oversight

Fannie Mae Purchased Single-Family Mortgages, Including those Purchased through Master Agreements, in Accordance with Selected Credit Terms Set Forth in its Selling Guide for 2015-2017

Published by the Federal Housing Finance Agency, Office of Inspector General on 2019-03-27.

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

          Federal Housing Finance Agency
              Office of Inspector General




  Fannie Mae Purchased Single-
Family Mortgages, Including those
    Purchased through Master
 Agreements, in Accordance with
Selected Credit Terms Set Forth in
  its Selling Guide for 2015-2017




 Audit Report • AUD-2019-006 • March 27, 2019
                 Executive Summary
                 The Federal Housing Finance Agency (FHFA or Agency), established by the
                 Housing and Economic Recovery Act of 2008 (HERA), is responsible for the
                 supervision, regulation, and housing mission oversight of Fannie Mae, Freddie
                 Mac, and the Federal Home Loan Bank System. Using its powers under
                 HERA, FHFA placed Fannie Mae and Freddie Mac (collectively, the
AUD-2019-006     Enterprises) into conservatorships on September 6, 2008. FHFA’s
                 conservatorship authority is separate and distinct from its supervision
March 27, 2019   authority.

                 Fannie Mae manages the quality of its mortgage purchases by requiring
                 mortgage sellers to comply with its Selling Guide. The Selling Guide sets
                 forth Fannie Mae’s underwriting standards and eligibility guidelines, as well
                 as its policies and procedures related to sales of single-family mortgages to it.
                 Fannie Mae’s underwriting standards are developed, in part, based on risk-
                 based criteria, which enables it to evaluate a borrower’s willingness and
                 capacity to repay a mortgage and the value of the property to ensure that it
                 provides adequate collateral for the mortgage. Risk-based criteria relating to
                 a borrower’s willingness and capacity include debt-to-income (DTI) ratio,
                 loan-to-value (LTV) ratio, and credit score, while collateral value is assessed
                 through property valuation. None of these criteria are considered in a vacuum
                 but are considered together to build a snapshot of the potential risk level of the
                 mortgage.

                 Historically, many mortgage sellers sought to sell mortgages to Fannie Mae
                 that did not meet the underwriting standards and/or eligibility requirements
                 in the Selling Guide. Fannie Mae captured these negotiated terms, referred
                 to as variances, with its mortgage sellers in a document called a “master
                 agreement.” Each master agreement supplemented the general requirements
                 of the Selling Guide and set forth the additional negotiated terms under which
                 Fannie Mae agreed to purchase mortgages from the mortgage seller.

                 This report sets forth the results of our audit in which we sought to assess
                 FHFA’s oversight of Fannie Mae’s master agreements with its single-family
                 mortgage sellers from 2015 through 2017 (review period). As part of the
                 audit, we analyzed master agreements for Fannie Mae’s top three single-
                 family mortgage sellers and found no variation between the terms in the
                 master agreements for DTI ratio, LTV ratio, credit score, and property
                 valuation method from the terms for the same elements set forth in the
                 Selling Guide.

                 We also obtained information from FHFA and Fannie Mae and analyzed loan-
                 level data in FHFA’s Mortgage Loan Integrated System (MLIS) for all single-
                 family mortgage sellers to determine whether the credit terms for DTI ratio,
                 LTV ratio, credit score, and property valuation method for the mortgages
                 purchased by Fannie Mae differed from those credit terms in the governing
                 Selling Guide. For the single-family mortgages purchased by Fannie Mae
                 during the review period, through our analysis, we identified some differences
                 with these credit terms, but those differences were not material (less than one-
                 tenth of one percent of the mortgages purchased by Fannie Mae during the
AUD-2019-006     review period).

March 27, 2019   We did, however, identify issues with the reliability of certain data fields in
                 MLIS. Specifically, we found instances where data fields for our selected
                 credit terms were either missing information or were shown as “unknown,”
                 particularly with respect to the data field for property valuation method.
                 FHFA subsequently brought this matter to Fannie Mae’s attention for
                 resolution.

                 We also identified examples where FHFA performed oversight of Fannie
                 Mae’s use of master agreements.

                 We make one recommendation in this report to address the MLIS property
                 valuation method data field. In a written management response, FHFA agreed
                 with the recommendation.

                 This report was prepared by Tara Lewis, Audit Director; Pamela L. Williams,
                 Auditor-in-Charge; and Sarah Florino, Auditor; with assistance from Bob
                 Taylor, Senior Advisor. We appreciate the cooperation of FHFA staff, as well
                 as the assistance of all those who contributed to the preparation of this report.

                 This report has been distributed to Congress, the Office of Management and
                 Budget, and others and will be posted on our website, www.fhfaoig.gov.

                 Marla A. Freedman, Deputy Inspector General for Audits /s/
TABLE OF CONTENTS ................................................................
EXECUTIVE SUMMARY .............................................................................................................2

ABBREVIATIONS .........................................................................................................................6

BACKGROUND .............................................................................................................................7
      Fannie Mae’s Role in Single-Family Housing Finance ............................................................7
      Fannie Mae’s Underwriting Standards for Single-Family Loans .............................................7
      Fannie Mae’s Use of Master Agreements ................................................................................9
      During the Review Period, Fannie Mae Purchased Nearly 6.5 Million Single-Family
      Loans with a Total Unpaid Principal Balance of $1.49 Trillion, of which $1.16
      Trillion (78%) Were Purchased through a Master Agreement .................................................9
      FHFA Oversight of Fannie Mae’s Use of Master Agreements ..............................................10

FACTS AND ANALYSIS.............................................................................................................12
      For the Top Three Mortgage Sellers with Master Agreements, Testing Identified No
      Variances in these Master Agreements from the Credit Terms of DTI Ratio, LTV
      Ratio, Credit Score, or Property Valuation Method in the Selling Guide ..............................12
      Testing Identified Very Few Mortgages Purchased by Fannie Mae that Had DTI
      Ratios, LTV Ratios, Credit Scores, and/or Property Valuation Methods Outside
      Those Permitted by the Selling Guide ....................................................................................12
             Testing Identified Issues that Affected the Reliability of Certain Data Fields in
             FHFA’s MLIS .................................................................................................................13
      Testing Identified Some Oversight by FHFA of Fannie Mae’s Use of Master
      Agreements .............................................................................................................................14

FINDING .......................................................................................................................................17
      Information about Purchased Mortgages Was Incomplete for Certain MLIS Data
      Fields.......................................................................................................................................17

CONCLUSION AND RECOMMENDATION .............................................................................17

FHFA COMMENTS AND OIG RESPONSE ...............................................................................18

OBJECTIVE, SCOPE, AND METHODOLOGY .........................................................................19

APPENDIX: FHFA MANAGEMENT RESPONSE ....................................................................21


                                            OIG • AUD-2019-006 • March 27, 2019                                                                   4
ADDITIONAL INFORMATION AND COPIES .........................................................................22




                                 OIG • AUD-2019-006 • March 27, 2019                                       5
ABBREVIATIONS .......................................................................

DER                   Division of Enterprise Regulation

DHMG                  Division of Housing Mission and Goals

DOC                   Division of Conservatorship

DTI                   Debt-to-Income

DU                    Desktop Underwriter

Enterprises           Fannie Mae and Freddie Mac

FHFA or Agency        Federal Housing Finance Agency

HERA                  Housing and Economic Recovery Act of 2008

LTV                   Loan-to-Value

MLIS                  Mortgage Loan Integrated System

OIG                   Federal Housing Finance Agency Office of Inspector General

UPB                   Unpaid Principal Balance

VWE                   Variances, Waivers, and Exceptions




                         OIG • AUD-2019-006 • March 27, 2019                        6
BACKGROUND ..........................................................................

Fannie Mae’s Role in Single-Family Housing Finance

Fannie Mae performs an important role in the nation’s housing finance system by providing
liquidity, stability, and affordability to the mortgage market. Fannie Mae purchases single-
family mortgages from mortgage companies, commercial banks, credit unions, state and local
housing finance agencies, and other financial institutions (mortgage sellers) and either holds
these mortgages in its portfolio or packages them into mortgage-backed securities.

Fannie Mae’s Underwriting Standards for Single-Family Loans

Fannie Mae manages the quality of its mortgage purchases by requiring mortgage sellers to
comply with its Selling Guide. The Selling Guide sets forth Fannie Mae’s underwriting
standards and eligibility guidelines, as well as its policies and procedures related to sales of
single-family mortgages to it. Except where any such changes are significant, Fannie Mae has
been delegated authority to amend its Selling Guide: Fannie Mae updated its Selling Guide 29
times with regard to underwriting standards and/or eligibility guidelines during the review
period. Fannie Mae has developed a proprietary automated underwriting system, Desktop
Underwriter (DU), for use by mortgage sellers to determine whether a single-family mortgage
meets its underwriting standards as specified in the Selling Guide. The majority of single-
family mortgages purchased by Fannie Mae during the review period were processed by
mortgage sellers through DU and met the underwriting standards programmed into DU
(including any permissible exceptions to those standards). During that period, Fannie Mae
also purchased single-family mortgages underwritten using automated underwriting systems
other than DU, as well as manually underwritten single-family mortgages that met its standard
underwriting requirements and eligibility guidelines or agreed-upon (i.e., negotiated) terms
that differ from the standard requirements.

Fannie Mae’s underwriting standards are developed from different criteria, including risk-
based criteria. Risk-based criteria enable Fannie Mae to evaluate a borrower’s willingness and
capacity to repay a mortgage and the value of the property to ensure that it provides adequate
collateral 1 for the mortgage. None of these criteria are considered in a vacuum but are
considered together to build a snapshot of the potential risk level from the mortgage. Risk-
based criteria relating to a borrower’s willingness and capacity include DTI ratio, LTV ratio,
and credit score, while collateral value is assessed through property valuation. During the




1
    Property pledged as the security for a mortgage loan is the mortgaged real estate referred to as collateral.



                                      OIG • AUD-2019-006 • March 27, 2019                                          7
review period, Fannie Mae’s Selling Guide established the following standards for each of
these four elements:

    1. DTI ratio – The DTI ratio is derived by dividing the borrower’s total monthly
       obligations (including housing expenses) by his or her stable monthly income. During
       the review period, the maximum DTI ratio authorized in the Selling Guide for single-
       family mortgages eligible for purchase was 50%. 2

    2. LTV ratio – The LTV ratio is the relationship between the original loan amount of
       the first mortgage and the property’s appraised value (or sales price, if it is lower).
       During the review period, the maximum LTV ratio authorized in the Selling Guide for
       single-family mortgages eligible for purchase loans was 97%. 3

    3. Credit score – The credit score is a numerical value derived from a statistical
       evaluation of information in an individual’s credit file, at a fixed point in time. During
       the review period, the minimum credit score authorized in the Selling Guide for
       single-family mortgages eligible for purchase was 620. 4

    4. Property valuation – Property valuation methods provide an estimate of the current
       value of the property intended to secure the mortgage. During the review period,
       various property valuation methods, including traditional appraisals 5 and automated
       valuation models, were authorized by the Selling Guide.



2
  Fannie Mae Selling Guides in force during the review period included different maximums for the DTI
ratio based on various factors, such as method of underwriting, credit score, and product type. For example,
loans underwritten manually had a maximum allowable DTI ratio of 45% if the borrower met established
requirements for credit score and financial reserves, while loans underwritten through DU had a maximum
allowable DTI ratio of 50%.
3
  Fannie Mae Selling Guides in force during the review period included different maximums for LTV ratio
based on various factors, such as method of underwriting and product type. For example, a loan underwritten
manually for the purchase of a single unit principal residence could have an LTV ratio of up to 95%. The same
type of loan underwritten using DU could have an LTV ratio of up to 97%. Fannie Mae’s Selling Guide also
included loan programs that allowed an LTV ratio above 97%, such as: (1) the Community Seconds program,
which allowed a combined LTV ratio of up to 105%, and (2) government guaranteed mortgages, such as those
under Section 502 Guaranteed Rural Housing Loan Program that allowed an LTV ratio up to 100%.
4
  Fannie Mae Selling Guides in force during the review period included different minimums for credit scores
based on various factors, such as method of underwriting, LTV ratio, and product type. For example, while
loans with an LTV ratio up to 75% may have a minimum credit score of 620, a higher credit score was
required for loans with an LTV ratio exceeding 75%.
5
  An appraisal is a report that provides an opinion or estimated value based on interior and/or exterior property
inspections. An appraisal report is written by a state-licensed or state-certified appraiser. The appraiser must
have the knowledge required to perform a quality appraisal for the specific geographical location and property
type and have access to appropriate data sources.



                                    OIG • AUD-2019-006 • March 27, 2019                                             8
Fannie Mae’s Use of Master Agreements

Historically, many mortgage sellers sought to sell mortgages to Fannie Mae that did not meet
the underwriting standards and/or eligibility requirements in the Selling Guide. Fannie Mae
captured negotiated terms with its mortgage sellers in a document called a “master
agreement.” Each master agreement supplemented the general requirements of the Selling
Guide and set forth the additional negotiated terms under which Fannie Mae agreed to
purchase mortgages from the mortgage seller. Differences in terms between those in the
Selling Guide and those in a master agreement are called variances. Generally, variances can
be grouped into the following categories:

    •   Borrower willingness and capacity to repay the mortgage: exceptions to evaluation
        of the borrower’s credit history, liquid reserves, reliable and recurring income, equity
        investment, and/or the cumulative effect of these exceptions on loan performance;

    •   Valuation of property to secure the mortgage: exceptions to property valuation
        methods; and

    •   Other: exceptions for unique products (e.g., government loans or loans with unique
        eligibility requirements), and exceptions to process and/or operational requirements
        (e.g., document custody, servicing, etc.).

During the Review Period, Fannie Mae Purchased Nearly 6.5 Million Single-Family
Loans with a Total Unpaid Principal Balance of $1.49 Trillion, of which $1.16 Trillion
(78%) Were Purchased through a Master Agreement

According to data maintained by FHFA in MLIS, 6 Fannie Mae purchased 6,455,307 single-
family mortgages with an unpaid principal balance (UPB) of $1.49 trillion for the years 2015
through 2017. 7 See Figure 1.




6
  MLIS is an FHFA database that maintains loan-level data provided by the Enterprises. According to FHFA
officials, MLIS is used by several divisions within FHFA as a resource in support of policy decisions and
official publications. The loan acquisition data for Fannie Mae maintained in MLIS dates back to 1979 and the
loan performance data dates back to 1991. FHFA established requirements for the Enterprises’ input to MLIS,
including reporting frequency and submission requirements.
7
  These totals do not include “Refi Plus” mortgages. Refi Plus mortgages purchased by Fannie Mae were
refinances of mortgages that had been owned or guaranteed by Fannie Mae. Since the credit risk associated
with Refi Plus mortgages essentially replaced the credit risk on the mortgages that Fannie Mae had already
held, we excluded the Refi Plus mortgages from our audit.



                                   OIG • AUD-2019-006 • March 27, 2019                                          9
                     FIGURE 1. SINGLE-FAMILY LOANS PURCHASED BY FANNIE MAE

                                                    2015                 2016           2017       2015-2017
      Number of Mortgages Purchased                       1.9               2.4            2.1              6.4
      (in millions)
      UPB (in billions)                               $442                $559           $488            $1,489
Source: FHFA’s MLIS

Fannie Mae purchased these mortgages from mortgage sellers, pursuant to the negotiated
terms in master agreements and/or the Selling Guide requirements.

During the review period, single-family mortgages with a total UPB of $1.16 trillion (78% of
the total UPB of $1.49 trillion) were purchased by Fannie Mae through master agreements.
See Figure 2.

    FIGURE 2. SINGLE-FAMILY LOANS PURCHASED BY FANNIE MAE THROUGH MASTER AGREEMENTS
                                    AND THE SELLING GUIDE

                                     Master Agreementsa                                  Selling Guide
                                                                2015-                                       2015-
                             2015       2016      2017          2017            2015    2016     2017       2017
Number of Mortgages            1.5        1.8       1.6            4.9            0.4      0.6      0.5        1.5
Purchased (in millions)
UPB (in billions)             $347       $429      $379     $1,155                $95    $130      $109       $334
a
 According to Fannie Mae, these figures include loans delivered under master agreements that met Selling
Guide requirements.
Source: Fannie Mae

Fannie Mae officials reported to us that Fannie Mae began to transition away from master
agreements to tracking each negotiated term (i.e., variance) individually, beginning in
December 2016. Fannie Mae officials stated that tracking individual variances should allow
Fannie Mae to better track and analyze active variances. According to these officials, Fannie
Mae tracks these variances in its “Enterprise Contract Management System,” and as of July
2018, Fannie Mae had migrated approximately 60% of its mortgage sellers away from master
agreements to tracking by individual variance.

FHFA Oversight of Fannie Mae’s Use of Master Agreements

As conservator, FHFA oversees the authority delegated to Fannie Mae. During the review
period, Fannie Mae had general delegated authority from FHFA (as conservator) to enter
into master agreements, save for one exception. That exception required Fannie Mae to seek
conservator approval for material changes to master agreements for its top five single-family



                                  OIG • AUD-2019-006 • March 27, 2019                                             10
mortgage sellers. 8 FHFA, through its Division of Housing Mission and Goals (DHMG),
Office of Housing and Regulatory Policy, established guidelines for reviewing variances from
the Selling Guide.

As supervisor, FHFA conducts examination activities pursuant to its Examination Manual
modules and supplemental guidance. Depending on the scope and objective of the
examination activity, supervision of master agreements may incorporate relevant information
from FHFA’s Credit Risk Management module and Single-Family Mortgage Underwriting
and Acquisitions guidance.

                                                      *****

In this audit, we sought to assess FHFA’s oversight of Fannie Mae’s master agreements with
its single-family mortgage sellers during the review period. As part of the audit, we analyzed
master agreements for Fannie Mae’s top three single-family mortgage sellers to determine
whether there were variations between the terms in the master agreements for DTI ratio, LTV
ratio, credit score, and property valuation method from the terms for the same element in the
Selling Guide. We also obtained information from FHFA and Fannie Mae and analyzed loan-
level data from FHFA’s MLIS to: (1) determine the number of loans purchased by Fannie
Mae, pursuant to master agreements; (2) determine whether the credit terms for DTI ratio,
LTV ratio, credit score, 9 and property valuation method for the mortgages purchased by
Fannie Mae differed from those credit terms in the governing Selling Guide; and (3) identify
FHFA’s oversight activities of Fannie Mae’s single-family loan acquisitions outside of the
credit terms defined in the Selling Guide during the review period.




8
 FHFA imposed this requirement in its 2012 Letter of Instruction. FHFA uses its Letter of Instruction to
define and outline the scope of delegated and undelegated authorities. This requirement for FHFA approval
was removed in its revised Letter of Instruction, effective March 31, 2018. FHFA intends, in its role as
conservator, to rely on routine management reporting and surveillance activities to monitor master agreements.
9
  There are other underwriting standards and eligibility requirements used by Fannie Mae that were not
included in the scope of this audit, such as mortgage insurance, loan purpose, and borrower’s financial
reserves.



                                   OIG • AUD-2019-006 • March 27, 2019                                           11
FACTS AND ANALYSIS ...............................................................

For the Top Three Mortgage Sellers with Master Agreements, Testing Identified No
Variances in these Master Agreements from the Credit Terms of DTI Ratio, LTV Ratio,
Credit Score, or Property Valuation Method in the Selling Guide

Of the 6,455,307 single-family mortgages purchased by Fannie Mae during the review period,
approximately 1.6 million mortgages were purchased from its top three mortgage sellers,
pursuant to 155 master agreements and amendments. We tested whether any term in these 155
master agreements and amendments setting DTI ratio, LTV ratio, credit score, and property
valuation method was an exception to the DTI ratio, LTV ratio, credit score, and property
valuation method established in the Selling Guide.

Result of Test: Our review identified no variation between the terms in the master agreements
for DTI ratio, LTV ratio, credit score, and property valuation method from the terms for the
same element in the Selling Guide.

Testing Identified Very Few Mortgages Purchased by Fannie Mae that Had DTI Ratios,
LTV Ratios, Credit Scores, and/or Property Valuation Methods Outside Those
Permitted by the Selling Guide

We analyzed all Fannie Mae single-family acquisition data from FHFA’s MLIS to determine
whether Fannie Mae purchased loans with terms for DTI ratio, LTV ratio, credit score, and
property valuation method that were different from the terms in the Selling Guide.

Result of Test: We reviewed loan-level data for 6,455,307 mortgages and found that for those
mortgages where we identified differences with the credit terms for DTI ratio, LTV ratio,
credit score, and property valuation method allowed by the Selling Guide, we consider those
differences to be immaterial, accounting for less than one-tenth of one percent of total
mortgages purchased during the review period. Further, as of June 2018, 100% of those
mortgages purchased with differences in DTI ratio, 100% of those mortgages purchased with
differences in LTV ratio, 98.6% of those mortgages purchased with differences in credit
score, and 100% of those mortgages purchased with differences in property valuation were in
good standing.




                             OIG • AUD-2019-006 • March 27, 2019                                12
     Testing Identified Issues that Affected the Reliability of Certain Data Fields in FHFA’s
     MLIS

To perform this audit, we analyzed loan-level data housed in FHFA’s MLIS. 10 MLIS source
data for Fannie Mae is populated by data transmitted and uploaded from a number of Fannie
Mae systems. 11 When we tested Fannie Mae’s single-family acquisition data at the loan-level
using MLIS, we found instances where data fields for our selected credit terms were either
missing information or were shown as “unknown.” The largest instance that we identified in
this audit was with the data field related to whether or not there was an authorized property
valuation method (e.g., appraisal) present at the time of loan delivery. Specifically, we found
647,820 loans were coded as “unknown” in MLIS (representing more than 10% of the loans
reviewed). When we inquired as to the reasons for this, Fannie Mae officials told us that
based on their loan-level research, they determined that computer glitches had occurred when
data was transmitted from Fannie Mae’s systems and uploaded to MLIS. As a result of these
glitches, some data fields in MLIS either were not populated or were populated with the value
of “unknown.” Fannie Mae officials also advised that some of the data fields and/or allowable
values for those fields in MLIS did not directly correspond or map to the data fields in their
systems (Fannie Mae’s data fields are based on the Uniform Loan Delivery Dataset 12). This
also resulted in some missing data fields and/or unknown values in MLIS.

As part of this audit, we also inquired how FHFA ensures that MLIS data is complete and
accurate. FHFA officials told us that they performed validation testing on certain data fields
by comparing MLIS data with data in Fannie Mae’s 10-K, public use databases, and other
Enterprise reports. These officials reported that as a result of the validation testing, any
material exceptions would be addressed with the Enterprises. At the time of our audit, we
were informed that over 100 open material exceptions were under review by FHFA that
related to Fannie Mae. Examples of open material exceptions under review during the audit
period included: the loan delinquency field (missing loan numbers); the credit score field
(submitted source code was not an allowable value); and the special feature code for the
MyCommunity Mortgage, an affordable housing product (code was incomplete). We also
learned that during the time of our review FHFA did not consider a data field populated with
the value of “unknown” to be an exception (i.e., FHFA allowed/accepted “unknown” as a
valid value in MLIS). It was not until after we inquired about the “unknown” value for

10
  MLIS is an FHFA database that maintains loan-level data provided by the Enterprises. Among other things,
FHFA requires the Enterprises to attest that all data is checked for accuracy before submission.
11
  Examples of Fannie Mae systems used to populate the data in MLIS include SF RDW (Single-Family
Relational Data Warehouse), CLM ODS (Credit DataMart), and AMTM (Asset Mark-to-Market) to name a
few.
12
   The Uniform Loan Delivery Dataset is the common set of data elements that FHFA requires Fannie Mae
(and Freddie Mac) to use for single-family loan deliveries.



                                  OIG • AUD-2019-006 • March 27, 2019                                        13
authorized property valuation that FHFA identified the matter as a material exception with
Fannie Mae.

In technical comments to a draft of this report, FHFA maintained that “MLIS is designed to
support research and policy analyses and is not intended to have a level of accuracy necessary
to support audit activities (‘MLIS is not audit quality data’).” FHFA also stated that the
explanation for why information did not flow from the system of record to MLIS was
generally due to the design and intended use of the research data set, and not from computer
glitches. According to FHFA, “unknown” is accepted as a valid value in MLIS “in
recognition of the fact that a data element may not have been required historically or that a
data element may not be required for all programs or products.”

We do not challenge FHFA’s explanation for why “unknown” is a valid data value for some
elements at the loan-level, given the historical nature of the information and that not all data
elements apply to all loans. We are concerned, however, by FHFA’s insistence that MLIS is
not meant to “support audit activities” or does not have “audit quality data.” FHFA, by its
own admission, uses the MLIS data for critical purposes – research, policy analyses, and
official publications. Using MLIS data for these purposes requires a high level of data quality,
similar to the data quality required for audits. Because FHFA relies on the completeness and
accuracy of the data in MLIS, FHFA has adopted internal controls, including reporting
requirements for the Enterprises and attestations to the accuracy of the data before
submission, to ensure accuracy and completeness. For those reasons, we do not credit FHFA’s
claim that the data in MLIS is not sufficiently accurate to support audit activities.

FHFA also asserted, in its technical comments, that the reason why information did not flow
from the system of record to MLIS was unrelated to computer glitches. However, Fannie Mae
officials provided the explanation of computer glitches, based on Fannie Mae’s detailed
research. Because FHFA provided no basis in its technical comments to dispute Fannie Mae’s
explanation, we do not accept FHFA’s characterization.

Testing Identified Some Oversight by FHFA of Fannie Mae’s Use of Master Agreements

Three FHFA divisions execute oversight responsibilities for Fannie Mae’s use of master
agreements with single-family mortgage sellers. Set forth below is a snapshot of the oversight
offered by these divisions:

   Division of Conservatorship

The Division of Conservatorship (DOC) works to assist the FHFA Director in carrying out
the responsibilities as conservator, which includes taking the actions necessary to restore the
Enterprises to a sound and solvent condition, preserve and conserve their assets, and operate
them in the public interest. As conservator, FHFA has delegated authority for general


                              OIG • AUD-2019-006 • March 27, 2019                                  14
corporate governance and day-to-day matters to the Enterprises’ boards of directors and
executive management and retained authority for certain significant matters. All matters
requiring FHFA approval are to be submitted by the affected Enterprise through a
“conservatorship portal,” and FHFA internal policies and procedures task DOC with
responsibility for processing Enterprise requests for FHFA approval.

FHFA reported to us the oversight activities relating to Fannie Mae’s master agreements
conducted by DOC during the review period. Our testing confirmed that DOC performed the
following oversight activities. 13

      •    Analyzed Fannie Mae’s requests for approval of revisions to master agreements
           that require conservator action. As discussed, Fannie Mae was required during the
           review period to obtain FHFA approval for material changes to master agreements
           for its top five single-family mortgage sellers. We reviewed supporting records of the
           portal submissions made during the review period and found that DOC processed one
           request from Fannie Mae for a material change to a master agreement for one of its
           top five single-family mortgage sellers. Our review of FHFA records found that the
           Agency reviewed Fannie Mae’s supporting documentation and approved the request.

      •    Prepared reports on emerging risks. During the review period, DOC prepared
           regular reports on key and emerging risks. We reviewed a sample of these reports and
           noted that under emerging risks, one report cited a variance in a master agreement
           with a large single-family mortgage seller related to homeownership education
           requirements. 14

      Division of Housing Mission and Goals

Included in its portfolio of responsibilities, DHMG coordinates, reviews, and/or makes
recommendations on Enterprise requests for FHFA conservator approval and matters
submitted for conservator notice, including submissions involving single-family credit
policies and master agreements, and manages FHFA’s MLIS.

FHFA reported to us the oversight activities relating to Fannie Mae’s master agreements
conducted by DHMG during the review period. Our testing confirmed that DHMG performed
the following oversight activities. 15



13
     We take no position on the sufficiency of these activities.
14
     DOC issued a non-objection email to Fannie Mae in May 2017 related to this variance.
15
     We take no position on the sufficiency of these activities.




                                      OIG • AUD-2019-006 • March 27, 2019                           15
      •    Prepared Variances, Waivers, and Exceptions (VWE) Monitoring Dashboard
           reports. Beginning in June 2016, DHMG prepared monthly VWE Monitoring
           dashboard reports, which according to DHMG officials were used to track negotiated
           terms contractually executed through the Enterprise’s master agreements. 16 17 We
           reviewed the monthly dashboards prepared by DHMG based on monthly Fannie
           Mae data from June 2016 through December 2017 and found that they included
           information on active variances (including those in master agreements) and various
           credit terms, including two of our selected credit terms – LTV ratio and credit score.

      Division of Enterprise Regulation

Included in its portfolio of responsibilities, Division of Enterprise Regulation (DER) executes
FHFA’s obligations under HERA to supervise the Enterprises. According to FHFA, DER
examiners conduct risk-based supervisory activities, including ongoing monitoring and
targeted examinations, related to Fannie Mae. 18

DER conducted the following supervisory activities related to Fannie Mae’s master
agreements during the review period. 19

      •    2015 Examination Cycle. DER performed an ongoing monitoring activity of Fannie
           Mae titled “Single-Family Credit Risk Management,” which, among other things,
           addressed trends for variances in the master agreements. This activity identified no
           adverse examination findings.

      •    2016 Examination Cycle. DER performed an ongoing monitoring activity of Fannie
           Mae titled “Single-Family Mortgage Business,” which included a monthly review
           of Fannie Mae’s variance reports. This activity identified no adverse examination
           findings. Also, as part of its 2016 examination cycle, DER performed an ongoing

16
  During much of the review period, variances represented negotiated terms that were set forth in a master
agreement.
17
   DHMG did not prepare monthly VWE Monitoring Dashboard reports based on Fannie Mae data from
January 2015 through May 2016. Further, as OIG reported in December 2015, DHMG had failed to implement
a process for reviewing Enterprise variances. See OIG, Compliance Review of FHFA’s Implementation of Its
Procedures for Overseeing the Enterprises’ Single-Family Mortgage Underwriting Standards and Variances
(Dec. 17, 2015) (COM-2016-001). In an update to that 2015 report, based on a limited review period, we
reported in March 2018 that DHMG did follow its process for reviewing Enterprise variances. See OIG,
Update on FHFA’s Implementation of its Revised Procedures for Overseeing the Enterprises’ Single-Family
Mortgage Underwriting Standards and Variances (Mar. 27, 2018) (COM-2018-003).
18
   According to the FHFA Examination Manual, the purpose of ongoing monitoring is to analyze real-time
information and to use those analyses to identify Enterprise practices and changes in an Enterprise’s risk
profile that may warrant supervisory attention, while targeted examinations allow for a deep or comprehensive
assessment of the area under review.
19
     We take no position on the sufficiency of these activities.



                                      OIG • AUD-2019-006 • March 27, 2019                                       16
       monitoring activity to review Fannie Mae’s corrective actions to remediate adverse
       examination findings issued during the 2014 examination cycle. DER closed both
       findings.

   •   2017 Examination Cycle. DER started a targeted examination titled “Program
       Variances/High LTV Loan Programs.” In February 2019, FHFA completed the
       targeted examination.


FINDING ...................................................................................

Information about Purchased Mortgages Was Incomplete for Certain MLIS Data Fields

We found instances where MLIS data fields for our selected credit terms were either missing
information or were shown as “unknown.” For example, for the data field related to property
valuation method present at the time of loan delivery, we found 647,820 loans (more than
10% of the loans reviewed) were coded as “unknown.” We learned that, during our review
period, FHFA did not consider a data field populated with the value of “unknown” to be an
exception (i.e., FHFA allowed/accepted “unknown” as a valid value in MLIS). After we
inquired about the high instances of “unknown” values for property valuation method, FHFA
did act to identify the matter as a material exception with Fannie Mae. According to Fannie
Mae officials, computer glitches occurred when data was transmitted from Fannie Mae’s
systems and uploaded to MLIS. Further, some of the data fields and/or allowable values for
those fields in MLIS did not directly correspond or map to the data fields in Fannie Mae’s
systems, which resulted in data fields in MLIS that either were not populated or were
populated with the value of “unknown.”


CONCLUSION AND RECOMMENDATION ...................................

We found no variations between the terms in the master agreements for Fannie Mae’s single-
family mortgage sellers for DTI ratio, LTV ratio, credit score, and property valuation method
from the terms for the same element in Fannie Mae’s Selling Guide. For the single-family
mortgages that we analyzed at the loan-level, where we identified differences with the credit
terms for DTI ratio, LTV ratio, credit score, and property valuation method allowed by the
Selling Guide, we considered those differences to be immaterial for the review period
(accounting for less than one-tenth of one percent of total mortgages purchased during the
review period). From our testing, we found no evidence that Fannie Mae’s master agreements
with mortgage sellers were “opening the credit box” beyond its Selling Guide for the credit
terms we reviewed.


                             OIG • AUD-2019-006 • March 27, 2019                                17
We also identified examples of FHFA’s oversight of Fannie Mae’s use of master agreements,
but take no position on the sufficiency of those activities.

In our tests using Fannie Mae’s single-family acquisition data in MLIS, we found instances
where data fields for our selected credit terms either were missing information or were shown
as “unknown.” The largest instance – more than 10 percent of single-family acquisitions
during our review period – was the MLIS data field for the property valuation method. When
we inquired as to the reasons, Fannie Mae advised that “computer glitches” caused some data
fields in MLIS either not to be populated or to be populated with the value of “unknown.” We
make one recommendation to FHFA to address this issue.

Fannie Mae also advised that some of the data fields and/or allowable values for those fields
in MLIS do not directly correspond or map to the data fields in their systems. The scope of
our audit did not include a review into these matters. However, we urge FHFA to continue to
work with the Enterprise to ensure that the Enterprise’s information for all MLIS data fields is
complete and accurate.

We recommend that FHFA:

   1. Work with Fannie Mae to resolve the high instance (over 10 percent of loans during
      our review period) of the property valuation method being reported as “unknown” in
      MLIS.


FHFA COMMENTS AND OIG RESPONSE .....................................

We provided FHFA an opportunity to respond to a draft of this audit report. FHFA provided
technical comments on the draft report and those comments were considered in finalizing this
report. FHFA also provided a management response, which is included in the Appendix to
this report. In its response, FHFA agreed with the recommendation and stated that it will
continue to discuss with Fannie Mae the possibility of improving the reporting of the property
valuation method. FHFA will provide a status update on those discussions to OIG by June 1,
2019. We consider FHFA’s planned corrective action responsive to our recommendation.




                              OIG • AUD-2019-006 • March 27, 2019                                  18
OBJECTIVE, SCOPE, AND METHODOLOGY .................................

In this audit, we sought to assess FHFA’s oversight of Fannie Mae’s master agreements with
its single-family mortgage sellers during the review period. As part of the audit, we analyzed
master agreements for Fannie Mae’s top three single-family mortgage sellers to determine
whether there were variations between the terms in the master agreements for DTI ratio, LTV
ratio, credit score, and property valuation method from the terms for the same element in the
Selling Guide. We also obtained information from FHFA and Fannie Mae and analyzed loan-
level data from FHFA’s MLIS to: (1) determine the number of loans purchased by Fannie
Mae, pursuant to master agreements; (2) determine whether the credit terms for DTI ratio,
LTV ratio, credit score, and property valuation method for the mortgages purchased by Fannie
Mae differed from those credit terms in the governing Selling Guide; and (3) identify FHFA’s
oversight activities of Fannie Mae’s single-family loan acquisitions outside of the credit terms
defined in the Selling Guide during the review period.

To address our objectives, we:

   •   Researched applicable laws, regulations, policies, and procedures applicable to
       FHFA’s oversight of Fannie Mae’s single-family loan acquisitions.

   •   Obtained and reviewed Fannie Mae’s Selling Guide to determine the requirements for
       the credit terms of LTV ratio, DTI ratio, credit score, and property valuation method.

   •   Obtained and reviewed Fannie Mae’s master agreements with its top three single-
       family mortgage sellers for 2015 through 2017 to determine if there were any
       variances related to DTI ratio, LTV ratio, credit score, and property valuation method
       that were outside of the credit terms defined in the Selling Guide.

   •   Determined whether FHFA performed oversight activities related to Fannie Mae’s
       master agreements with its single-family mortgage sellers from 2015 through 2017.
       This included obtaining an understanding of the oversight activities conducted during
       the audit period by interviewing FHFA officials and reviewing documents related to
       FHFA’s oversight activities.

   •   Obtained and analyzed 2015 through 2017 loan-level acquisition data provided by
       FHFA from FHFA’s MLIS to determine whether Fannie Mae purchased loans outside
       of the credit terms defined in the Selling Guide for DTI ratio, LTV ratio, credit score,
       and property valuation method.

   •   Assessed the reliability of data received for this audit as determined necessary by
       (1) corroborating the information with other source data, (2) interviewing FHFA


                              OIG • AUD-2019-006 • March 27, 2019                                  19
       officials regarding data reliability tests performed, and (3) reviewing documentation
       related to FHFA’s data reliability tests.

We conducted this performance audit between April 2018 and March 2019 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 objectives. We believe that the evidence
obtained provides a reasonable basis for our findings and conclusions based on our audit
objectives.




                             OIG • AUD-2019-006 • March 27, 2019                                 20
APPENDIX: FHFA MANAGEMENT RESPONSE .............................




                   OIG • AUD-2019-006 • March 27, 2019        21
ADDITIONAL INFORMATION AND COPIES .................................


For additional copies of this report:

   •   Call: 202-730-0880

   •   Fax: 202-318-0239

   •   Visit: www.fhfaoig.gov



To report potential fraud, waste, abuse, mismanagement, or any other kind of criminal or
noncriminal misconduct relative to FHFA’s programs or operations:

   •   Call: 1-800-793-7724

   •   Fax: 202-318-0358

   •   Visit: www.fhfaoig.gov/ReportFraud

   •   Write:

                FHFA Office of Inspector General
                Attn: Office of Investigations – Hotline
                400 Seventh Street SW
                Washington, DC 20219




                               OIG • AUD-2019-006 • March 27, 2019                         22