OFFICE OF AUDIT DRAFT
REGION 7
For Discussion and Comment Only - Subject to Review and Revision
KANSAS CITY, KS
U.S Department of Housing and Urban
Development, Office of Single Family Housing
FHA’s Loss Mitigation Program
2014-KC-0004 SEPTEMBER 24, 2014
Issue Date: September 24, 2014
Audit Report Number: 2014-KC-0004
TO: Kathleen Zadareky, Deputy Assistant Secretary for Single Family Housing, HU
//signed//
FROM: Ronald J. Hosking, Regional Inspector General for Audit, 7AGA
SUBJECT: Lenders Generated $428 Million in Gains From Modifying Defaulted FHA Loans
Attached is the U.S. Department of Housing and Urban Development (HUD), Office of
Inspector General’s (OIG) final results of our review of lender gains from Federal Housing
Administration (FHA) loan modifications.
HUD Handbook 2000.06, REV-4, sets specific timeframes for management decisions on
recommended corrective actions. For each recommendation without a management decision,
please respond and provide status reports in accordance with the HUD Handbook. Please furnish
us copies of any correspondence or directives issued because of the audit.
The Inspector General Act, Title 5 United States Code, section 8M, requires that OIG post its
publicly available reports on the OIG Web site. Accordingly, this report will be posted at
http://www.hudoig.gov.
If you have any questions or comments about this report, please do not hesitate to call me at
913-551-5870.
Office of Audit Region 7
400 State Avenue, Suite 501, Kansas City, KS 66101
Phone (913) 551-5870, Fax (913) 551-5877
Visit the Office of Inspector General Web site at www.hudoig.gov.
September 24, 2014
Lenders Generated $428 Million in Gains From
Modifying Defaulted FHA Loans
Highlights
Audit Report 2014-KC-0004
What We Audited and Why What We Found
We audited the U.S. Department of Lenders generated an estimated $428 million in gains
Housing and Urban Development’s from the sale of Government National Mortgage
(HUD) Federal Housing Association securities when modifying defaulted FHA
Administration (FHA) loss mitigation loans in fiscal year 2013. These loan modifications
program. We initiated this audit due to were completed as part of FHA’s loss mitigation
our concern that FHA might have program. None of these lender generated gains were
incurred costs while allowing lenders used to offset FHA’s insurance fund costs. As a result,
to make large amounts of money by FHA missed opportunities to strengthen its insurance
modifying defaulted FHA-insured fund.
loans. Our audit objective was to
determine the extent to which loans
modified under the FHA program
generated gains for the lenders.
What We Recommend
We recommend that HUD perform a
study of the loan modification program
and evaluate whether any changes are
needed to strengthen the insurance
fund.
TABLE OF CONTENTS
Background and Objective 3
Results of Audit
Lenders Generated $428 Million in Gains From Modifying Defaulted FHA Loans 4
Scope and Methodology 6
Internal Controls 10
Appendixes
A. Schedule of Funds To Be Put to Better Use 11
B. Auditee Comments and OIG’s Evaluation 12
2
BACKGROUND AND OBJECTIVE
The U.S. Department of Housing and Urban Development (HUD) initiated its loss mitigation
program to provide maximum opportunities for Federal Housing Administration (FHA)-insured
borrowers to retain home ownership. The program delegates loss mitigation responsibility and
authority to lenders. They must consider the comparative effects of their elective servicing
actions and take steps that can reasonably be expected to generate the smallest financial loss to
FHA. Such actions include but are not limited to deeds in lieu of foreclosure, preforeclosure
sales, partial claims, assumptions, special forbearance, and recasting of mortgages. Regulations
at 24 CFR 203.605 require lenders to evaluate all loss mitigation techniques available before four
full monthly installments due on the mortgage have gone unpaid. FHA pays lenders an incentive
of $750 for each loan modification they complete.
Lenders are authorized to package FHA loans into securities as mortgage-backed securities and
offer them to investors willing to purchase them. The Government National Mortgage
Association (Ginnie Mae), through its mortgage-backed securities program, guarantees these
securities. Security holders receive a “pass-through” of the principal and interest payments on a
pool of mortgages, less amounts required to cover servicing costs and Ginnie Mae guaranty fees.
The Ginnie Mae guaranty ensures that the security holder receives the timely payment of
scheduled monthly principal and any unscheduled recoveries of principal on the underlying
mortgages, plus interest at the rate provided for in the securities. The pools are composed of
mortgages that are insured or guaranteed by FHA and other government entities.
Lenders issuing securities must pay security holders on time the full amount specified by the
terms of the securities. If pooled loans are delinquent or in foreclosure, lenders must use their
own funds to pay security holders when payment is due. For loans backing a Ginnie Mae
security which has an issue date on or after January 1, 2003, lenders may repurchase a loan
without written permission from Ginnie Mae if the borrower fails to make a payment for 3
consecutive months. Lenders may repurchase any pooled loan for an amount equal to 100
percent of the loan’s outstanding balance and may recover advances from funds remaining after
they pay off the security holders.
Modified loans that have successfully completed the modification process and have been
permanently modified may be repooled. To be eligible for repooling, the permanently modified
loan must be current as of the issuance date of the related security. When a security is sold for a
premium, it is sold for more than its face value (unpaid principal balance), and when it is sold at
a discount, the lender receives less than the face value. The lenders receive this lump-sum
payment at the time of the sale of each mortgage-backed security. Typically, the higher the
interest rate of the security, the higher the premium it is sold for. Currently, there is no
requirement or restriction on how the lenders use these gains.
Our audit objective was to determine the extent to which loans modified under the FHA program
generated gains for the lenders.
3
RESULTS OF AUDIT
Lenders Generated $428 Million in Gains From Modifying Defaulted
FHA Loans
Lenders generated an estimated $428 million in gains from the sale of Ginnie Mae securities
when modifying defaulted FHA loans in 2013. These loan modifications were completed as part
of FHA’s loss mitigation program. FHA does not have requirements governing the use of these
lender gains and it does not receive information on the amounts of the gains generated. As a
result, FHA missed opportunities to strengthen its insurance fund.
Lenders Generated Gains
Lenders generated an estimated $428 million in gains from the sale of Ginnie Mae
securities related to modified FHA loans. These loan modifications were
completed as part of FHA’s loss mitigation program. Lenders received these
gains from modifying 67,048 defaulted FHA loans during fiscal year 2013 and
packaging them into Ginnie Mae securities between September 2012 and April
2014. The sale of these securities was made possible by the FHA loss mitigation
program. When the original loans became delinquent, Ginnie Mae allowed the
lenders to repurchase the loans for an amount equal to 100 percent of the loans’
principal balance (that is, at face value or par). After successfully undergoing
loan modifications, these repurchased loans were eligible to be repooled in
mortgage-backed securities. The lenders were able to obtain these gains as a
result of the loan modifications completed as part of the FHA program.
FHA Lacked Requirements and
Data Access
FHA does not have requirements governing the use of the lender gains from the
sale of Ginnie Mae securities nor does it have access to the details of these
transactions.
Specifically, FHA does not impose any limitations on how the lenders use the
gains from the sale of these securities or how much they can generate in gains.
Lenders are free to sell the securities at a price the market will bear and are
unencumbered in their use of these gains.
In addition, FHA does not have a mechanism to record or track Ginnie Mae
securities sales data as lenders do not currently share that information with it and
there is no central repository for this information.
4
FHA Missed Opportunities
FHA may have missed opportunities to strengthen its insurance fund. Lenders
could be required to offset gains they obtained from the sale of securities for
incentive fees and claims for modified loans that redefault.
At a minimum, FHA could have reduced or eliminated the incentive fees paid to
lenders for modifying loans. For the 67,048 loans modified in fiscal year 2013
and repooled, FHA paid about $50 million in loan modification incentives to the
lenders.
Lenders could be required to utilize the gains to offset the losses realized when
the modified loans became nonperforming and the associated property was
foreclosed upon. If FHA required lenders to apply the gains against its claims for
FHA insurance on failed loans, the insurance fund would pay out smaller claim
amounts on the loans that fail.
Another opportunity to strengthen the insurance fund would be to decrease the
allowable interest rate for modified loans. While such a decrease would reduce
the lenders’ gains from repooling the loans, it should assist some homeowners in
maintaining their modified loan in good standing by lowering their payments,
and/or reducing the deferred principal amount in the form of a Partial Claim,
which is a subordinate lien that does not have to be repaid unless the modified
loan is sold or refinanced.
Conclusion
Lenders generated an estimated $428 million in gains from the sale of Ginnie Mae
securities related to modified defaulted FHA loans in 2013 and FHA did not seek
a portion of the gains to offset its incentive fees for loan modifications or claims
of modified loans that redefaulted. Therefore, it missed opportunities to
strengthen its insurance fund and should explore potential program modifications
to reduce future payments.
Recommendation
We recommend that the Deputy Assistant Secretary for Single Family Housing
1A. Perform a study of the loan modification program and evaluate whether any
changes are needed, such as (1) developing procedures to offset lender
gains from insurance fund payments, (2) reducing the allowable interest
rate for loan modifications, and (3) examining the incentives paid to
lenders modifying loans, to put $50 million to better use.
5
SCOPE AND METHODOLOGY
To accomplish our objective, we
Interviewed HUD staff,
Reviewed Federal regulations and mortgagee letters,
Reviewed the Ginnie Mae mortgage-backed securities guide, and
Selected and reviewed a statistical sample of successfully completed loan modifications.
We performed our audit between March and August 2014. Our audit generally covered October
1, 2012, through September 30, 2013.
The Single Family Data Warehouse is a large and extensive collection of database tables
organized and dedicated to support the analysis, verification, and publication of Single Family
Housing data. Using this system, we identified loans with loss mitigation claims in the
“loss_mitigation” table by a code of “32” (loan modification) in the “clm_typ” field. We
included only claims that had been paid, indicated by a value of “1” (claims paid without
corrections required) or “2” (claims suspended then paid) in the “clm_sts_cd” field. Using this
process, we identified 107,689 loan modification claims processed between October 1, 2012, and
September 30, 2013. We matched these loans against the system’s monthly transaction table for
Ginnie Mae, known as “gnma_loan_level.” This match identified 67,048 loan modifications that
had resold on the secondary market as Ginnie Mae mortgage-backed securities between
September 2012 and April 2014. Because our universe was determined based on claim dates
rather than the date the loan was actually modified, some loans repooled before their loan
modification claims were processed; therefore some loans were repooled in September 2012
even though the claims were processed in fiscal year 2013.
We used a sampling frame to represent the universe. Our sampling frame consisted of the
56,504 loan modifications that were handled by the top 9 servicers of FHA loans and were resold
as mortgage-backed securities. The single, first occurrence of a sale to a secondary market
mortgage-backed security between September 2012 and April 2014 was the sampling unit. In
the event that a loan was pulled from a pool and sold again on the secondary market as
sometimes occurred, we did not include the second sale in our universe as that sale might have
occurred with or without the loan modification.
For the purpose of stratifying the sample and testing the viability of the design, we estimated the
amount of gain on resale for each loan. We did so by applying the published market value of the
mortgage-backed security for the coupon rate that was closest to the yield of the associated
Ginnie Mae pool during the month when the loan entered that pool.
Resale rates can vary, and for the purposes of stratification, we used the published rates for the
end of each month based on common coupon rates of 3.0, 3.5, 4.0, 4.5, 5.0, and 5.5 percent. For
coupon rates between published rates, such as 3.75 percent, we used prorated resale values,
averaging the values above and below. For pools with yield rates below 3.0 percent, we applied
6
the resale value of 3.0 percent pools for that month. Similarly, for yield rates above 5.5 percent,
we used the 5.5 percent resale value for that month.
For the purposes of stratification, each sampling unit was given a valuation equal to the
outstanding principal at the time of pooling times the resale markup value in accordance with the
estimated markup values mentioned above.
The sample was designed as a stratified, optimized sample with seven strata. The strata are
designed to control for the variable amounts of income that result from applying changing resale
values to varied mortgage amounts. Modified loans were sorted and ranked by the sample unit
valuation calculated above and then stratified by percentile points along this ranking to control
for variance in dollar amounts.
To control variance and minimize effects from random selections within influential strata, we
used an optimized Neyman sample to assign samples to each stratum. A Neyman sample
distributes the selection of sampling units according to how much uncertainty an area has and
how much it will affect the final projection, thereby making the most effective use of samples.
Optimized samples require some basic knowledge of the relative variance between strata.
Because Neyman samples respond to relative variance between strata rather than exact variance
amounts, we can model a variance profile that is sufficient to design the sample without knowing
the exact rate of error that will be found during the execution of the audit. The sample valuation
method mentioned above is sufficient for establishing relative variance. The sample design was
stratified as shown in the table below.
Sample design and results
Loans in
lower Sample Loans in Sampling
Stratum sampling
boundary size universe weights
frame Estimated resale
net_loss Neg. $ 2 1,145 1,164 582.000 Loss on resale
0-10pct $0 3 5,533 6,339 2,113.000 $0
10-30pct $8,192 6 11,074 12,915 2,152.500 $2,780
30-50pct $11,986 7 11,068 13,079 1,868.429 $5,490
50-70pct $15,334 9 11,077 13,201 1,466.778 $8,178
70-90pct $22,592 16 11,070 13,351 834.438 $1,1672
90-100pct $30,654 27 5,537 6,999 259.222 $18,350
Total n/a 70 56,504 67,048 n/a n/a
We selected a sample of 70 pooled loans. We used computer-replicated sampling (audit
simulations) to test the performance of the sample design, with sample counts ranging from 60 to
150. Because the markup rate found will vary some from the reference rates used to design the
sample, our testing randomly varied the markup rate found in our simulated audits by +/- 30
percent to ensure that the sample design can accommodate that level of uncertainty.
7
After replicating the audit findings that would occur with this number set at various sample sizes,
we compared these typical audit results with the dollar amounts underlying our tests. The
recommended sample size was found to be extremely effective in preventing errors, and the
accuracy of probabilistic statements made with this sample design exceeded the stated
confidence interval – a one-sided confidence interval of 95 percent.
Sample records were randomly selected with the number of samples in each stratum being
optimized to get the most accurate dollar estimate for a given sample size. The audit sample was
selected by means of computer routines written in SAS®, using the surveyselect procedure and a
seed of 7.
To quantify the earnings that loan servicers accrue from reselling loans that have undergone loss
mitigation, we sampled the universe of 67,048 modified loans that underwent HUD’s loss
mitigation refinance process during HUD’s fiscal year 2013 and were successfully resold to the
secondary market during this period or during the 6 months that followed it. Using the loan
modifications originated by the top nine servicers as a sampling frame, we pulled a stratified,
optimized, statistical sample of 70 loans from this frame and then projected the results to the
entire universe of loans. The top nine frame we sampled from represented 84.2 percent of the
universe.
The sample was stratified according to the expected markup on the loan (in dollars) when it was
resold to a mortgage-backed security. Expected markup value of the resale was established by
applying the published, end-of-the-month market value for Ginnie Mae mortgage-backed
security pools with the coupon rate closest to that of the Ginnie Mae pool that bought the loan.
The coupon rate of the pool that bought the loan was calculated specifically for the month when
the loan was purchased. Later analysis showed that the expected resale value calculated in this
manner had a correlation coefficient of .91 when compared with the resale value found during
the audit, thereby showing a very strong correlation between the estimated value used to group
the sample and the mortgage-backed security resale value of the loan.
Samples were then randomly selected in accordance with a 70-count, optimized sample design,
using the surveyselect procedure in SAS®. The sample count was verified extensively, using
replicated sampling, and the design was found to be more than sufficient for making these
projections.
The audit team acquired records pertaining to resale on the secondary market, computed the
dollar amount of the resale gain, calculated a margin of error, and made a final projection on that
basis. This was done by computing the mean and standard error of the resale amounts, using the
means estimating procedure (surveymeans) in SAS®. Variances were calculated by using a
Taylor series.
Regarding the 67,048 loss-mitigated loans in our universe, we can say, with a one-sided
confidence interval of 95 percent that servicers earned at least $428 million for the loans that had
been restored by loan modification in fiscal year 2013 and resold on the secondary market to a
mortgage-backed security.
8
We relied in part on data maintained by HUD in its Single Family Data Warehouse database.
Specifically, we relied on the data to identify loans that were successfully modified during our
audit period. Although we did not perform a detailed assessment of the reliability of the data, we
corroborated the fields used to determine our sample universe against documentary evidence
supplied by the lenders for our 70 sample loans. Based on the work performed, we determined
that the computer-processed data were sufficiently reliable for the purposes of this report.
We conducted the audit in accordance with generally accepted government auditing standards.
Those standards require that we plan and perform the audit to obtain sufficient, appropriate
evidence to provide a reasonable basis for our findings and conclusions based on our audit
objective(s). We believe that the evidence obtained provides a reasonable basis for our findings
and conclusions based on our audit objective.
9
INTERNAL CONTROLS
Internal control is a process adopted by those charged with governance and management,
designed to provide reasonable assurance about the achievement of the organization’s mission,
goals, and objectives with regard to
Effectiveness and efficiency of operations,
Reliability of financial reporting, and
Compliance with applicable laws and regulations.
Internal controls comprise the plans, policies, methods, and procedures used to meet the
organization’s mission, goals, and objectives. Internal controls include the processes and
procedures for planning, organizing, directing, and controlling program operations as well as the
systems for measuring, reporting, and monitoring program performance.
Relevant Internal Controls
We determined that internal controls were not relevant to the audit objective because
this audit addressed an area without established rules or guidance. HUD could not
be reasonably expected to have controls limiting lender gains since it did not have
requirements limiting these gains.
Significant Deficiencies
We did not identify internal controls related to the audit objective to evaluate in
accordance with generally accepted government auditing standards. Our evaluation
of internal controls was not designed to provide assurance regarding the
effectiveness of the internal control structure as a whole. Accordingly, we do not
express an opinion on the effectiveness of FHA’s internal control.
10
APPENDIXES
Appendix A
SCHEDULE OF FUNDS TO BE PUT TO BETTER USE
Recommendation Funds to be put
number to better use 1/
1A $50,286,000
1/ Recommendations that funds be put to better use are estimates of amounts that could be
used more efficiently if an Office of Inspector General (OIG) recommendation is
implemented. These amounts include reductions in outlays, deobligation of funds,
withdrawal of interest, costs not incurred by implementing recommended improvements,
avoidance of unnecessary expenditures noted in preaward reviews, and any other savings
that are specifically identified.
In this instance, if HUD implements our recommendations, it will ensure that FHA will
reduce payments from its insurance funds due to modified loans as it will be able to
recapture incentive payments made to lenders or use them to offset future claims. In
fiscal year 2013, we determined that FHA made more than $50.2 million in incentive
payments to lenders processing loan modifications that were sold as Ginnie Mae
securities ($750 per loan modification times 67,048 loans in our universe). We expect a
similar payment next year if FHA does not take advantage of opportunities to strengthen
its insurance fund as loan modification volumes in fiscal 2014 are similar to fiscal year
2013 levels. This amount is conservative as it does not take into account FHA’s
capturing all the gains of the lenders.
11
Appendix B
AUDITEE COMMENTS AND OIG’S EVALUATION
Ref to OIG Evaluation Auditee Comments
Comment 1
Comment 2
12
Ref to OIG Evaluation Auditee Comments
Comment 3
Comment 4
Comment 5
13
Ref to OIG Evaluation Auditee Comments
Comment 6
14
Ref to OIG Evaluation Auditee Comments
Comment 7
15
OIG Evaluation of Auditee Comments
Comment 1 Regulations at 24 CFR 203.605 require lenders to evaluate all loss mitigation
techniques available before four full monthly installments due on the mortgage
have gone unpaid and take the appropriate loss mitigation action. Further, 24
CFR 203.501 requires that lenders must consider the comparative effects of their
elective servicing actions and take steps that can reasonably be expected to
generate the smallest financial loss to FHA. Therefore, lenders cannot just decide
to avoid all loss mitigation procedures and file a claim for the loss even if there
were no incentive payments. Once participating in the FHA program, they are
bound by its rules and regulations.
Comment 2 The National Housing Act, 12 USC § 1701, et seq, (the Act) imposes an
affirmative obligation upon the Secretary to ensure that the insurance fund
remains financially sound. 12 USC § 1708(a)(3). To accomplish this, among
other obligations, the Secretary is specifically authorized and directed to make
such rules and regulations as may be necessary to carry out the provisions of the
Act. 12 USC § 1715b. Regarding the payment of insurance benefits, the
Secretary is specifically authorized to pay insurance benefits that “shall be equal
to the original principal obligation of the mortgage (with such additions and
deductions as the Secretary determines are appropriate)….” (emphasis
added). 12 USC § 1710 (a)(5). HUD has already implemented this authority
through regulations specifying additions and deductions to the claim amount. See
24 CFR 203.400 et seq. Moreover, in the area of loss mitigation, “[t]he Secretary
may pay insurance benefits to the mortgagee to recompense the mortgagee for all
or part of any costs of the mortgagee for taking loss mitigation actions….” 12
USC § 1710 (a)(2). Therefore, the Secretary is authorized to determine what
additions to and subtractions from insurance claims are appropriate, as well as
what part of any loss mitigations costs to reimburse. Therefore, HUD can decide
if it is appropriate or necessary to deduct the lender gains from the sale of Ginnie
Mae securities on the secondary market from such claims.
Comment 3 While it is good that FHA reduced the maximum interest rate on modified loans
by 25 basis points last year, it needs to constantly monitor trends in the industry to
ensure that it is minimizing insurance fund costs.
Comment 4 FHA’s decision to reduce the incentive fee should not be based on the comparison
of the incentive fee with the one percent origination fee but on the actual costs of
the lenders and of the insurance fund. FHA needs to conduct its own study to
determine the actual costs of the lender and to determine if its loss mitigation
efforts are done with the least payout from the insurance fund. The National
Housing Act authorizes the Secretary to “pay insurance benefits to the mortgagee
to recompense the mortgagee for all or part of any costs of the mortgagee for
taking loss mitigation actions….” 42 USC § 1710(a)(2). Therefore, FHA is not
obliged to cover all lender costs.
16
Comment 5 We agree that this is a required step.
Comment 6 We only used published mortgage backed securities coupon rates for preliminary
grouping of loans into statistical strata. This allowed us to compute and blend
similarly sized amounts to further tighten our accuracy and to prevent wide
margins of error. The projected earnings from reselling loans modified in fiscal
year 2013 are based on actual, final resale amounts received from the lenders.
Comment 7 We did not classify the gains as “profit” because we do not know the costs of the
lenders. The objective of the audit was to identify the amount of the gains lenders
are obtaining from modifying the loans and the costs FHA incurs to enable these
gains so that it can determine if there are any changes needed for its loss
mitigation program.
17
Lenders Generated $428 Million in Gains From Modifying Defaulted FHA Loans
Published by the Department of Housing and Urban Development, Office of Inspector General on 2014-09-24.
Below is a raw (and likely hideous) rendition of the original report. (PDF)