Potential Losses From the Rental Housing Inventory: Soundness of Current Estimates

Published by the Government Accountability Office on 1990-02-28.

Below is a raw (and likely hideous) rendition of the original report. (PDF)

.                       United States Genera3Accounting Office   /qa730
:-GAO                   Testimony


    For Release         Potential Losses from the Rental Housing
    on Delivery         Inventory:  Soundness of Current Estimates
    Expected at
    9:30 a.m.
    February 28, 1990

                        Statement of
                        Eleanor Chelimsky
                        Assistant Comptroller General
                        Program Evaluation and
                          Methodology Division
                        Before the
                        Subconnnittee on Housing
                          and Community Developent
                        Committee on Banking, Finance
                         and U&an Affairs
                        House of Representatives

Mr. Chairman      and Members of the             Subcommittee:
        I am very pleased       to be here today to discuss GAO's efforts
to measure potential         losses in federally     subsidized     units    from
Low- and moderate-income           rental housing.     My testimony      today is
based on a study that GAO is currently             conducting    at the
Subcommittee’s      request.      The results    I report   are preliminary       and
subject    to revision    as we complete our work.
        In recent years,     several   studies have suggested that a Large
number of federally      subsidized     units could be Lost from the Low-
and moderate-income      housing inventory     as private  owners end their
participation     in federal     housing programs by prepaying   their
mortgages.      Losses from the more than 600,000 units supported        by
mortgage     insurance      and    Fntefitit     ,3~lJ~i~j,~fl   under   the   section
221(d)(3)  and section   236 programs have been estimated      at more
than 300,000 early in the next century;       for some other programs,
the numbers are even higher.      This potential   loss of existing
low- and moderate-income    housing poses a serious     policy  problem,
given that little    new housing of this type has been constructed
over the last few years.
Background     on federal         housing      programs
       Federal housing assistance              for Low- and moderate-income
families      changed dramaticaly           after    the passage of the Housing
Act of 1937.         Initially,      the federal       government,     working through
Local public       housing authorities,           assumed primary       responsibility
for meeting the goal of "a decent home and a suitable                         Living
environment       for every American family,"               as enunciated     in the
Housinq Act of 1949.              Beginning    in the 1950's,       however, this
responsibility         began to shift       to the private        sector through       a
series    of programs designed to provide                 federal   mortgage
insurance      and interest-rate          subsidies     to reduce the financing
costs of new construction.                During the 1960's        rent subsidies        were
added to the support            for the private        development     and operation       of
Low- and moderate income housing.                   In the 1970's and 1980's           these
subsidies      were changed from project-based                to tenant-based
programs,      culminating        in the introduction         of vouchers in 1983.
       In our work, we focus on the Low- and moderate-income
housing deveLoped under sections             221(d)(3)      and 236 of the Housing
Act, as amended.          Under these programs,          the Federal Housing
Administration       insures    40-year mortgages on the covered
properties.      Initially,      the Loans were made under section
221(d)(3)     at the prevailing       market interest          rate (MR), but a 1961
amendment provided          for below-market      interest       rates (BMIR) of 3
percent through        federal   subsidies     paid up front.          In 1968, this
was replaced     by monthly subsidies          to reduce the interest          rate
under section      236 to 1 percent.         In return       for the mortgage
insurance     and interest-rate       subsidies,       the owners of these
projects    agreed     to restrict       the   use    of the   units    to   low-   and
moderate-income        househol.ds.
Threats    to the     subsidized      inventory
       Over the past few years,           a major concern of the Congress,        as
exemplified      by the emphasis at this hearing,           has been that owners
of section      221(d)(3)     and section    236 properties     generally  have an
option     to prepay their      mortgages    at the end of 20 years,      rather
than holding       them for the full. 40-year       term.     In many cases, such
prepayment     would remove the use restrictions            that keep the
property     available     for low- and moderate-income         renters.     The
properties     could then be used for other purposes,              such as market-
rate rentals,        condominium    conversions,    or even commercial
        The ability      to prepay and convert             the property     is limited       in
several      ways.    First,     projects      owned by nonprofit         or public
sponsors      and those owned by for-profit                owners who purchased        them
from nonprofit        owners remain subject            to the use restrictions           for
the entire        40 years of the original            mortgage,     unless the
Secretary       of the Department         of Housing and Urban Development
 (HUD) specifically          approves the prepayment            and lifting     of use
restrictions.          In addition,       for-profit       sponsors whose projects
have received        rent supplements          or flexible      subsidies     must
maintain      their   properties       as low- and moderate-           income rental
units     for the life       of the o*riqinal        mortgage.      Under these
circumstances,         concerns     about prepayment          among these owners are
         However, many other owners, for-profit                  sponsors who are not
affected     by the restrictions        mentioned        above, appear to be
eligible     to prepay and convert         their     properties       to presumably
more lucrative      uses in the next few years.                  Some of these owners
may not be able to end use restrictions                   immediately,      because they
also have section        8 subsidies     that require          that the properties     be
used for low- and moderate-income                rentals     throughout     the life  of
the section      8 agreements.       But for many such properties,               the
section     8 agreements    will   end during        the first       half of the
        The major reasons why property              owners might choose to prepay
revolve      around the potential           for higher     profits      if the properties
are put to other uses.               Currently,    profits      in these programs are
limited      to a pretax      return     of 6 percent      annually       on the ori inal
equity     investment,       usually     a down payment       amounting       to 1 lrgkzt
of the project         cost.     This could mean that pretax               profits     from
these projects         are rather      modes%, given the overall              change in
prices    over the past 20 years.               But after-tax        returns      could be
much higher,       because mortgage payments in the early years of the
mortgage would go largely              for tax-deductible          interest,       and
generous depreciation            rules have already         allowed       much of the cost
of the property      to be written   off.   However, by the 20-year
point,    many of these projects     no longer benefit  from these tax
provisions,     because mortgage payments are largely      for
nondeductible     principal   payments, and often the depreciation    has
been fully    taken.
        In 1987, the Congress addressed the prepayment problem
temporarily      by requiring      that HUD be notified      of an owner's
intent    to prepay and that,         in all prepayment cases, the owner
develop a plan of action,            acceptable    to HUD, to deal with the
effects     of prepayment on low- and moderate-income             households.
These restrictions,        effectively      providing   a moratorium    on
prepayments,      are scheduled       to expire    on September 30, 1990.        The
major issue now confronting             the Congress is whether action        is
necessary to prevent         the loss of housing units through
prepayments      and what such action         should be.    To answer these
questions,     information      is needed on the likely       scope of the
prepayment problem.
       To help address this concern,           we were asked to review the
estimates     of possible     losses to low- and moderate-income             housing
to determine     the soundness of the methods the researchers                 used
and the reasonableness          of the estimates     of loss they reported.
In conducting      this review,      we used bibliographic        data bases and
the advice of a panel of experts            to identify     studies     that address
the potential      loss of low-hand moderate-rental            housing.      We found
only four studies       that presented      original    estimates     of potential
losses in the subsidized          low- and moderate-rental         housing
universe.      Three of these were conducted by government agencies:
GAO, the Congressional          Budget Office     (CBO), and HUD. The fourth
was carried     out by the National       Low Income Housing Preservation
Commission,     a private     concern financially       supported     by the
National    Corporation     for Housing Partnerships         and the Ford
       In addition,        we gathered    information        on experience       up to now
with prepayments         under the Housing and Community Development Act
of 1987.     We found that by December 31, 1989, only 92 property
owners had notified          HUD of their      intent    to preoay.         Of these,  45
had filed    plans of action,         7 of which HUD had approved for
incentives.       In no case had an owner actually                been allowed to
erww .      Thus,   little     is  known    about    the   actual    effect    of
preoayments     on the properties         or tenants       involved.
         Finally,    during the autumn of 1989, we interviewed         housing
officials,        market experts,     and property   owners in four cities     to
collect      information    on the likely     effect  of local market factors
on prepayment decisions.           Two of the cities,      Boston and Los
Angeles,       are "tight"   markets,    where demand for low- and moderate-
income housing is high relative            to supply (although    there has
been some softening    of the Boston market in recent months), while
the two others,    Denver and Houston, have demand that is low
relative to supply.
Findings   on prepayment   estimates
       With respect to our analysis       of the four studies,   two of
them--those      by CBO and GAO-- assumed that all owners would prepay
when they can.1        Both HUD and the Preservation    Commission made
estimates     of the probable    Losses that might be expected from the
insured    low- and moderate-income     housing inventory.     As one would
expect,    their   estimates  are lower than the CBO and GAO estimates,
which aimed at providing       figures  on the maximum losses that might
be expected.
       Table 1 shows that the four studies     provide   a wide range of
estimates    of the inventory   covered by sections    221(d)(3)    and 236,
ranging from 581,000 to 645,000 units.       This results      from some
technical.   differences   in how the data were extracted      from HUD's
data bases and the exclusion      of section 221(d)(3)-MR      properties
from the CBO study.

      *'The GAO study also included     an estimate      of the minimum
potential   losses.   This estimate   was  6,000    units   by 2005,
assuming all owners held their      mortgages for the fuLL term.
Table      1:    Estimated         Prepayments        Under Sections             221(d)(3)      and

                                         Commission                  HUD            CBO          GAO
                                            (1988)                  (1986)         (1987)       (1986)

Total      Inventory                         645                     604            581b         627

Eligible        to prepay                    367                     364            334          42Sc

Projected        prepayments
by 2005
   Maximum                                   not                     not            334          255
                                             pr0j.d                  pr0j.d
   Probable                                  243e                    154f           not          not
                                                                                    oroj.d       proj.d
Maximum and probable
prepayments    as % of
total  inventory                                 38                   25             57           41

aIn     thousands      of units.
bExcludes        section      221(d)(3&MR         properties.
cIncludes    some properties              that could prepay                   but would still         be
subject   to use restrictions                because of rent                  supplementsor
flexible    subsidies.
dNot projected.
eEstimate  of the National Low Income Housing                                  Preservation
Commission is for prepayments  by 2002.
fHUD estimate          is    for    prepayments           "ever."
Source:     National     Low Income Housing Preservation            Commission,
Preventing    the Disappearance        of Low-Income Housing, (Washington,
D.C. : 1988):      testimony    by Thomas Demery, Assistant            Secretary  for
Housing and Fede bra1 Housing co ~mmissioner , before the Hous e
Subcommittee     on Housing and coImmunity DeNvelopment (Octobe !r 21,
1986);    Congressi onal Budget Of fice. , The Potential            Loss of
Assisted    Housing Units as Cert ain MO=                          t Subsi 2Y
Programs Mature (Washington 7 1.c.: March y$+yjf                    S. Genelral
Accountina    Offic Fe, Rental Housing:         Pote ntial    Redu .ction i n the
Privately-Owned       and Federal ly Assisted       I nventory,    GAO/RCEID-86--
1761% (Washingto In, D.C.:       Ju .ne 1986) .

      However, three of the studies         had similar      estimates       of the
number of units eligible       to prepay,    334,000      to 367,000      (the
Larger GAO estimate     includes   specific    for-profit        properties     with
rent subsidies    and flexible    subsidies    that can prepay but would
still  be subject   to use restrictions).         The importance          of this
broad agreement on the number of units eligible                for prepayment        is
that it provides    an approximate     upper limit        to the scope of the
prepayment problem.       That is, it appears that at most about
367,000 units could be lost from the inventory               through
       The estimated       losses in these programs by early in the next
century    range from 154,000 to 334,000 units.               At the high end,
CBO estimated      that,     by 2005, all the 334,000 units eligible            to
prepay could do so, representing                about 57 percent   of the totaL.
At the Low end, HUD concluded              that of 364,000 units eligible         to
prepay,    84,000      "definitely"       would prepay and an additional
70,000 "likely"       would do so, for a total          of 154,000   units,   or
about 25 percent         of the total      inventory.    The Preservation
Commission's      analysis      predicted     that by the year 2002, a total         of
243,000 units       (38 percent)       would be Lost, a figure      similar   to
GAO's estimate       of a maximum of 255,000 units           (41 percent    of the
inventory)     by 2005.

       These differences             Largely     reflect      the methods that were used
to generate       the estimates.             As I noted earlier,          CBO and GAO
assumed alL eligible              units would prepay when able to do so to
arrive    at their       maximum es&mates;              differences      between their
estimates      largely       reflect     different       assumptions      about the effect
of related       subsidy arrangements.                In contrast,       HUD asked Loan-
service    officers        in each region to fill               out a questionnaire      on
the financial        and physical          condition        of each property     in the
inventory      and then to reach a judgment on the likelihood                       that
each would prepay.              The Preservation            Commission developed       a model
of the factors         likely      to affect       owners' prepayment decisions           and
collected      data on a sample of 300 properties,                     to apply to the
model, from HUD, housing market experts                         and the property    owners.
         Although      both HUD and the Preservation               Commission made
serious      efforts       to estimate     probable     Losses, we believe           that
there are problems with both studies.                     The HUD method of relying
on Loan-service            officers'    opinions     of whether specific
properties        were likely        to prepay at any time over the term of the
40-year mortgage seemed particularly                    weak, especially          given that
consistent        criteria       were not specified.          The Preservation
Commission's         model was a far more useful              analytical       tool,    but as
with most such models, its predictions                    are subject        to error      from   .
uncertainty         about important        parameters      (for example, the
projected        rate of inflation),          variables      omitted     from the model,
and, of course,            unforeseen     events.     Moreover,      the data the
Preservation          Commission used were based on only 198 properties
eligible      to prepay,          a sample size far too small to account for
differences     among housing markets.  (See the appendix for a more
thorough    discussion  of the HUD and Preservation Commission
Findings    from   our case studies
       Our interviews      with housing officials,        market experts,         and
owners of section        221(d)(3)      and section   236 properties       in Boston,
Denver, Houston and Los Angeles point to the importance                      of Local
market conditions        in prepayment decisions.         We interviewed
property     owners eligible        to prepay by September 30, 1994--that
is, within     about 5 years of the date of the interview.                   These
owners included      about 60 percent of those eligible             to prepay in
Denver and Houston and a judgment sample of about one third                       of
those in Boston;       in Los Angeles,       the large number of owners
meeting our selection          criteria    meant that our judgment sample was
only about 6 percent of those eligible.                Our interviews       were
intended     to be illustrative,         and not to provide     nationally-
representative      estimates       of owners' plans.
        In the tight    housing markets Boston and Los Angeles,               all
but one of the owners we interviewed             would Like to prepay as soon
as possible.      Our analysis      of owner responses showed that
current    market value and built-up        equity    outweighed    all other
factors    in these decisions.        As a result     of the rapid
appreciation     of property     values in these areas over recent years,
the owners' equity       in the properties       has increased     significantly.
We found that,       in most cast&; the built-up         equity  is so great
that even substantial        conversion    costs will     not dissuade these
owners from converting        the property     to market-rate      housing.
       Owner responses in the Low-demand markets Denver and
Houston were markedly different:               none of the owners indicated
they planned to prepay.             In both cities,      relatively      Low rents and
high vacancy rates in the general housing market disposed owners
not to prepay the mortgage and convert                the property       to market
uses.      In general,     the owners we interviewed           told us subsidies
such as HUD's section          8 program provide        guaranteed     rent revenue
and occupancy for their           properties.       For example, in the Denver
properties       we examined, 325 of the total           of 418 units       received
section      8 Loan management set-aside          funding.       The owners'
responses showed that the average vacancy rate for these
properties       is estimated     to be approximately         3 percent,     far below
the overall        Denver housing market figure          of 10 to 12 percent.          In
addition,      the cost of conversion         coupled with the uncertainties
of operating        in the general market appear to be factors                 in
owners' decisions         against     prepayment    in these markets.
        Thus, our findings    suggest that the prepayment problem is
closely    tied to the opportunities     available to property owners in
the local     market.    This means that the extent of the problem may
vary widely     from city to city.
       Overall,   we conclude that consistent       estimates     of the
maximum loss of low- and moderate-income          insured     rental   units
throug6    mortgage prepayments   are available,       based on reasonable
agreement about the size of the total         inventory     and about the
number of units that are eligible       for prepayment.          However,
estimates    of probable   losses are uncertain.        Under these
circumstances,     if the Congress wants to avoid the loss of low-
and moderate-income      housing units,   two broad responses may be
       First,     the Congress could decide to aim preservation                            efforts
at all the approximately             367,000 units eligible                 to prepay.        This
strategy      would recognize        that there is, after              all,      an upper limit
on the number of units eligible                  for prepayment and that while the
probable      number of prepayments            is hard to predict              accurately,
all current       predictions      suggest       that many of the eligible                 units
wT1 prepay.          Precise    efforts      to prevent        prepayment could involve
increasing      the current      profit      Limits      under sections           221(d)(3)       and
236 through       some combination         of (1) tax credits,                (2) changes in
the basis on which profit              levels      are computed (for example,
making the basis current             appraised        value     rather     than original
equity    investment),        or (3) other incentives.                 The disadvantage            of
using tax credits          is that it involves             losses in federal           revenues.
Changing the basis for computing profits,                        however, has the
disadvantage        of being Likew'to            increase       rents to tenants           and
costs to the federal           government through             increases        in rental
assistance      to those tenants.
       Of course,   if this approach is adopted,      some owners who
would not have prepaid      in any case will   receive   windfatl
benefits.     But this cost may be partly    offset    by the likelihood
that some owners who might have defaulted        would not do so.
         An alternative      strategy,      however, would be to recognize
that,     to some extent,        the prepayment problem is closely              tied to
prevailing       conditions      in specific      local     housing markets.       Using
this strategy,         the Congress could extend incentives                to owners of
low- and moderate-income             housing projects         that incorporate      a
flexible      approach based on local           housing market conditions.             That
is, HUD could be directed             to continue,         much as it does now,
offering      incentives      in selected      areas case by case, taking
account of the appraised             value of the property,           the costs of
conversion       to other uses,       the    demand    for   rental   or condominium
units     in the Local market,          and other such factors.            This means
that more assistance            might be needed in tight            housing markets
than in those where demand is Low relative                      to supply and also
that.the      level    of assistance       in any given market might change as
market conditions          change.

        This strategy       has the advantage of being potentially                  less
costly    than offering         incentives     to all eligible      owners.      It
accords with our findings               that in relatively      low-demand markets,
most owners benefit           from the virtually        guaranteed    cash flow and
relative      protection      from market risks afforded           by participation
in the assisted-housing             programs and are, therefore          unlikely        to
convert     their    buildings      to other uses or to be able to sell them
as a way of extracting             built-up    equity.
        But to make such a strategy          work, HUD wilt       need to improve
its collection      and maintenance        of information      on the properties
covered by these programs.           More importantly,         close management
and careful     oversight     would also be needed.           But the fact    that
incentives     would be offered      in Limited        numbers of cases should
simp'lify   management oversight.           In addition,      HUD's experience
with using current        incentives     to address the prepayment         issues
should prove useful         in optimatly      targeting    the effort.
        WhiLe alternative        uses of some eligible     properties      would be
so attractive       financially       that HUD would not be able to offer
adequate incentives          to retain     them among Low- and moderate-rental
housing,      no other plan would prevent         this situation,      either.     Of
course,      %he moratorium      on prepayments     could be made permanent,
but it is also possible           that either    Legal challenges      or owners'
actions      (such as refusal       to maintain   the property    or massive
numbers of defaults)           would make any blanket     continuation       of these
restrictions       impractical.        1
     This concludes  my prepared  statement,                 Mr. Chairman.  I wiL1
be happy to respond to any questions    that                you or members of the
Subcommittee may have.

      As I discussed   in the text,  only the HUD and Preservation
Commission studies   attempted   to derive  estimates  of the probable
losses that could be expected from prepayment actions.        Both of
these studies   used methods with serious    problems.
       HUD in its study,     presented    in congressional      testimony    in
March 1987, estimated      that about 154,000      units    (25.5    percent   of
HUD's inventory      of Sections   221(d)(3)   and 236 units)        would "ever"
be probable    losses from prepayment.        Further,     HUD concluded     that
75 percent    of the losses would occur in three of its regions.
HUD concluded     that this number of units at risk in a few
locations    would not create the crisis       that was feared by many
housing experts.
        To develop this estimate          of the probable         number of
prepayments,      HUD directed      each of its field          offices     to determine
the properties      under its jurisdiction             that were either       likely   or
unlikely    to prepay their        mortgages.        We believe      that the method
HUD used to predict        these losses is seriously              flawed
methodologically.         First,    HUD's effort         is based on opinion.         The
judgments of HUD's loan-service              officers,      based on their        own
records and knowledge of the properties,                   were the sole source of
information     on likely      prepayments.        But no consistent         criteria
for judgment were prescribed,             making it impossible           to know
whether other loan-service           officers      would have reached the same
        Second, in reviewing        the questionnaire      used to make these
estimates,     we found that it lacked many key financial                indicators,
such as market value and alternative                uses, cost data on needed
repairs    and improvements,         and whether the local        real estate
market and individual          property's     financial   situation    is such that
conversion     costs are prohibitive.            However, there were questions
about other relevant         factors,     such as proximity       to downtown or to
schools,    parks,   libraries,       and playgrounds.       Questions     about
negative    neighborhood       amenities    and safety    were also addressed.
       Third,     no information   was gathered       from property    owners
regarding      their   intentions,  possible     alternative      uses for the
property,      or its financial    and physical       conditions.     The
opinions      queried were those of government loan-service              officers,
not those of the most relevant          population,       private   owners.
       Fourth,    we found     significant problems with the            data    HUD
loan-service      officers     used to make their  estimates.            We
interviewed       HUD field      and reqional         office     loan-service       officers
on selected       properties        in Boston, Denver, Houston, and Los
Angeles.      We designed these interviews                   to learn about the
property-specific          data that HUD field             offices     have and about how
much loan-service          officers       know about their          assigned properties.
We found a wide variation               in the amount of property-specific                    data
available     from location          to location.          For example, loan-service
officers    thought       some properties         were owned by for-profit
interests     when in fact they were nonprofit                     owners ineligible          to
prepay.     Some     officers      had    a significant         amount   of   information
about the financial            condition     of properties          while others did not.
Given their       responses,       we believe       that the HUD officials              often
lacked data on property-specific                   real estate       values and had at
best vague ideas of possible                alternative         uses for properties.
        We also found some discrepancies                in the numbers reported.
For example,         one HUD Denver region determined               that of its 145
properties        (12,347  units)    eligible       for prepayment,        as few as 15
properties        (1,182  units)   will     likely     prepay,     and 79 properties
(7,587     units)     are not likely      to prepay.         For the remaining       51
properties,        the region's    loan-service         officers      responded that
they did not know what the owners' likely                     action     would be.
However, HUD's 1987 report places                  57 properties        in the unknown
category      for the entire      nation.        This suggests        that some changes
may have been made after           data were submitted             by HUD's regional
         Given all these probl&ns,             we believe       that    HUD's estimates
cannot      be regarded as reliable.
        The consensus of experts         we consulted      was that the
Preservation       Commission's    study is the best available              addressing
likely    housing losses.        The report     builds    on a carefully         selected
sample of HUD properties          and uses a soohisticated           modeling
technique     to estimate     the likelihood       that property       owners will
default     or prepay their      mortgages under various          alternative
scenarios.        The sample was drawn from the older portion                   of the
Federal Housing Administration            multifamily      rental    inventory
 insured    before   1975.    Thus, it included        properties      slightly
closer    to the date of possible         prepayment than a random sample of
all properties       would have produced (but only 17 percent of the
properties      in the universe      of concern were developed             after    1975).
For each sample property,          the Preservation        Commission obtained
data from HUD's files,          market experts,       and the property          owners.
        In general,   we believe   that the commission's       approach
provides    a reasonable    basis for making estimates        of how economic
factors    might influence     the overall  likelihood     of prepayments
under the Section      221(d)(3)   and Section     236 programs.     However,
several    problems with this analysis      limit    the reliability    of the
estimates    generated     by the model.    First,  the sample size for
this study,     300 properties,    included    only 198 eligible         to prepay.
This sample size was too small to permit predictions                  about
specific    markets.      However, our own fieldwork       suggests      that
markets vary greatly        in the extent to which they provide
attractive    alternative     uses for properties     eligible      for
        Second, the model assumes no government action         and therefore
does not consider      HUD's ongoing efforts      to prevent tenant
displacement.      For example, it assumes that no section          8 loan
management set-aside      contracts   will    be renewed or replaced     at
expiration,     whereas HUD currently      extends these contracts.        It
also does not take into account state limits            on use conversions
that may apply to many projects.
        Third,  while the model examines the economic behavior                 of
owners, it is not designed              (as the study notes) to take into
account other influences            on owners' decision      making.    These
include     changes in local         real estate markets,      local politics'
owners' concern about the effect               on tenants,   risks inherent      in
changing the character           of the real estate,       and the possibility
that financial       information       will   not always be perfect.        Such
limitations      are usually       inherent    in formal forecasting      models,     of
         Fourth,    the model appears to be sensitive       to assumptions
about a number of parameters:            For example, when the inflation
rate is assumed to be zero, rather            than 5 percent,     the estimated
number of units lost through prepayments             drops from 243,000 to
131,000,      or 23,000 fewer than HUD's estimate.          Given that
increases        in rents may be near zero, or even negative,           in some
markets,      this    is a serious limitation     on the  model's    predictive
       Thus, while the Preservation      Commission's     estimates      may be
the best available,    they are subject     to significant       limitations.
From this analysis    and our own fieldwork,      we believe       that these
problems could result     in an unreliable    estimate     of the likely
losses in federally    supported    low- and middle-income         housing