable or
should be discontinued. HUD Office of Housing officials told us they
developed a proposal for a contractor study of the program but that the
proposal was never funded. HUD officials said they did not suspend the
program because they felt the improvements made following the Inspector
General's review would prevent further problems.

In 2002, the Discount Sales Program Cost HUD at Least $18.8 Million

We estimate that the Discount Sales Program cost HUD between $18.8 and
$23.9 million in calendar year 2002. Most of this cost, between $15.1 and
$20.2 million, was a reduction in net revenue resulting from HUD's selling
properties through the program rather than through its regular sales
process.10 Personnel expenses for administering the program accounted for
the remaining $3.7 million of HUD's cost.

10We defined net revenue as the price for which HUD sells a property minus
HUD's holding and selling costs. These costs include discounts to
nonprofits, other homebuyer sales incentives, payments to management and
marketing contractors, property taxes, broker's fees and commissions, and
financing and closing costs. HUD was not able to provide us with
property-specific data on certain overhead expenses that it allocates
among the properties in its inventory. As a result, we did not include
these expenses in our analysis. Because these expenses are relatively
small, we believe this omission did not have a significant effect on our
estimates.

The Discount Sales Program Reduced HUD's Net Revenue

The net revenue HUD receives from each property it sells is less than the
property's selling price because HUD incurs certain holding and selling
costs. Some of these costs are common to both discounted and regular HUD
property sales, while others are not. For example, on a regular sale, HUD
pays the homebuyer's closing and financing costs and the sales commission
of the successful selling broker, within certain guidelines. HUD does not
pay either of these costs for properties sold through the Discount Sales
Program. HUD does not require a selling broker for the properties
nonprofits purchase through the program; consequently, there is generally
no selling broker's commission for these transactions. HUD incurs other
types of costs for all properties whether or not they are part of the
program. These costs include (1) the fees and reimbursable expenses it
pays to management and marketing contractors responsible for inspecting,
appraising, securing, maintaining, and selling HUD-owned properties; (2)
sales incentives in the form of cash allowances that HUD periodically
offers to homebuyers-including nonprofits-that close relatively quickly on
executed sales contracts; (3) the listing broker's fee; and (4) the
property taxes for the period when HUD owned the home.

To determine the impact of the Discount Sales Program on HUD's net
revenues (i.e., the selling price minus holding and selling costs) we
compared the estimated net revenues HUD received for the discounted
properties with the estimated net revenues HUD would have received if it
had sold the properties through its regular sales process.11 Using data
from HUD's SAMS and the U.S. Census Bureau, we made this determination for

11We derived our estimates from a model we developed that predicts the net
revenue HUD received from each property it sold during calendar year 2002
as a function of several variables, including whether HUD sold the house
through the Discount Sales Program. The model allowed us to estimate the
effect of the program on HUD's net revenue, while holding constant the
effects of other factors. We made this estimate by comparing (1) the net
revenue our model predicted HUD would have received for each discounted
property had the property been sold through the regular sales process with
(2) the net revenue our model predicted HUD would have received by selling
the property through the Discount Sales Program. Even though we knew the
actual amount HUD received by selling the property through the Discount
Sales Program, we made the comparison between two estimated values in
order to be consistent. Each estimated value contained an error term that
captured the effects of omitted variables unavailable for the modeling
process. By comparing two estimated values, we removed the influence of
the omitted variables from our comparison, leaving the effect of the
Discount Sale Program. Appendix II provides additional information about
our model.

1,19412 properties that HUD sold through the program in calendar year
2002. We used a statistical model that included data for these properties
and approximately 4,00013 properties HUD sold through its regular process
during the same year.14 We found that by selling the 1,194 properties
through the Discount Sales Program instead of its regular sales process,
HUD reduced the net revenue it received in calendar year 2002.
Specifically, we estimate that the total reduction in HUD's net revenue
was between $15.1 and $20.2 million,15 an average of between $12,672 and
$16,945 per property. (See app. II for a detailed discussion of our
statistical analysis.) Without the program, and with all other things
remaining equal, cash flows into HUD's insurance fund would have increased
by that amount.

As shown in table 2, the overall and average reductions in net revenue
varied according to the discount level. The properties sold with 10
percent discounts accounted for about two-thirds of the homes that HUD
sold through the program in calendar year 2002, but for less than half the
total estimated reduction in net revenue. In contrast, the properties sold
with 30 percent discounts represented less than one-third of the total
properties sold but more than 40 percent of the overall reduction in net
revenue. Finally, the properties with 15 percent discounts accounted for
about 9 percent of the total properties and between 7 and 10 percent of
the overall reduction in HUD's net revenue.

According to HUD officials, the agency's database somewhat overstates the
number of properties sold with a 10 percent discount (the most common
type) and somewhat understates the number sold with a 15 percent discount
(the least common type). The officials said this overstatement occurs
because HUD does not always update its database to reflect the fact

12HUD sold a total of 1,226 properties through the Discount Sales Program
in calendar year 2002. Because we were unable to match 32 of the
properties to census tract data, we dropped them from our analysis,
leaving 1,194 properties.

13These were all of the homes HUD sold through its regular process that
were located in the same census tracts as the properties it sold through
the Discount Sales Program.

14Our analysis assumed that in the absence of the program, HUD would have
sold the 1,194 properties through its regular sales process. Accordingly,
our statistical analysis excluded data on properties that were not sold
through either the Discount Sales Program or HUD's regular process.

15We are 90 percent confident that HUD's actual loss in net revenue was
between these two values.

that a property, indicated in the database as being sold with a 10 percent
discount, may actually have been sold at a 15 percent discount if it was
part of a group of five or more properties bought in a single transaction.
As a result, our analysis may overestimate the reduction in net revenue
for properties discounted by 10 percent and underestimate it for those
discounted by 15 percent.

Table 2: Estimated Reduction in Net Revenue Due to the Discount Sales
Program in Calendar Year 2002, by Discount Level

                                            Total estimated Average estimated 
                                           reduction in net  reduction in net 
                             Number of             revenues      revenues per 
        Discount level properties sold (dollars in                   property 
                                       millions)            
            10 percent             744          $6.6 - $9.3  $8,844 - $12,453 
            15 percent             102            1.0 - 1.9   10,223 - 18,997 
            30 percent             348            7.5 - 9.0   21,572 - 25,947 
                 Total           1,194        $15.1 - $20.2 $12,672 - $16,945 

Sources: HUD and the U.S. Census Bureau.

Note: Our confidence level for the estimates is 90 percent.

As shown in table 3, the overall and average reductions in net revenues
also varied by HOC. A major reason for this variance was differences among
the centers in the proportion of their properties sold with 30 percent
discounts. Because 30 percent discount properties cost HUD more in net
revenue than properties at the other discount levels, the HOC with the
highest proportion of 30 percent properties-Santa Ana-had the greatest
reduction in net revenue. The HOC with the lowest proportion-Denver- had
the smallest reduction in net revenue.

Table 3: Estimated Reduction in Net Revenue Due to the Discount Sales
Program in Calendar Year 2002, by Homeownership Center

                                            Total estimated Average estimated 
                                           reduction in net  reduction in net 
        Homeownership       Number of              revenues      revenues per 
               center properties sold (dollars in millions)          property 
              Atlanta             296           $2.8 - $3.8   $9,317- $12,744 
               Denver             183             1.2 - 2.3    6,551 - 12,689 
         Philadelphia             359             3.5 - 4.7    9,712 - 13,061 
            Santa Ana             356             7.7 - 9.4   21,592 - 26,542 
                Total           1,194         $15.1 - $20.2 $12,672 - $16,945 

Sources: HUD and the U.S. Census Bureau.

Note: The estimates may not add up to totals due to rounding. Our
confidence level for the estimates is 90 percent.

HUD officials told us they were aware that the program reduced the
agency's net revenue from property sales. However, HUD has not evaluated
whether the program is reducing HUD's property inventory in a timely,
efficient, and cost-effective manner, as intended.

HUD Incurred Administrative Costs Operating the Discount Sales Program

To determine the impact of the Discount Sales Program on HUD's
administrative costs in calendar year 2002, we compared the administrative
costs HUD incurred under the program to what HUD would have incurred had
the discounted properties been sold through HUD's regular process.
According to HUD, in administering the Discount Sales Program, HOC staff
perform tasks that are not part of HUD's regular home-selling process. For
example, for the Discount Sales Program, center staff approve and
recertify participating nonprofit organizations and monitor the
nonprofits' compliance with program requirements-tasks they do not perform
for the regular sales process. As a result, HUD's administrative cost per
property is lower for its regular sales process than it is for the
Discount Sales Program. HUD officials told us that for this reason and the
small volume of properties sold through the Discount Sales Program,
selling the discounted properties through HUD's regular process would have
had no measurable effect on the administrative costs for the regular sales
process.

The bulk of HUD's administrative costs for the Discount Sales Program are
the salaries and benefits of staff who work on the program. According to
HUD officials, most of these staff split their time among several
programs,

but HUD's time and attendance system does not record the time they spend
on each one. Therefore, we relied on estimates from HUD to determine how
many staff years were spent on the program in calendar year 2002 and the
associated costs. Although HUD incurred other types of costs to administer
the program, such as mailing costs and travel expenses for visiting
nonprofit offices, these were minor compared with the personnel costs and
were not included in HUD's estimate.

HUD estimates that its personnel costs for the Discount Sales Program were
approximately $3.7 million in calendar year 2002.16 (See table 4.) HUD's
estimate was based on information provided by its HOCs and Office of
Housing, which showed that their staffs devoted a total of 45 staff years
to the program. The number of staff years and the associated cost varied
across offices, however. Among the HOCs, the Atlanta center had the most
staff years and highest personnel costs and the Denver center the fewest
staff years and lowest personnel costs. HUD's Office of Housing devoted
the equivalent of about one staff year to the program. HUD headquarters
and HOC officials told us that in the absence of the Discount Sales
Program, these staff years would have been dedicated to administering
other HUD programs, so that HUD would have incurred the personnel costs
with or without the program.

  Table 4: HUD's Estimate of Personnel Costs for the Discount Sales Program in
                  Calendar Year 2002, by Homeownership Center

               Atlanta  Denver  Philadelphia    Santa          HUD 
                                                  Ana              
                center  center        center   center headquarters      Total 
    Staff         17.0   5.2            11.8     10.4          0.8 
    years                                                          
  Personnel $1,386,740 $423,930   $957,919   $844,599      $75,263 $3,688,451 
    costs                                                          
Source:                                                         
    HUD.                                                           

16HUD officials stressed that the estimate of personnel costs was only a
rough approximation of the agency's actual costs.

The Discount Sales Program Is Not Likely to Benefit Most Homebuyers
Financially but Can Help Them Access Homeownership Services and Assistance

Our analysis of 238 properties sold under the Discount Sales Program in
calendar year 2002 indicates it is likely that most homebuyers did not
benefit financially from the program. Specifically, assuming that
nonprofits and homebuyers had the same rehab costs, we estimate that 76
percent of the homebuyers would have spent less if they had purchased the
properties through HUD's regular sales process and paid for the rehab work
themselves. In part, the lack of financial benefit to homebuyers is
attributable to the program's rules, which authorize nonprofits to pass on
costs that homebuyers would likely not incur using the regular sales
process. Despite the program's limited financial benefits, it may help
homebuyers access a range of services and assistance-such as homeownership
counseling, down payment assistance, and home maintenance courses-that are
beneficial but also available from other sources. The program may also
help improve neighborhood conditions by rehabbing and putting back on the
market homes that might otherwise remain vacant or in disrepair.

The Financial Benefit of the Program to Most Homebuyers is Doubtful

To determine the extent to which homebuyers benefited financially from
purchasing a rehabilitated home through the Discount Sales Program, we
performed a statistical analysis comparing what the homebuyers actually
paid for these homes with our estimate of what they would have spent had
they purchased the homes under HUD's regular process and paid for the
rehab work themselves. Our analysis assumed that in the absence of the
Discount Sales Program, a homebuyer would be able to (1) purchase the same
home and rehabilitate it to the same extent as the nonprofit and (2) incur
the same rehab costs as the nonprofit.17 We also assumed that the
homebuyer would inhabit the home during the rehabilitation and therefore
would not incur housing expenses for two residences during that period.18
We performed this analysis on 238 properties that nonprofits purchased

17HUD does not collect the data necessary to test this assumption. In
addition, the data HUD provided us on nonprofits' rehab costs did not
always include other, relatively minor, expenses that an individual
homebuyer could also incur, such as fees for rehab consultants,
construction permits, and termite inspection services. However, as
discussed later in this section, our overall finding does not change even
under the assumption that a homebuyer would pay significantly more than a
nonprofit to rehab a home.

18While this assumption would not hold true in all cases, more than half
of the properties covered by our analysis received less than $15,000 in
rehab work and some received none at all, suggesting that the rehab work
was generally not so extensive as to require vacating the home.

and resold between February 1 and December 31, 2002.19 These properties
were the only ones for which HUD could provide the rehab costs and selling
prices to homebuyers at the time of our review. (See app. II for a
detailed discussion of our statistical analysis.)

Assuming equal rehab costs for nonprofits and homebuyers, we estimate that
182 of the 238 homebuyers, or 76 percent, did not benefit financially by
purchasing a rehabbed property from a nonprofit that bought the property
through the Discount Sales Program. That is, the buyers would have spent
less had they purchased the property through HUD's regular sales process
and paid for the rehab work themselves. Under the assumptions of our
analysis, we estimate that these purchasers spent an average of $9,200
more buying the house through the program than they would have spent
otherwise. Our analysis indicated that the other 24 percent of the
homebuyers benefited financially from the program, because purchasing the
homes through HUD's regular sales process and rehabbing them would have
been more expensive. We estimate that these homebuyers saved $9,200, on
average, by purchasing through the Discount Sales Program. Because
nonprofits may, in some circumstances, be able to rehab a home more
cheaply than an individual homebuyer, we also performed the analysis
assuming that a homebuyer would pay 25 percent more than a nonprofit for
the same rehab work. Even under that assumption, we estimate that 59
percent of the homebuyers would not have benefited financially from the
program. More specifically, we estimate these purchasers spent an average
of $8,000 more buying the house through the program than they would have
spent otherwise. We estimate that the remaining 41 percent of homebuyers
saved $10,100, on average, by purchasing a home through the program.

Our estimates of the extent to which homebuyers did or did not benefit
varied according to the discount level of the property purchased. Assuming
equal rehab costs for nonprofits and homebuyers, we estimate that 79
percent of the homebuyers purchasing houses that had been discounted 10
percent saw no financial benefit. For the properties with 15 percent
discounts, we estimate that more than 90 percent did not benefit. However,
for the properties with 30 percent discounts, we estimate that one-half of
the homebuyers saw some financial benefit. (See table 5.)

19We used this time frame because HUD made significant changes to the
program that became effective on February 1, 2002. For example, HUD
expanded its annual reporting and resale price requirements to properties
sold with 10 and 15 percent discounts.

One reason homebuyers did not benefit financially, according to our
analysis, was that nonprofits sometimes resold the properties for more
than the program allowed.20 This finding was especially strong for
purchases in the 15 and 30 percent discount categories. Had the nonprofits
not overcharged the homebuyers in these cases, we estimate that more than
one-third of the homebuyers who bought properties with a 15 percent
discount and more than three-quarters of those who bought properties with
a 30 percent discount would have benefited financially.

Table 5: Estimated Number of Homebuyers Who Did and Did Not Benefit
Financially by Purchasing Homes through HUD's Discount Sales Program

                                                         Number of homebuyers 
                        Number of Number of homebuyers    who did not benefit 
        Discount level properties who benefited                   financially 
                                  financially            
            10 percent        143                     30                  113 
            15 percent         51                      4 
            30 percent         44                     22 
                 Total        238                     56                  182 

Sources: HUD and the U.S. Census Bureau.

Our analysis did not take into account certain factors that are difficult
to quantify but may make the program either more or less beneficial from a
homebuyer's perspective. For example, some homebuyers may be willing to
incur significant costs to avoid the time, difficulty, and inconvenience
involved in selecting materials, obtaining and evaluating contractor bids,
residing in a property undergoing rehab work, and possibly obtaining a
separate loan to finance the rehab work.21 Conversely, some homebuyers may
not view these tasks as major obstacles and may see significant benefits
to controlling the rehab process themselves, such as the ability to select
the materials used and the ability to oversee the rehab work as it
progresses.

20The issue of nonprofits violating resale price restrictions is discussed
in more detail in the next section of the report.

21Although a homeowner would incur fees and interest costs for such a
loan, homebuyers financing properties purchased through the Discount Sales
Program incur similar expenses because the cost of the rehab work is
included in the nonprofit's selling price and therefore is reflected in
the homebuyer's mortgage costs.

The Discount Sales Program's Rules Reduce the Likelihood That Buyers Will
Benefit Financially

Some of the Discount Sales Program's rules make it unlikely that
purchasing a property from a nonprofit that purchased the property from
HUD at a 10 or 15 percent discount will benefit homebuyers more than
purchasing the same property from HUD through the regular sales process.
For example, HUD allows nonprofits to resell discounted properties for up
to 110 percent of the "net development cost," or the cost of buying the
property plus allowable rehab, holding, and selling costs. The 10 percent
markup helps nonprofits to cover the overhead expenses they incur by
participating in the program. However, taking a 10 percent markup on a
property purchased at a 10 or 15 percent discount effectively cancels out
all or most of the discount. As a result, the price of the home to the
eventual homebuyer reflects little, if any, of HUD's discount to the
nonprofit.

Furthermore, program rules authorize nonprofits to include in their
calculations of net development cost certain "allowable" financing and
closing costs they incur in buying discounted HUD properties. As a result,
a homebuyer who purchases a property from a nonprofit pays not only his or
her own financing and closing costs but-through the sales price-the
nonprofit's as well. In contrast, when a homebuyer purchases a property
using HUD's regular sales process, HUD pays allowable financing and
closing costs on the buyer's behalf. Also, a nonprofit's net development
cost may include the principal and interest payments for the mortgage on
the property while the property is being renovated for up to 6 months.
Raising the nonprofit's net development cost effectively raises the price
for the eventual homebuyer, who could have avoided some of these expenses
by purchasing the house directly from HUD and, if possible, inhabiting it
during the renovation.

The Discount Sales Program May Help Homebuyers Access Certain Services and
Assistance and Improve Neighborhood Conditions

According to HUD and nonprofit officials, many of the families who
purchase properties through the program are first-time homebuyers.
Accordingly, HUD strongly encourages nonprofits to provide homeownership
counseling services and requires participants to submit "affordable
housing plans" detailing, among other things, the services and assistance
that low-and moderate-income homebuyers using the program can expect to
receive. During our visits to HUD's homeownership centers, we reviewed the
affordable housing plans for a judgmental sample of 17 nonprofits. The
plans showed that the nonprofits offered a wide range of services and
assistance to homebuyers, either directly or through referrals to other
agencies. The services included mortgage credit counseling, "hotlines"
homebuyers could call with questions, and courses on budgeting

and home maintenance. Some nonprofits also offered assistance with down
payments and closing costs. HOC staff told us that these services and
assistance were typical of those provided by most participating
nonprofits.

Both HUD and nonprofit officials told us they believed that providing such
services to new homebuyers facilitated homeownership among low-income
families that might otherwise have a hard time purchasing a home. These
officials also said that the services helped minimize the likelihood of
default by preparing families for the responsibilities of homeownership.
However, we found that similar services were widely available outside the
Discount Sales Program. For example, HUD itself provides financial support
to hundreds of housing counseling agencies across the country. Any
prospective homebuyer can access these services at no cost.

According to HUD, the Discount Sales Program also generates benefits and
serves policy objectives, such as neighborhood revitalization and
stability, that extend beyond the individual households that purchase
properties. Some HUD and nonprofit officials told us they believe that the
Discount Sales Program may help to improve neighborhood conditions by
supporting the rehabilitation and sale of properties that would otherwise
be vacant and in disrepair, reducing surrounding property values, and
becoming magnets for vandalism and trespassing. For example, one nonprofit
official told us that by purchasing and rehabbing multiple properties over
a period of several years, the organization had not only improved the
housing stock of one community but also helped create an environment that
encouraged economic development and social service opportunities nearby.
HUD officials also told us that many prospective owner-occupants are not
willing to purchase homes requiring significant rehab work because of the
difficulty and risks of undertaking a rehab project. They said that even
if owner-occupants were to purchase these properties, they might do less
rehab work than a nonprofit would or not rehab them at all. Finally, HUD
believes that by promoting homeownership and property rehabilitation, the
program has stabilizing effects on neighborhoods and contributes to
property values-factors that reduce the risk of foreclosure and losses to
FHA's insurance fund. Although expansion of homeownership opportunities
and neighborhood revitalization are objectives and potential benefits of
the program, HUD has not studied the extent to which the program is
serving these ends.

Limited Monitoring Efforts Have Uncovered Numerous Violations, but
Effective Enforcement Has Been Difficult

The HOCs use two monitoring tools to assess nonprofits' compliance with
program requirements: desk reviews of annual reports and on-site
evaluations. However, the HOCs had trouble conducting desk reviews because
many program participants turned their reports in late or not at all, and
many reports were incomplete. In addition, the HOCs' use of on-site
reviews was uneven, with two centers conducting them routinely and the
other two doing few or none. Even with these problems, the HOCs'
monitoring efforts uncovered numerous violations of program rules, such as
making excess profits by reselling discounted properties for more than the
program allowed. The HOCs have removed many nonprofits for noncompliance
but lack an effective mechanism for enforcing requirements concerning
excess profits.

Effectiveness of Desk Reviews is Hampered by Missing or Incomplete
Information

Nonprofits participating in the Discount Sales Program are required to
submit annual reports to the HOCs each February 1 that provide information
on the properties purchased under the program the previous calendar
year.22 The required information includes the status of the property
(i.e., whether it has been rehabbed and resold), the rehab costs, and the
selling price to the homebuyer. Nonprofits must also provide
documentation, such as settlement statements, giving detailed financial
information on the purchase and resale of the properties.

Desk reviews of these reports are HUD's primary method of determining
whether nonprofits comply with key program requirements, such as those
restricting the resale prices of rehabilitated homes and the purchasers'
income levels. However, the effectiveness of desk reviews as a monitoring
tool has been limited because many nonprofits have not submitted annual
reports on time, and others have provided incomplete information.
Specifically, as of July 2003-more than 5 months after the annual reports
were due-HUD lacked reports from nonprofits accounting for more than half
of the 626 properties it estimates were bought, rehabbed, and resold in
calendar year 2002. The HOCs had reports for properties resold to
homebuyers from only 44 of the 166 nonprofits that purchased discounted
properties in calendar year 2002. Other nonprofits submitted reports that
lacked all of the data the HOCs needed to assess the participants'

22HUD has required annual reports on properties sold at a 30 percent
discount since December 1994. Beginning with reports due in February 2003,
HUD expanded the requirement to include all properties sold at a discount.

compliance with program requirements. For example, as of April 2003, more
than half of the annual reports received at the Denver HOC did not contain
the information necessary to determine the nonprofits' net development
costs for the properties. As a result, staff could not determine whether
the nonprofits had sold their properties at prices that were within
program limits. Similarly, more than one-third of the reports received by
the Atlanta HOC as of July 2003 did not contain the required certification
of the homebuyers' income levels. Without this information, the HOC had no
assurance that the properties were sold to homebuyers with low and
moderate incomes, as required.

According to HOC officials, efforts to collect missing data and resolve
other reporting issues can take months. For example, staff at the Atlanta
HOC told us that they spent large amounts of time calling and writing
nonprofits to gather the information missing from the annual reports.
Although three of the HOCs-Atlanta, Denver, and Santa Ana---had originally
planned to complete their desk reviews by the end of April 2003, these
reviews were still under way in mid-July 2003. At that time, the remaining
HOC-Philadelphia-had completed reviews of just 16 percent of its
properties.

The reporting problems occurred despite the HOCs' efforts to remind
nonprofits of the reporting requirements and to provide training on the
program rules. For example, all four HOCs sent reminder letters to
nonprofits several weeks before the annual reports were due. In addition,
HOC officials told us they provided either one-on-one or group training to
nonprofits on submitting annual reports and meeting other program
requirements. HOC officials speculated that a major reason for the
reporting problems was that many nonprofits lacked adequate administrative
capacity. However, they also said that carrying out in-depth assessments
of administrative capacity as part of the initial approval process would
require a costly on-site evaluation of every applicant.

In October 2003, HOC officials told us that as a result of their follow-up
efforts, they had made significant progress in obtaining annual reports
from nonprofits that had not reported earlier in the year. However, during
the long time it takes the HOCs to obtain and review annual reports,
nonprofits may continue to purchase more homes and violate program rules.
In addition, according to HOC officials, many nonprofits that had failed
to report had either withdrawn or been removed from the program, leaving
little incentive to report. Consequently, it is unlikely that HUD will

ever know whether these nonprofits followed program requirements in
rehabbing and reselling discounted properties.

The Number of On-Site Reviews Varied Across Homeownership Centers

HUD's guidelines not only require that nonprofits allow on-site reviews of
their operations as a condition of program participation but also outline
the types of records participants must maintain and make available for
HUD's on-site review. On-site reviews generally allow for a more in-depth
assessment of a nonprofit's program activities than desk reviews. For
example, on-site reviews may include examining invoices and cancelled
checks to determine the validity of claimed rehab costs. They may also
involve inspection of rehabbed homes and interviews with homeowners.

We found that only two of the four HOCs-Atlanta and Denver-routinely used
on-site reviews as a monitoring tool in calendar year 2002. Atlanta HOC
officials told us that they tried to review each of their medium- and
high-risk nonprofits at least once every 2 years.23 Consistent with this
policy, the center performed 22 on-site reviews in calendar year 2002,
covering about half of the nonprofits that had purchased discounted homes
that year. The Atlanta HOC had trained staff stationed throughout the
center's geographic jurisdiction to perform the on-site reviews and also
employed two specialists with backgrounds in home construction. The Denver
HOC performed on-site reviews of 10 of the 23 nonprofits it sold
properties to in calendar year 2002. Officials there said that they
targeted nonprofits using desk reviews, homebuyers' complaints, and
applications to the program. The reviews were performed by staff working
out of the HOC, with assistance from other HUD staff stationed near the
nonprofits.

In contrast to the Atlanta and Denver centers, the other two HOCs-
Philadelphia and Santa Ana-used on-site reviews rarely or not at all. The
Philadelphia HOC, which sold properties to 48 nonprofits in calendar year
2002, conducted just two on-site reviews during that year, citing a
shortage of staff as the primary reason. Center officials told us that
they plan to use a contractor to conduct on-site reviews of nonprofits
with known performance problems and that the contractor will be required
to have construction specialists assist in these reviews. The Santa Ana
HOC, which sold properties to 52 nonprofits in calendar year 2002, did not
perform any on-site reviews. Santa Ana HOC officials told us that the
center's large

23The Atlanta HOC designates nonprofits that purchase five or more
properties in a year as medium- or high-risk.

geographic jurisdiction made it impractical for them to travel to
nonprofits' offices. To compensate for the lack of on-site reviews, the
Santa Ana center uses an in-depth version of the desk review, requiring
nonprofits to submit large amounts of supporting documentation, including
invoices, with their annual reports. A Santa Ana official said that this
is the same documentation that HOC staff examine during on-site reviews
but acknowledged that desk reviews do not include property inspections.

Despite Monitoring Limitations, HOCs Uncovered Numerous Program Violations

The HOCs identified violations of program requirements during both desk
and on-site reviews. Primary among these were violations of the ceiling
for resale prices: 110 percent of the net development cost.24 The desk
reviews the HOCs had conducted as of July 2003 showed that nonprofits
often did not comply with the resale restriction. This problem occurred
despite the fact that in December 2001 HUD had issued guidance to
nonprofits clarifying net development costs and providing detailed
instructions for calculating them.

As shown in figure 1, 28 of the 44 nonprofits that had submitted annual
reports as of July 2003 overcharged homebuyers for one or more properties.
These violations occurred on about 124 (47 percent) of the 265 properties
covered by the annual reports and resulted in homebuyers being overcharged
an estimated total of $704,720.25 The amount of the estimated overcharges
varied significantly from property to property, ranging from under $10 to
more than $40,000. For example, one nonprofit closed on a discounted home
in New York for $117,600 in October 2002 and spent about $41,000 to rehab
the property. The nonprofit subsequently resold the property for $234,000,
or $43,333 more than the program allowed. Assuming the homebuyer had
secured a 30-year loan at 6 percent interest (the prevailing rate at the
time the homebuyer made the purchase), the overcharge increased the
homebuyer's annual mortgage payments by more than $3,100.

24In purchasing a discounted property, a nonprofit must sign an addendum
to the sales contract stating that it will adhere to this rule.

25This figure is an estimate because as of July 2003 the HOCs had not made
final determinations of excess profits for all of the properties they had
reviewed. As a result, the actual amount of excess profits for these
properties may be higher or lower than this amount.

Figure 1: Resale Price Violations Disclosed by Desk Reviews of Nonprofits'
Annual Reports on Calendar Year 2002 Program Activity

Denotes number of nonprofits or properties with violations out of total.

Source: HUD.

The HOCs' desk reviews also identified three cases in which nonprofits
violated program requirements by reselling discounted properties to
homebuyers whose incomes exceeded 115 percent of the area median income.
For example, the Denver HOC found that one of its nonprofits sold a Texas
home to a buyer whose income was 141 percent of the area median income,
adjusted for family size.

The HOCs' on-site reviews also revealed instances of serious
noncompliance, underscoring the importance of these reviews as a
monitoring tool. Among these violations were a lack of auditable records,
unallowable rehabilitation costs, and conflicts of interest between
nonprofits and their rehabilitation contractors. For example, one review
disclosed that the nonprofit had made bulk purchases of the materials it
needed to rehabilitate discounted properties but, contrary to program
requirements, did not maintain records showing the costs of these
purchases or the materials that were used for each home. As a result, the
reviewers could not verify the net development cost for any of these
properties. In another on-site review, the Atlanta staff found that the
nonprofit was not in control of the day-to-day operations of its discount
property purchases. Instead, the nonprofit allowed its affiliated
contractors and realtors to control the buying, rehabbing, and selling of
the properties

acquired through the Discount Sales Program and to share in the profits
from the sale.

HOCs Have Removed Violators from the Program but Lack Effective Means to
Enforce Requirements on Excess Profits

HOCs Frequently Removed Nonprofits from the Program, but the Process Has
Limitations

To hold nonprofits accountable for program violations, the HOCs may use
two main enforcement tools: (1) removing participants from the program and
(2) requiring them to use excess profits to pay down overcharged
homebuyers' mortgages. The HOCs have often exercised their authority to
remove nonprofits but lack an effective mechanism for enforcing
requirements concerning excess profits. As a result, some nonprofits that
did not comply with program requirements have retained excess profits, and
homebuyers who were overcharged have not received financial restitution.

HUD issued regulations in June 2002 authorizing the agency to remove a
nonprofit from its roster of approved organizations for any cause HUD
judged to be detrimental to the agency or to any of its programs.26 These
causes include failure to comply with HUD guidance and instructions and
failure to respond within a reasonable time to HUD inquiries, including
requests for documentation. In recent years, the HOCs frequently used
removal to hold nonprofits in the Discount Sales Program responsible for a
variety of compliance problems. In calendar year 2002, for example, the
Santa Ana HOC removed 31 nonprofits from the program, mostly for failure
to submit annual reports, conflicts of interest, selling homes for more
than program limits or to families that were not low-or moderate-income,
and lack of administrative or financial capacity. In calendar year 2003,
all four HOCs removed nonprofits that had failed to file annual reports or
committed other program violations. For example, as of October 2003, the
Atlanta HOC had removed 18 nonprofits from the program-more than one-third
of the nonprofits that had purchased discounted properties in calendar
year 2002. The other three HOCs removed a total of 63 nonprofits that had
purchased properties that year.

Despite its importance as an enforcement tool, HOC officials told us that
removing nonprofits from the program had significant limitations. First,
the process can take months to complete, allowing nonprofits to continue
purchasing properties and possibly to commit additional violations. HOC

26Prior to that time, HUD did not have regulatory procedures for taking
this action and removed nonprofits through a less formal process.

staff must carefully document their justification for the action and must
follow due process procedures that can be lengthy, particularly if a
nonprofit appeals the removal decision. HOC officials said that once they
decide to remove a nonprofit, they often restrict the number of properties
it may purchase in an effort to reduce the potential for further
noncompliance during the due process period. Second, once a nonprofit is
removed, it has little incentive to report to HUD on the discounted
properties it resold or to surrender any of the excess profits it may have
earned. As a result, the centers may never learn whether these properties
were resold at reasonable prices to low- or moderate-income homebuyers,
and the homebuyers who were overcharged for their properties do not
receive any financial restitution.

HOCs Had Limited Success In December 2001, HUD issued instructions
requiring nonprofits to sign an

Recovering Excess Profits	addendum to every sales contract that limited
the resale price of discounted homes to 110 percent of the net development
cost. The instructions also mandated that nonprofits use the excess
profits they earn by exceeding the 110 percent limit to pay down the
mortgages of the homebuyers they overcharged.

The four HOCs have attempted to implement this requirement by requesting
mortgage "pay downs" from nonprofits that are making excess profits.
However, Philadelphia HOC officials and attorneys from HUD's Office of
General Counsel also told us that the agency's authority to enforce the
requirement was not specified in regulation and was therefore in question.
Furthermore, the attorneys said that as a practical matter the requirement
would be difficult to enforce, as HUD would have to refer nonprofits that
refused to comply to the Department of Justice for legal action. The
officials said they doubted whether Justice would accept these cases
because the amounts of money involved are generally relatively small-often
less than $10,000-and it would be cost-prohibitive for Justice's attorneys
to pursue them. HUD officials added that obtaining enough documentation to
build a convincing legal case was difficult, because many nonprofits do
not keep proper financial records that adequately document the amount of
excess profits earned.

As a result of these problems, the HOCs had limited success getting
nonprofits to use excess profits to pay down homebuyers' mortgages.
According to HOC officials, the nonprofits that have paid down mortgages
did so voluntarily because they wanted to stay in the program. As of July
2003, the HOCs' desk reviews had identified 28 nonprofits that made an
estimated total of $704,720 in excess profits. At that time, seven of the

nonprofits had made combined mortgage pay downs of $62,002, or only about
9 percent of the total estimated excess profits. (See table 6.)

Table 6: Percentage of Estimated Excess Profits Used to Pay Down Homebuyer
Mortgages, as of July 2003

                       Number of                     Amount of  Percentage of 
                      properties                     estimated      estimated 
                      resold for                excess profits excess profits 
                       more than                 used to pay      used to pay 
        Homeownership    program      Estimated           down           down 
               center     limits excess profits      mortgages      mortgages 
              Atlanta         27       $123,059             $0             0% 
               Denver         45       $253,926        $33,330            13% 
         Philadelphia         18       $186,942             $0             0% 
            Santa Ana         34       $140,793        $28,672            20% 
                Total        124       $704,720        $62,002             9% 

Source: HUD.

Note: The estimated excess profits are for discounted properties that HUD
sold between February 1 and December 31, 2002, which nonprofits resold
that same year.

HOC officials told us that although some of the remaining 21 nonprofits
had withdrawn or were removed from the Discount Sales Program, others were
still being evaluated by HOC staff and were continuing to purchase
discounted properties. Moreover, these nonprofits retained all of the
excess profits they had made. For example, one nonprofit still under
review as of October 2003 made an estimated $28,700 in excess profits on
five discounted properties that it resold in 2002. The nonprofit did not
pay down any mortgages, purchased 27 additional properties in 2003, and is
still in the program. HUD's inability to enforce its requirements on
excess profits in a vigorous and timely manner not only deprives
homebuyers that have been overcharged but also puts the financial
interests of other prospective homebuyers at risk.

Conclusions	HUD's Discount Sales Program is intended, among other things,
to help make homes affordable for low-and moderate-income homebuyers.
However, deficiencies in the program's design and implementation have
undermined its ability to serve this end. Our analysis suggests that the
program is of questionable benefit to most homebuyers. In addition, the
HOCs' monitoring and enforcement efforts do not adequately ensure that

nonprofits are complying with requirements designed to protect homebuyers'
financial interests. And despite the program's significant cost, HUD has
not determined whether the program is meeting its objectives of expanding
affordable housing opportunities, helping to revitalize neighborhoods, and
reducing HUD's property inventory in a timely, efficient, and
cost-effective manner.

Measures to address the cost, benefit, and compliance issues raised in
this report are likely to be expensive or have adverse effects. For
example, to reduce the cost of the program, HUD could reduce the size of
its discounts, dedicate fewer staff resources to the program, or both.
However, these actions would likely reduce the program's financial benefit
to homebuyers and weaken HUD's ability to oversee participating
nonprofits. Conversely, to increase financial benefits to homebuyers, HUD
could increase the size of its discounts, but doing so would effectively
raise the cost of the program.

Taken together, the program's problems and the lack of clear solutions
raise serious questions about whether HUD should continue to operate it.
We recognize that contemplating the termination of the program involves
trade-offs. In the absence of the program, it is possible that some
individuals and families would have difficulty purchasing suitable homes
and that neighborhood conditions would suffer if HUD-owned properties were
not rehabilitated or sold as quickly. However, unless significant changes
are made to the program, it will likely continue to experience the
problems raised in this report and by HUD's Inspector General.

Recommendations for Executive Action

GAO recommends that the Secretary of HUD take the following two actions:

o 	Evaluate options to improve the program's benefit to homebuyers, the
agency's monitoring of nonprofits, and enforcement of excess profits
requirements.

o  Assess the extent to which the program is meeting its objectives.

If the Secretary determines that the current cost of the program plus the
resources needed to improve it exceed the program's benefits, GAO
recommends that the program be terminated.

Agency Comments and Our Evaluation

We provided HUD with a draft of this report for review and comment. In a
letter from the Assistant Secretary for Housing (see app. I), HUD agreed
with our recommendations to further assess the program and said it would
proceed accordingly. Also, HUD disagreed with some aspects of our
methodology and said that our analysis overstated the program's costs and
understated its benefits. Lastly, HUD said that the report should
acknowledge the significance of the multiple public policy objectives the
program serves. However, HUD did not respond to our third recommendation
concerning the possible termination of the program if, after further
evaluation, HUD determines that the program's current costs plus the
resources needed to improve it exceed its benefits.

More specifically, HUD stated that any conclusions about program costs
should be based on actual program performance rather than on "hypothetical
extrapolations." We designed our analysis to estimate the effect of the
Discount Sales Program while holding constant the effect of other factors,
such as neighborhood and property characteristics, that might also
influence the ratio of the net revenue HUD receives from selling a
property to the property's appraised value. To implement this approach, we
developed a statistical model using data from actual HUD property sales
and compared (1) the net revenue our model estimated HUD would have
received for each discounted property had the property been sold through
the regular process with (2) the net revenue our model estimated HUD would
have received by selling the property through the Discount Sales Program.
Each estimated value contained an error term that captured the effects of
omitted variables unavailable for the modeling process. As appendix II of
our draft report explained, comparing two estimated values removed the
influence of the omitted variables from our comparison, leaving the effect
of the program on HUD's net revenue. HUD said that its own preliminary
comparison of net revenues from program and nonprogram property sales had
shown that the loss in net revenue from the program was substantially less
than our estimate indicated, and that our report significantly overstated
the program's cost. Because HUD did not provide us details of its
analysis, we cannot determine why HUD's results differed from ours. We
continue to believe that isolating the effect of the program from other
influences, rather than making comparisons that do not control for these
factors, is the most appropriate way to estimate the impact of the
Discount Sales Program on HUD's net revenue.

HUD also said that our conclusion that most homebuyers did not benefit
financially from the program rested on the assumption that individual

homebuyers would have access to rehab financing on terms as favorable as
nonprofits and would have the skills to oversee the construction. HUD
disagreed with this assumption and said it believes that nonprofits have
both the ability to obtain financing that is unavailable to average
homebuyers and the capacity to oversee the rehab work at a cost that "may
be less than that" charged by profit-motivated firms. As a result, HUD
said that our report significantly understated the program's benefit to
homebuyers. Our draft report recognized the possibility that individual
homebuyers might not be able to rehab a home as cheaply as a nonprofit.
For this reason, we estimated the program's financial benefits under two
scenarios. The first assumed equal rehab costs for nonprofits and
homebuyers; the second assumed that homebuyers would pay 25 percent more
than a nonprofit for the same rehab work. Under both scenarios, our
estimates indicated that most homebuyers did not benefit financially from
the program. Our draft report also recognized that obtaining financing for
rehab work and overseeing this work are obstacles for some homebuyers, and
that these and other factors that are difficult to quantify may make the
program either more or less beneficial from a homebuyer's perspective.
However, it is important to note that homebuyers financing properties
purchased through the Discount Sales Program in effect pay financing costs
for the rehab work because the cost of this work is included in the
nonprofit's selling price (i.e., the homebuyer's purchase price) and
therefore is reflected in the homebuyer's mortgage costs. Furthermore,
because more than half of the properties we reviewed received less than
$15,000 in rehab work-including some that received none at all-it is
unlikely that the oversight costs for these projects would be very
substantial.

Finally, HUD said that the Discount Sales Program serves several public
policy purposes and was neither conceived as nor intended to be only a
source of revenue for FHA. HUD stated that the program contributes in a
"direct and positive manner" to the promotion of homeownership and the
revitalization of neighborhoods and that our report should acknowledge the
significance of these objectives. Our draft report did not indicate that
the program was intended to be solely a source of revenue. In addition,
our draft report recognized that the goals of the program include the
expansion of affordable housing opportunities and neighborhood
revitalization and included statements from HUD and nonprofit officials
about how the program may help to improve neighborhood conditions.
Furthermore, HUD's comments do not recognize the potential of the agency's
regular sales process to help achieve the same goals. Nevertheless, we
added language to the final report to reflect HUD's views about the
program's

potential benefits. As our report notes, HUD has not provided any analysis
to support its assertion that the program is contributing to the stated
policy objectives. Accordingly, we believe that HUD needs to undertake
such an analysis before making decisions about the program's future.

We are sending copies of this report to the appropriate congressional
committees, and it will be available at no charge on GAO's Web site at
http://www.gao.gov. If you or your staff have any questions about this
report, please call me at (202) 512-8678. Key contributors to this report
are
listed in appendix IV.

Sincerely yours,

David G. Wood
Director, Financial Markets and
Community Investment

Appendix I

                       Objectives, Scope, and Methodology

Our objectives were to assess (1) the cost of the Discount Sales Program
to HUD, (2) the benefits of the program to homebuyers, and (3) HUD's
efforts to monitor participating nonprofits and enforce program
requirements. Our work focused on the approximately 1,200 properties HUD
sold through the program in calendar year 2002 and the monitoring and
enforcement activities the four HOCs performed in connection with those
properties.

To determine the cost of the Discount Sales Program to HUD, we examined
the program's impact on HUD's net revenue from property sales and the cost
of administering the program. To determine the effect of the program on
HUD's net revenue, we performed a statistical analysis of data from HUD's
Single-Family Acquired Asset Management (SAMS) and the U.S. Census Bureau.
This analysis allowed us to estimate how much less net revenue HUD
received by selling properties through the program instead of its regular
sales process, while controlling for other factors. Appendix II provides
detailed information on our statistical model. Because HUD does not keep
records that would identify the costs to administer the program, we asked
HUD to estimate them. HUD developed its estimate by querying HOC and
Office of Housing officials about the number of staff years they allotted
to the program in calendar year 2002. HUD used this information and salary
and benefit data to derive an approximate personnel cost for the program.
HUD's estimate of administrative costs did not include comparatively minor
expenses, such as travel and mailing costs. We did not assess the
reliability of HUD's estimate.

To determine the benefits of the program to homebuyers, we examined the
program's potential financial benefits and the types of homeownership
services and assistance provided by participating nonprofit organizations.
To determine the extent to which homebuyers benefited financially from the
program, we performed a statistical analysis using data from HUD's SAMS,
the Bureau of the Census, and the four HOCs. Our analysis was limited to
238 discounted properties that HUD sold between February 1 and December
31, 2002,1 that nonprofits resold to homebuyers that same year. These were
the only properties for which the HOCs had the rehab and resale data
necessary for our analysis. This analysis allowed us to compare what
homebuyers actually paid for the discounted homes to our estimate of what
they would have spent had they purchased the homes under HUD's regular
sales process and paid for the rehab work themselves. Our analysis

1We used this time frame because HUD made significant changes to the
program that became effective on February 1, 2002.

Appendix I
Objectives, Scope, and Methodology

assumed that in the absence of the Discount Sales Program, a homebuyer
would be able to purchase the same home and rehabilitate it to the same
extent as the nonprofit. We also assumed that a homebuyer would inhabit
the home during the rehabilitation and, therefore, would not incur housing
expenses for two residences during that period. We performed the analysis
twice using different assumptions about rehab costs each time. The first
time we assumed that a homebuyer would incur the same rehab costs as a
nonprofit; the second time we assumed that homebuyers would incur 25
percent higher rehab costs than a nonprofit. Appendix II provides detailed
information on our statistical model. In assessing the program's financial
benefits, we also reviewed HUD's rules and instructions for both the
Discount Sales Program and the agency's regular sales process and
interviewed officials from HUD's Office of Housing. To determine the types
of services and assistance provided by participating nonprofits, we
visited 6 nonprofits that were actively involved in the program and
interviewed officials from these organizations. During our visits to the
four HOCs, we reviewed the affordable housing plans for a judgmental
sample of 17 nonprofits and interviewed the nonprofit coordinator at each
HOC. We did not evaluate the impact of the Discount Sales Program on
neighborhood conditions. However, we discussed this issue with HUD and
nonprofit officials and visited six properties that nonprofits had either
rehabbed or were in the process of rehabbing.

To assess HUD's efforts to monitor participating nonprofits, we reviewed
HUD's program guidance and instructions, SAMS data on the number of
discounted properties each nonprofit purchased in calendar year 2002,
nonprofits' annual reports on these properties, and the results of the
HOCs' desk and on-site reviews. We analyzed SAMS data, nonprofits' annual
reports, and desk reviews to determine the extent to which (1) nonprofits
submitted required monitoring information through their annual reports and
(2) the HOCs' monitoring efforts identified noncompliance with
requirements governing the resale price of discounted properties and the
income level of the homebuyers. From the desk review information, which
was current as of July 2003, we also determined the estimated excess
profits earned by the nonprofit organizations that submitted annual
reports. For each HOC, we used SAMS data and on-site review logs to
determine how many of the nonprofits that purchased discounted properties
were subject to an on-site visit. We also examined the results of these
reviews and discussed them with cognizant HOC officials to determine how
they were conducted and the types of problems they uncovered. Finally, we
interviewed HUD Office of Housing and HOC officials about factors that
hampered their ability to monitor nonprofits.

Appendix I
Objectives, Scope, and Methodology

To assess HUD's efforts to enforce program requirements, we reviewed the
agency's regulations and guidance to determine the major enforcement
actions available to the HOCs. We interviewed officials from HUD's Office
of Housing, Office of General Counsel, and the four HOCs about their
ability to take these actions. At each HOC, we collected information on
how frequently and in what situations they used these enforcement tools.
We also collected information on the amount of excess profits nonprofit
organizations used to pay down homebuyers' mortgages as of July 2003.
Finally, we obtained data from each HOC showing the calendar year 2003
discounted home purchases of nonprofits that sold properties for more than
program limits the previous year.

We tested the data we obtained from HUD's SAMS and the HOCs' desk reviews
for reasonableness and completeness and found them to be reliable for the
purpose of our analyses. In addition, we reviewed existing information
about data quality and controls supporting SAMS and discussed the data we
analyzed with agency officials to ensure that we interpreted them
correctly.

We conducted this review from December 2002 through November 2003. We
performed our work in accordance with generally accepted government
auditing standards.

Appendix II

Statistical Models Used to Estimate Program Costs and Benefits

Two objectives of the study were to determine (1) the cost of the Discount
Sales Program to HUD and (2) the benefits of the program to homebuyers.
For the cost objective, the scope of the study included properties that
HUD sold to nonprofit organizations in calendar year 2002. For the
benefits objective, the scope included the properties that HUD sold to
nonprofits between February 1 and December 31, 2002, that nonprofits
rehabilitated and resold to homebuyers during the same year. The empirical
analysis used to address these objectives was based upon two procedures.
The first procedure was the estimation of a model for which the dependent
variable is the fraction of the appraised value that HUD recovers for each
property after taking the agency's selling costs into account. Controlling
for a number of factors discussed below, the difference between the
estimated net revenue HUD would have received by selling the properties at
a discount versus an estimate of what HUD have would received if the
property had been sold under HUD's regular sales process represented the
cost of the Discount Sales Program to HUD. The second procedure was an
estimate of the financial benefits a homeowner received from purchasing a
rehabilitated Discounted Sales Program property from a nonprofit
organization rather than purchasing the property through HUD's regular
sales process and paying for the rehab work personally.

This appendix is organized in the following manner. First, there is a
brief discussion of the data. Second, there is an explanation of the
specification of the two econometric models that differ only in their
dependent variables, as mentioned above. Next, there is a discussion of
the estimation results of the two models. Finally, these results are used
to calculate estimates of the costs and benefits of the Discount Sales
Program.

Data	For our analysis, we obtained from HUD's Single-Family Acquired Asset
Management System (SAMS) computerized files for the 65,039 properties sold
by the agency during calendar year 2002. Each record provided financial
information, such as selling price, appraised value, and various
transactions costs. Each record also contained information on the
structural characteristics of the property, such as the number of bedrooms
and bathrooms. To describe the impact of neighborhood characteristics, we
obtained data at the census tract level on the percent of the population
living within an urbanized area, median household income in 1999, and
median real estate taxes. The source of these data was the Census 2000
Summary File 3 prepared by the U.S. Census Bureau.

                                  Appendix II
                  Statistical Models Used to Estimate Program
                               Costs and Benefits

A review of the data identified a number of outliers and missing values.
These observations were replaced using the means from the overall data set
minus these observations. In addition, 3,776 observations were lost during
the merger of the Census data because of the inability to identify the
census tract in which they were located. As a result, these observations
were not included in our analysis. In order to make the properties more
comparable, we also restricted the set of properties to only those
properties in census tracts in which at least one discount sale was made.
Finally, we excluded some observations with extreme values for the
dependent variable in our first model, which resulted in 5,189
observations being used in the regression analysis.

Specification of the Models	We developed two econometric models possessing
the same set of explanatory variables. We used the first model to estimate
the cost of the Discount Sales Program to HUD. In this model, the
dependent variable is HUD's net revenue from the sale of a property
divided by the property's appraised value. Net revenue equals HUD's
selling price minus property taxes paid by HUD, management and marketing
contractor costs, financing and closing costs (including discounts) paid
by HUD, the listing broker fee, and the cost of any HUD sales incentives.
We used the second model to estimate the benefits of the program to
homebuyers. In this model, the dependent variable is simply HUD's selling
price for a property divided by its appraised value. The explanatory
variables used in both models are of four types: dummy variables
identifying the homeownership center (HOC) that administered the sale;
dummy variables describing a combination of HOC and discount level, for
those properties sold through the Discount Sales Program; characteristics
of the property; and characteristics of the neighborhood. We chose the
final specification of the models to obtain a good fit while avoiding
problems with multicollinearity.

We chose the property and neighborhood characteristics based on a review
of the specifications employed in hedonic housing models.1 Such models
treat the housing market as an integrated series of submarkets for various
housing characteristics such as house size and neighborhood quality. A set
of housing and neighborhood characteristics can serve to describe a

1See, for example, Ann D. Witte, Howard J. Sumka, and Homer Erekson, "An
Estimation of a Structural Hedonic Price Model of the Housing Market: An
Application of Rosen's Theory of Implicit Markets," Econometrica, Vol. 47
(September 1979): 1151-1173.

Appendix II
Statistical Models Used to Estimate Program
Costs and Benefits

product like housing because individual housing units can be
differentiated from one another in many ways.

Selling price is the amount paid to HUD by either the nonprofit or the
private individual purchasing the property. Property taxes are the annual
property tax multiplied by the number of days the property was owned by
HUD divided by 365. Management and marketing contractor costs encompass
all fees and reimbursable costs HUD pays to these contractors. Financing
and closing costs paid by HUD include any discount granted to a nonprofit
organization. The selling fee is the commission that HUD paid to the
selling broker. This fee is generally zero for properties sold to
nonprofits. All HUD properties are placed with a listing broker, so this
generates a listing broker fee. Incentives are cash back allowances that
HUD periodically pays to homebuyers who close on an executed sales
contract relatively quickly-for example, within 30 to 60 days. Appraised
value is the amount at which the property was appraised before the sale.

The explanatory variables belong to one of four groups of variables: those
identifying the HOC that administered the sale; those describing a
combination of HOC and discount level, for those properties sold through
the Discount Sales Program; those describing characteristics of the house;
and those describing characteristics of the neighborhood. This
specification allows the net revenue as a fraction of the appraised value
for properties sold through the regular program to vary by HOC, and it
allows the effect of the Discount Sales Program on HUD's net revenue to
vary by HOC and by discount level. We defined a set of HOC dummy variables
that take on a value of 1 for all sales (discounted and not discounted)
administered by that HOC, with Atlanta as the omitted category, and 0
otherwise. We then defined a set of dummy variables that represented
combinations of HOC and discount program level. For example, there is a
dummy variable that takes on a value of 1 for all sales with a 10 percent
discount made by the Atlanta HOC and 0 otherwise, etc. In this way, the
coefficient on that variable represents the difference in the fraction of
the appraised value obtained on property sales with a 10 percent discount
and made by the Atlanta HOC, compared with the fraction obtained by the
Atlanta HOC on sales made through the regular program. Similarly, the
coefficient on the dummy variable that takes on a value of 1 for all sales
with a discount level of 30 percent made by the Santa Ana HOC represents
the difference in the fraction of the appraised value obtained on property
sales made at a 30 percent discount and made by the Santa Ana HOC,

Appendix II
Statistical Models Used to Estimate Program
Costs and Benefits

compared with the fraction obtained by the Santa Ana HOC on sales made
through the regular program.2

There are three variables that provide characteristics for each property.
These variables measure the property's number of bedrooms, bathrooms, and
stories. Although the likely association between those variables and HUD's
net revenue from a property sale relative to the property's appraised
value is not clear-cut, we anticipated that there may be a positive
association for the number of bedrooms and bathrooms. HUD's costs
associated with property management and sale include a fixed cost
component not related to property value as well as a component that varies
with selling price. Accordingly, if properties with more bedrooms and
bathrooms tend to be valued more, increasing both their selling prices and
appraised values, then HUD's net revenue relative to the appraised value
will be higher for higher-valued properties because HUD's fixed costs will
be lower relative to the appraised value. Because the relationship between
number of stories and market value is less clear, we had no clear
expectation for the sign of the coefficient for that variable. Because any
effect of these variables on market value is likely to similarly affect
both selling price and appraised value, for the equation with selling
price divided by appraised value as the dependent variable we had no clear
expectation of the signs of the coefficients of these variables. However,
we included them to keep our equations consistent and because these
variables may exhibit statistically significant effects if HUD is able to
obtain more accurate appraisals for certain types of properties.

Finally, there are four variables for neighborhood characteristics: the
percent of the population in the census tract living within an urbanized
area, median household income in 1999, median real estate taxes, and
whether the property is located in a revitalization area. As for the
property characteristics, we anticipated that to the extent that these
neighborhood characteristics are associated with higher (lower) property
values, they would be positively (negatively) associated with HUD's net
revenue from a property sale relative to that property's appraised value
because for higher(lower-) valued properties HUD's fixed costs will be
lower (higher) relative to appraised values. Accordingly, we anticipated
positive coefficients in the equation for which the dependent variable is
HUD's net revenue relative to

2There were no discounted sales at the 15 percent level made by the
Atlanta HOC, and as a result the dummy variable combining 15 percent
discount with the Atlanta HOC is also omitted.

Appendix II
Statistical Models Used to Estimate Program
Costs and Benefits

appraised value for the percentage of the population living within an
urbanized area, median income, and median real estate because all of these
variables are likely to be positively associated with property value. In
contrast, we anticipated a negative coefficient for location in a
revitalization area because that variable may indicate lower property
values that would raise the ratio of HUD's fixed costs to a property's
appraised value and because HUD may also incur additional costs to
maintain properties in those areas. Again, similar to the property
characteristics, we had no clear expectations for the signs of the
neighborhood characteristics in the equation in which the dependent
variable is HUD's selling price relative to the appraised value because
any effect of these variables on property value is likely to similarly
affect selling price and appraised value. But, for the reasons cited
above, we included them in the equation.

Table 7 presents the variable names and descriptions, along with mean
values of the explanatory variables. In a previous version of the model
that we estimated using a data set representing all usable observations of
HUD property sales, rather than the data set containing only sales made in
census tracts in which discount sales were made, we included as variables
the year in which the structure was built and the average age of homes in
the census tract. These variables were significant at that time, but they
became insignificant in the final model when we limited our data set, so
we dropped them from the analysis. Although the rate of change in house
prices is another variable that may have some predictive value, we did not
include it in our models because information on the rate of house price
change was not available at the census tract level. Also, according to HUD
officials, the average time between the appraisal and the sale of a
property is about 3 months, so in most cases we did not anticipate that
there would be much price appreciation or depreciation resulting from
market conditions.

                                  Appendix II
                  Statistical Models Used to Estimate Program
                               Costs and Benefits

 Table 7: Variable Names, Descriptions, and Mean Values Variable name Variable
                   description Mean value HOC dummy variables

Santa Ana HOC 1 if property was sold with or without a discount by the Santa Ana
                               HOC, else 0 0.4039

Denver HOC 1 if property was sold with or without a discount by the Denver HOC,
                                 else 0 0.1243

Philadelphia HOC 1 if property was sold with or without a discount by the
                        Philadelphia HOC, else 0 0.2249

                Discount Sales Program category dummy variables

    Santa Ana, 10   1 if property sold at 10 percent discount level by 0.0412 
                                the Santa Ana HOC, else 0              
    Santa Ana, 15   1 if property sold at 15 percent discount level by 0.0019 
                                the Santa Ana HOC, else 0              
    Santa Ana, 30   1 if property sold at 30 percent discount level by 0.0250 
                                the Santa Ana HOC, else 0              
      Denver, 10    1 if property sold at 10 percent discount level by 0.0191 
                                  the Denver HOC, else 0               
      Denver, 15    1 if property sold at 15 percent discount level by 0.0120 
                                  the Denver HOC, else 0               
      Denver, 30    1 if property sold at 30 percent discount level by 0.0041 
                                  the Denver HOC, else 0               
Philadelphia, 10 1 if property sold at 10 percent discount level by 0.0397 
                               the Philadelphia HOC, else 0            
Philadelphia, 15 1 if property sold at 15 percent discount level by 0.0058 
                               the Philadelphia HOC, else 0            
Philadelphia, 30 1 if property sold at 30 percent discount level by 0.0224 
                               the Philadelphia HOC, else 0            
     Atlanta, 10    1 if property sold at 10 percent discount level by 0.0399 
                                 the Atlanta HOC, else 0               
                    1 if property sold at 30 percent discount level by 0.0150 
     Atlanta, 30                 the Atlanta HOC, else 0               

Property characteristics

Bedrooms The number of bedrooms in the structure 3.1134 Bathrooms The
number of bathrooms in the structure 1.8071 Stories The number of stories
in the structure 1.3472

Neighborhood characteristics

          Urban area        Percent of population in the census tract 92.9976 
                                 living within an urbanized area      
        Median income        Median household income in census tract  38.6993 
                                             (1999)                   
Median real estate taxes Median real estate taxes in census tract   1.2862 
                                  1 if property is located in a               
     Revitalization area    HUD-designated revitalization area, else   0.2631
                                               0.                     
         Source: GAO.                                                 

Estimation Results	We estimated the two models using ordinary least
squares due to its ease of calculation and interpretation. Table 8
presents the estimated coefficients, their standard errors, and the
summary statistics.

In general, both models were consistent with our expectations. In the
model for net revenue relative to appraised value, the coefficients on the
Discount Sales Program category showed the expected pattern. That is, for
each HOC, coefficients became more negative at higher discount levels.

                                  Appendix II
                  Statistical Models Used to Estimate Program
                               Costs and Benefits

In both models the coefficients for the characteristics for the property
possessed the same sign. In general, a higher fraction of the appraised
value was obtained from larger homes, as measured by more bedrooms and
bathrooms, as well as homes with fewer floors. The fraction of the
appraised value obtained by HUD tended to be higher for properties located
in more urban areas and where median real estate taxes were higher. In the
model for selling price relative to appraised value, the coefficient for
median household income was negative and statistically significant at less
than the 0.01 percent level. However, in the model for net revenue
relative to appraised value, the coefficient on this variable was
statistically insignificant.

Table 8: Coefficients from Estimated Models Explanatory variable name Net
 revenue relative to appraised value Selling price relative to appraised value

                   Intercept 0.7975 0.9085 (0.0139) (0.0136)

      HOC dummy variables Discount Sales Program category dummy variables

           Santa Ana HOC                 0.0461                        0.0231 
                                        (0.0056)                     (0.0054) 
             Denver HOC                  -0.0444                      -0.0520 
                                        (0.0079)                     (0.0077) 
          Philadelphia HOC               -0.0026                       0.0174 
                                        (0.0074)                     (0.0072) 

           Santa Ana, 10                 -0.1448                      -0.0874 
                                        (0.0100)                     (0.0098) 
           Santa Ana, 15                 -0.1959                      -0.1064 
                                        (0.0437)                     (0.0427) 
           Santa Ana, 30                 -0.2726                      -0.0506 
                                        (0.0131)                     (0.0128) 
             Denver, 10                  -0.0622                      -0.0088 
                                        (0.0152)                     (0.0149) 
             Denver, 15                  -0.1147                      -0.0385 
                                        (0.0186)                     (0.0182) 
             Denver, 30                  -0.1744                       0.0525 
                                        (0.0309)                     (0.0303) 
          Philadelphia, 10               -0.1053                      -0.0785 

                                  Appendix II
                  Statistical Models Used to Estimate Program
                               Costs and Benefits

(Continued From Previous Page)

Explanatory variable name Net revenue relative to appraised value Selling
price relative to appraised value

                               (0.0109) (0.0106)

          Philadelphia, 15               -0.1267                      -0.0903 
                                        (0.0258)                     (0.0252) 
          Philadelphia, 30               -0.2718                      -0.0611 
                                        (0.0140)                     (0.0137) 
            Atlanta, 10                  -0.0851                      -0.0536 
                                        (0.0106)                     (0.0103) 
            Atlanta, 30                  -0.2595                      -0.0173 
                                         (0.167)                      (0.164) 

             Property characteristics Neighborhood characteristics

           Bedrooms                            0.0052                  0.0063 
                                     (0.0022)                        (0.0022) 
           Bathrooms                           0.0155                  0.0083 
                                     (0.0034)                        (0.0033) 
            Stories                  -0.0280                          -0.0081 
                                     (0.0054)                        (0.0053) 

               Urban area                    0.0007                    0.0007 
                                            (0.00009)                (0.0001) 
              Median income                  0.0003                   -0.0006 
                                            (0.0002)                 (0.0002) 
        Median real estate taxes             0.0177                    0.0254 
                                            (0.0032)                 (0.0031) 
           Revitalization area               -0.0244                   -.0091 
                                            (0.0054)                 (0.0053) 
           Summary statistics                              
                   R2                         0.28         
         Number of Observations               5,189                     5,189 
              Source: GAO.                                 

                   Note: Standard errors are in parentheses.

                                 Program Costs

We used the estimated coefficients for the Discount Sales Program
variables from the model for net revenue relative to appraised value to
estimate the cost of the program to HUD. We made this estimate by
comparing the (1) net revenue that our model predicted HUD would have
received for each discounted property had it been sold through the regular
sales process with (2) net revenue our model predicted HUD would have

                                  Appendix II
                  Statistical Models Used to Estimate Program
                               Costs and Benefits

received by selling it through the Discount Sales Program. Even though we
knew the actual amount HUD received by selling the property through the
Discount Sales Program, we compared two estimated values in order to be
consistent. Each estimated value contained an error term that captured the
effects of omitted variables unavailable for the modeling process. By
comparing two estimated values, we removed the influence of the omitted
variables from our comparison, leaving the effect of the Discounted Sales
Program.

This analysis was conducted using 1,194 discounted properties that were
sold during calendar year 2002. The values of the Discount Sales Program
dummy variables reflect the estimate of the difference in the fraction of
appraised value obtained by HUD for each combination of discount level and
HOC. That is, for a given property's program characteristics-discount
level and HOC-these estimates represent the difference in value HUD
obtains compared to the case in which that property had been sold through
the regular program. Because we included HOC dummies to capture otherwise
unmeasured factors that might vary by region, the program characteristic
dummy variables describe incremental revenue differences associated with
each discount level as compared to the level of nondiscount sales in each
HOC. For example, if a property was sold by the Philadelphia HOC at a 15
percent discount, the estimated difference in the net revenue value would
be about -13 percent of appraised value. The relevant coefficient in table
8 is -0.1267. Similarly, if a property was sold by the Santa Ana HOC at a
15 percent discount, the estimated difference in the net revenue value
would be about 20 percent of appraised value. The relevant coefficient in
table 8 is -0.1959.

We also estimated a range of costs using confidence intervals around the
estimates for the sales category dummy variables. We did this by
multiplying each coefficient estimate by 1.645 times its associated
standard error. We then added and subtracted this amount from the
coefficient estimate to create a 90 percent confidence interval around
each estimate. We then recalculated program costs assuming that the
estimates associated with the lower and upper bounds of the confidence
interval reflected the difference between the value HUD obtains through
the Discount Sales Program as compared to the value obtained through the
regular sales process.

Benefits to Homebuyers	In the model for selling price relative to
appraised value, the benefit to the homebuyer of purchasing a property
from a nonprofit organization was

Appendix II
Statistical Models Used to Estimate Program
Costs and Benefits

taken to be the estimated price a homeowner would have paid to buy the
house through HUD's regular sales process, minus the estimated financing
and closing costs that HUD would have paid in that situation, plus the
rehabilitation costs, minus the actual sales price paid to the nonprofit.
This analysis was conducted using 238 observations for discounted
properties purchased by nonprofits between February 1 and December 31,
2002, and resold to homebuyers that same year. These were the only
properties for which data on the nonprofits' rehabilitation costs and the
selling price to the final homeowners were available. We performed the
analysis twice using different scenarios. In the first scenario, we
assumed that the rehabilitation costs incurred by a homebuyer operating
outside of the Discount Sales Program would be the same as the
rehabilitation costs actually incurred by the nonprofit. In the second
scenario, we assumed that a homebuyer's rehabilitation costs would be 25
percent higher than those incurred by the nonprofit. Under both scenarios,
we assumed that the financing and closing costs HUD would pay for under
its regular sales process were equal to 3 percent of the estimated selling
price.

Because these properties were originally bought from HUD by a nonprofit
organization, we needed to estimate the prices homebuyers would have paid
to purchase the properties directly from HUD through the regular sales
process to calculate the difference in price paid by homeowners purchasing
a property after it had been rehabilitated under the Discount Sales
Program. In a manner parallel to our estimate of net revenue described
earlier, we used the values of the estimated Discount Sales Program dummy
variables from our second model to make this calculation.

Appendix III

Comments From the Department of Housing and Urban Development

Appendix III
Comments From the Department of Housing
and Urban Development

Appendix IV

                     GAO Contacts and Staff Acknowledgments

GAO Contacts	David G. Wood, (202) 512-8678 Steven K. Westley, (202)
512-8678

Acknowledgments	In addition to those named above, Kimberly Berry, Gwenetta
Blackwell, Stephen Brown, Emily Chalmers, Rudy Chatlos, Jay Cherlow, David
Dornisch, John McGrail, John Mingus, Mark Molino, David Pittman, Steve
Ruszczyk, Stewart Seman, and Mark Stover made key contributions to this
report.

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

Obtaining Copies of GAO Reports and Testimony

The fastest and easiest way to obtain copies of GAO documents at no cost
is through the Internet. GAO's Web site (www.gao.gov) contains abstracts
and fulltext files of current reports and testimony and an expanding
archive of older products. The Web site features a search engine to help
you locate documents using key words and phrases. You can print these
documents in their entirety, including charts and other graphics.

Each day, GAO issues a list of newly released reports, testimony, and
correspondence. GAO posts this list, known as "Today's Reports," on its
Web site daily. The list contains links to the full-text document files.
To have GAO e-mail this list to you every afternoon, go to www.gao.gov and
select "Subscribe to e-mail alerts" under the "Order GAO Products"
heading.

Order by Mail or Phone	The first copy of each printed report is free.
Additional copies are $2 each. A check or money order should be made out
to the Superintendent of Documents. GAO also accepts VISA and Mastercard.
Orders for 100 or more copies mailed to a single address are discounted 25
percent. Orders should be sent to:

U.S. General Accounting Office 441 G Street NW, Room LM Washington, D.C.
20548

To order by Phone: 	Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202)
512-6061

To Report Fraud, 	Contact: Web site: www.gao.gov/fraudnet/fraudnet.htm

Waste, and Abuse in E-mail: [email protected]

Federal Programs Automated answering system: (800) 424-5454 or (202)
512-7470

Public Affairs	Jeff Nelligan, Managing Director, [email protected] (202)
512-4800 U.S. General Accounting Office, 441 G Street NW, Room 7149
Washington, D.C. 20548

                               Presorted Standard
                              Postage & Fees Paid
                                      GAO
                                Permit No. GI00

United States
General Accounting Office
Washington, D.C. 20548-0001

Official Business
Penalty for Private Use $300

Address Service Requested
*** End of document. ***