subject to portfoUo reengineering.
10
24
important to note that the study's results do not reflect the changes that HUD made to its
proposal in early 1996.
Study Confirms Excess Subsidy Costs
and Significant Physical Needs at Properties
Ernst & Young estimates that the majority of the properties have assisted rents exceeding
market rents and that the properties have significant amounts of immediate deferred
maintenance and short-term and long-term capital needs.^ Specifically, Ernst & Young's
study estimates that a mzgority of the properties-between 60 and 66 percent-have rents
above market zind between 34 and 40 percent are estimated to have below-market rents.
Ernst & Young's data also indicate a widespread need for capital-between $9.2 bilhon and
$10.2 billion-to address current deferred maintenance needs and the short- and long-term
requirements to maintain the properties. The study estimates that the properties have
between $1.3 billion and $1.6 billion in replacement and cash reserves that could be used
to address these capital needs, resulting in total net capital needs of between $7.7 billion
and $8.7 billion. The average per-unit cost of the total capital requirements, less the
reserves, is estimated to be between $9,116 and $10,366.
Study Indicates a Significant Level
of Debt Restructuring Would Be Needed
*rhe study defines capital needs as the cost of improvements needed to bring properties
into adequate physical condition to attract uninsured market rate financing. Three
categories of capital needs are defined: (1) immediate deferred maintenance, the
estimated costs to bring all property operating systems up to market conditions and
lender underwriting standards; (2) the short-term capital backlog, the estimated costs for
the expired life of property systems requiring replacement in 5 or fewer years; and (3) the
long-term capital backlog, the estimated costs for the expired life of property systems
requiring replacement in more than 5 years.
11
25
Ernst & Young's analysis also indicates that about 80 percent of the properties would not
be able to continue operations unless their debt was restructured. Furthermore, for
approximately 22 to 29 percent of the portfolio, writing the existing debt to zero would
not sufficiently reduce costs for the properties to address their immediate deferred
maintenance and short-term capital needs. The study estimates that between 11 and 15
percent of the portfolio would not even be able to cover operating expenses.
The study was designed to use the information on market rents and the properties'
physical condition gathered by Ernst & Young, as well as fmaincial and Section 8
assistance data from HUD's data systems, in a financial model designed to predict the
proposal's effects on the portfolio as a whole. Specifically, the model estimates the
properties' future cash flows over a 10-year period on the basis of the assumption that
they would be reengineered (marked to market) when their current Section 8 contracts:
expire.'"
The model classifies the loans into four categories-performing, restructure, full write-off,
and nonperforming-that reflect how the properties would be affected by HUD's proposal.
Placement in one of the four categories is based on the extent to which income from the
reengineered properties would be able to cover operating costs, debt service payments,
deferred maintenance costs, and short-term capital expenses. Table 1 shows the results
of Ernst & Young's analysis of how properties would be affected by HUD's proposal.
'"For properties with more than one Section 8 contract, the model assumes that the
property would be reengineered when the contract with the earliest expiration date
expires.
12
26
Table 1 : Effects of Reengineering on HUD's Insured Section 8 Portfolio
Costs covered with
Status of loan after
reengineered cash
reengineering
Percent of portfolio
flows
Performing
17 to 23
Existing debt, operating
expenses, all capital
needs
Restructure
50 to 58
Restructured debt,
operating expenses, all
capital needs
Full write-off
1 1 to 1 5
Operating expenses and
some capital needs but
no debt
Nonperforming
11 to 1 5
Some operating
expenses but no debt or
capital needs
Note: Capital needs represent immediate deferred maintenance and short-term (5 years or less)
capital needs. In addition, the financial model categorizing the loans assumes annual deposits to
replacement reserves.
The case study properties we examined reflect the full range of possible outcomes. Ernst
& Young's model concluded that two of the case study properties would fall into the
performing category, two were classified in the restructure category, three were full write-
offs, and two were classified as nonperforming."
"One of our 10 case study properties was dropped firom Ernst & Yoimg's sample prior to
their final analysis.
13
27
MODEL RESULTS INDICATE EVENTUAL
SECTION 8 SAVINGS BUT HIGH CLAIMS COSTS
Ernst & Young's model provides estimates of Section 8 subsidy costs before and after
reengineering and the amount of the FHA insurance claims resulting from writing down
the properties' mortgages and addressing deferred maintenance needs, although in a
manner that does not conform with budget rules or scoring methodology. The May 2,
1996, briefing report does not present the information gathered in the study on how
portfolio reengineering would affect rental subsidy and claims costs. According to the
Deputy Assistant Secretary for Operations, HUD Office of Housing, while the Department
plans to use Ernst & Young's cost data in developing future budget estimates relating to
portfoho reengineering, it never intended that the cost data be included in Ernst &
Young's May 1996 report.
In the model, claims costs include the amount of debt reduction needed for the property
to sustain operations at market rents and also include funding for some or all of the
immediate deferred maintenance and short-term capital needs. However, the claiins costs
are limited to a maximum of the unpaid principal balance of the loans at the time of debt
restructuring. In addition, the claims costs are based on an evaluation of the loan amount
the property could support using standard financial underwriting criteria without the
continuation of FHA insurance.
Our analysis of these data indicates that, while Section 8 costs would decrease over the
long-run, there may be little or no aggregate savings in Section 8 rental assistance costs
over the next 10 years if, as the model assumes, all insured Section 8 properties were
reengineered when their current Section 8 contracts expire. These data indicate that, for
the period fiscal yeju- 1996 through 2005, there may be little difference in aggregate
Section 8 costs after reengineering compared with the cost of continuiitg project-based
assistance at current levels:
14
28
- If project-based assistance is continued at current levels (including inflation), the costs
in present value terms are estimated to be between $27.2 billion and $31.0 billion.'^
- The cost of Section 8 assistance after reengineering is estimated to be between $26.5
bUUon and $29.8 billion."
A primary reason that 10-year Section 8 cost estimates are similar is that the model
assumes that projects will be reengineered when their current Section 8 contracts expire.
This analysis thus reflects HUD's contractual obligations, which the Department has
repeatedly indicated that it will not abrogate. Because memy properties with rents below
market have expiring contracts during the first part of the 10-year period and thus will be
reengineered early in the process, Section 8 costs increase during the early years but then
begin to decrease as more projects with rents above market are reengineered in the later
years. In fiscal year 2005, when Ernst & Young assumes that virtually all projects have
been reengineered. Section 8 costs are estimated to be between $1.9 bilUon and $2.2
billion a year on a present value basis. The model indicates that annual savings of
between $298 million to $493 million (between 13 to 19 percent) could be achieved with
reengineering compared with the costs of continuing Section 8 assistance at current
levels.
'^These and other total cost estimates contained in our statement are based on a universe
of 8,363 properties-the population fi-om which the sample used by Ernst & Young was
selected. The estimates contained in Ernst & Young's May 1996 report are based on a
population of 8,563 properties. The difference reflects properties that did not have a
chance at being included in the sample due to technical and cost considerations. In
general, the estimates in our report would increase by about 2 percent if they were
applied to 8,563 rather than 8,363 properties. This assiunes that the additional properties
HUD identified are similar to those in the original population.
"Both estimates assim\e Section 8 costs increase by 3 percent a year. We discounted
these costs by 6.75 percent a year to arrive at a present value estimate.
15
29
However, we note that Ernst & Young's model does not reflect the changes that HUD
made to its proposal in early 1996. Some of the changes offer the potential of additional
Section 8 cost savings. For example, HUD is proposing to use a proactive approach to
portfolio reengineering and hopes that owners will voluntarily agree to go through this
process (and terminate the Section 8 contracts) in advance of Section 8 contract
expirations. However, it is not clear to what extent HUD will be successful in attracting
owners to restructure in advance of the Section 8 contract expirations-or what additional
incentives HUD may have to offer to achieve this goal.
In addition, HUD now plans to focus initially on reengineering properties with rents above
market To the extent that portfolio reengineering focuses on such properties. Section 8
savings would increase. For example, Ernst & Young's data indicate that the lO-year
Section 8 costs for properties with assisted rents above market would be between $21.2
billion and $25.0 billion compared with between about $18.5 billion and $21.5 billion if the
loans for properties with rents above market were restructured when the Section 8
contracts expire.''' Furthermore, it is important to note that some savings would result if,
as Ernst & Young's model assumes, mortgage interest subsidies are terminated as projects
are reengineered. Ernst & Young estimates that without reengineering, mortgage interest
subsidies would range from between about $841 million to $1.1 billion over the next 10
years (on a present value basis). However, it should be noted that most properties that
receive interest subsidies are believed to have rents that are below market.
Regarding the FHA insurance claims costs, our analysis of Ernst & Young's data indicates
that FHA clainis for mortgage write-downs and deferred maintenance and other capital
"All estimates for projects with assisted rents above or below market rents based on the
Ernst & Young sample may be misstated because the sample did not contain both types
of properties in each of the groups of properties, called strata, from which they sampled.
Thus, the estimates assume that none of the 510 projects from three strata containing
newer projects had assisted rents below market. The estimates also assume that none of
the 372 older projects from two strata were above market
16
30
costs for properties that need mortgage restructuring will be substantial. The mortgage
balances for properties needing mortgage restructuring-including those in the full write-
off and nonperforming categories that would have their mortgages totally written off-
would need to be reduced by between 61 to 67 percent. This reduction would result in
claims costs against the FHA multifamily insurance funds. According to the data, the 10-
year costs of claims paid, on a present value basis, would be between $6 bilUon and $7
billion. If HUD's proactive approach were successful, while Section 8 savings would
increase, the claims amounts related to debt write-down could also be higher than
indicated in Elrnst & Young's study because (1) the loans would be restructured earlier
when the unpziid principal balances are higher and (2) the present value of claims
occurring in the earlier years would be higher. However, HUD believes that without a
proactive approach, owners will disinvest in the properties. Such disinvestment would
have an adverse impact on the physical condition, resulting in higher claims costs for
deferred maintenance.
The claims payments estimated in Ernst & Young's study indicate substantial loan loss
rates for the goverrunent.'^ For example, portfolio reengineering clainis for properties
with assisted rents greater than market rents are estimated to be between $4.8 billion and
$5.8 billion, and the related unpaid principal balances at the time of restructuring are
between $6.9 billion and $8.1 billion. The estimated loss rate would be between 67 and 75
percent Table 2 provides claims, unpaid principal balance, and recovery data for the
properties subject to portfolio reengineering.
"The loss rate represents the ratio of claims to the unpaid principal balances at
restructure dates.
17
31
Table 2: Impact of Portfolio Reengineering on FHA's Insurance Fund. Fiscal Years 1996-2005 ^
Dollars in billions (present value)
Relative value of
assisted rents
before restructuring
Claims
Unpaid principal
balances at date of
restructuring
Loss rate
Greater than or equal
to market rents
Between $4.8 and
$5.8
Between $6.9 and
$8.1
67% to 75%
Less than market
rents"
Between $1.0 and
$1.5
Between $2.2 and
$3.1
40% to 51%
Total
Between $6.0 and
$7.0
Between $9.5 and
$10.8
61 % to 67%
'All estimates for projects with assisted rents above or below market rents based on the Ernst &
Young sample may be misstated because the sample did not contain both types of properties in
each of the groups of properties, called strata, from which they sampled. Thus, the estimates
assume that none of the 510 projects from three strata containing newer projects had assisted
rents below market. The estimates also assume that none of the 372 older projects from two
strata were above market.
This estimate may be misstated because no projects with claims were found among the sampled
projects with assisted rent less than market rent from four strata. Thus, the estimate assumes
that none of the 985 projects from these strata were projects with assisted rents less than market
rents that resulted in claims. The 985 projects included 807 newer and 178 older projects.
GAP'S ASSESSMENT OF THE
MODEL AND THE RESULTS
We are currently evaluating Ernst & Yoimg's financial model and expect to issue our
report late this summer. Oui preliminary assessment is that the model provides a
reasonable framework for studying the overall results of portfolio reengineering, such as
the number of properties that will need to have their debt restructured and to estimate
18
32
the related costs of insurjince claims. In addition, we did not identify any substantive
problems with Ernst & Young's sampling and statistical methodology. However, our
preliminary assessment of the study indicates that some aspects of Ernst & Young's
fmancial model and its assumptions may not reflect the way in which insured Section 8
properties will actually be affected by portfolio reengineering. Also, some of the
assumptions used in the model may not be apparent to readers of Ernst & Young's May
1996 briefing report.
For example, Ernst & Young's assumptions about the transition period that properties go
through in the reengineering process may be overly optimistic. During the transition, a
reengineered property changes fi-om a property with rental subsidies linked to its units to
an unsubsidized property competing in the marketplace for residents. The model
estimates that the entire transition will be completed within a year after the first Section
8 contract expires. In addition, the model assumes that during this year, the property's
rental income will move incrementally toward stabilization over 9 months. The lenders
with whom we consulted on the reasonableness of the model's major assumptions
generally believed that a longer transition period of 1 to 2 years is more likely. They also
anticipated an unstable period with less income and more costs during the transition
rather than the smooth transition assumed in the model. An Ernst & Young official told
us that the 9-month period was designed to reflect an average transition period for
reengineered properties. While he recognized that some properties would have longer
transition periods than assumed in the model, he believed that the transition periods for
other properties could be shorter than 9 months.
In addition, Elmst & Young's May 1996 report does not detail all of the assumptions used
in the firm's financial model that are useful to understanding the study's results. In
particular, the model assumes that the interest subsidies that some properties currently
receive will be discontinued after the first Section 8 contract expires, including those in
the performing category whose debts do not require restructuring. Furthermore, the
financial test identifying the loans that could cover all debt service, operating, and capital
19
33
needs costs if market rents replaced assisted rents, assumes that properties that currently
receive interest subsidies would have to pay the full mortgage interest amount without
the benefit of the interest subsidy. This assumption would identify fewer performing
loans than if the current debt service requirements were tested for. We are currently
examining how the assumptions contained in Ernst & Young's study affect its estimates of
the effects of portfolio reengineering. In addition, as discussed in appendix II, we also
performed sensitivity analyses to assess the extent to which the use of different
assumptions affects the results of Ernst & Young's study.
As part of our work evaluating the Ernst & Young model, we also compared Ernst &
Young's data on market rents and deferred maintenance with information from the three
licensed appraisal firms we retained to assess 10 of the HUD-insured Section 8 properties
included in Ernst & Young's sample. In 8 out of the 10 cases, the estimated market rents
provided by these appraisers are reasonably close to (i.e., within 10 percent of) the rents
Ernst & Young developed in their market surveys. In two instances, however, Ernst &
Young's market rent estimates are more than 20 percent lower than the market rent
estimates of the appraisal firms. This difference reflects in large measure a different
methodology that EIrnst & Young used in estimating market rents in neighborhoods
consisting primarily of assisted properties-where few, if any, comparable market
properties were identified. In these cases, Ernst & Young assumed that since the local
neighborhood was essentially maintained by non-market driven forces, there was no
market for unassisted rents in these neighborhoods other than that controlled by the local
housing authority. Thus, EIrnst & Young based its rent estimates on the rents subsidized
by the local housing authorities. In contrast, the appraisers we retained believed that
there were comparable properties that could be used to estimate market rents for the two
properties.
While the information we had on the market rents for our 10 case studies was generally
consistent with Elmst & Yoimg's estimates, the information on capital needs costs varied
widely. In general, the E^st & Young cost estimates were significantly higher. The
20
34
appraisers we retained conducted physical inspections of the properties but were not
tasked with performing engineering studies. In contrast, Ernst & Young retained a firm to
conduct engineering studies at the properties. Officials from Ernst & Young and the
engineering firm said the inspections were not full engineering studies which would be
used in financial underwriting or negotiations with the owners. However, the inspections
provide preliminary data that can be used for budgeting piuT)oses. Because the
appraisals conducted for us were more limited in scope than Ernst & Young's reviews and
thus not directly comparable, we provided the property owners and managers with the
capital needs estimates developed by Ernst & Young and by the appraisers for their
evaluation. We are currently exaimining their responses and the reasons for the
differences in the capital needs estimates.
COMPLEX ISSUES WILL SHAPE THE
POTENTIAL OUTCOMES OF REENGINEERING i
The Congress faces a number of significant and complex issues in evaluating HUD's
portfolio reengineering proposal. How these issues are resolved will, to a large degree,
determine the extent to which the problems that have long plagued the portfolio are
corrected and prevented from recurring and the extent to which the restructiuing process
results in any net savings to the goverrunent. Key issues include the following:
- How should HUD's problems in managing the insured Section 8 portfolio be
addressed?
- To what extent should FHA insurance be provided for restructured loans?
- Should rental assistance after reengineering be project-based or tenant-based?
- What protection should be given to residents of reengineered properties to protect
them from rent increases or displacement?
21
35
- To what extent should properties with assisted rents that are below market rents be
included in portfolio reengineering?
- What process or processes should be used to restructure mortgages?
- To what extent should the federal government finance rehabilitation costs?
- What actions should be taken to deal with properties that would have difficulty in
sustaining operations after portfolio reengineering?
- Whether and to what extent should tax relief be provided as a part of the
reengineering process?
- Will the current demonstration program be sufficient to test the range of options for
carrying out portfolio reengineering and its effects on properties and residents?
These issues are discussed further in appendix III.
OBSERVATIONS
HUD's portfolio reengineering initiative recognizes a reality that has existed for some
time-namely, that the value of many of the properties in the insured Section 8 portfolio is
far less than the mortgages on the properties suggest. Until now, this reality has not been
recogruzed and the federal government has continued to subsidize the rents at many
properties above the level that the properties could conunand in the commercial real
estate market.
As the Congress evaluates options for addressing this situation, it will be important to
consider each of the fundamental problems that have affected the portfolio, and their
imderlying causes. Any approach implemented should address not only the high Section 8
22
36
subsidy costs, but also the high exposure to insurance loss, poor physical condition, and
the underlying causes of these long-standing problems with the portfolio. As illustrated
by several of the key issues discussed above, questions about the specific details of the
reengineering process, such as which properties to include and whether or not to provide
FHA insurance, will require weighing the likely effects of various options and the trade-
offs involved when proposed solutions achieve progress on one problem at the expense of
another. Changes to the insiured Section 8 portfoUo should also be considered in the
context of a long-range vision for the Federal government's role in providing housing
assistance, and assistance in general, to low-income individuals, and how much of a role
the government is realistically able to have, given the current budgetary climate.
Addressing the problems of the portfoUo will inevitably be a costly and difficult process,
regardless of the specific approaches implemented. The overarching objective should be
to implement the process in the most efficient and cost-effective manner possible,
recognizing not only the interests of the parties directly affected by restructuring but also
the impact on the federal government and the American taxpayer.
As indicated earlier in our statement, we are continuing to review the results of Ernst &