[Senate Hearing 113-221]
[From the U.S. Government Publishing Office]
S. Hrg. 113-221
HOUSING FINANCE REFORM: FUNDAMENTALS OF
TRANSFERRING CRED IT RISK IN A FUTURE
HOUSING FINANCE SYSTEM
=======================================================================
HEARING
before the
COMMITTEE ON
BANKING,HOUSING,AND URBAN AFFAIRS
UNITED STATES SENATE
ONE HUNDRED THIRTEENTH CONGRESS
FIRST SESSION
ON
EXAMINING THE CONCEPT OF RISK SHARING CONSIDERED IN THE CONTEXT OF THE
HOUSING FINANCE MARKET AND THE RISK TRANSFER TRANSACTIONS ENTERED INTO
BY FANNIE MAE AND FREDDIE MAC
__________
DECEMBER 10, 2013
__________
Printed for the use of the Committee on Banking, Housing, and Urban
Affairs
Available at: http: //www.fdsys.gov /
U.S. GOVERNMENT PRINTING OFFICE
87-318 WASHINGTON : 2014
-----------------------------------------------------------------------
For sale by the Superintendent of Documents, U.S. Government Printing
Office Internet: bookstore.gpo.gov Phone: toll free (866) 512-1800; DC
area (202) 512-1800 Fax: (202) 512-2104 Mail: Stop IDCC, Washington, DC
20402-0001
COMMITTEE ON BANKING, HOUSING, AND URBAN AFFAIRS
TIM JOHNSON, South Dakota, Chairman
JACK REED, Rhode Island MIKE CRAPO, Idaho
CHARLES E. SCHUMER, New York RICHARD C. SHELBY, Alabama
ROBERT MENENDEZ, New Jersey BOB CORKER, Tennessee
SHERROD BROWN, Ohio DAVID VITTER, Louisiana
JON TESTER, Montana MIKE JOHANNS, Nebraska
MARK R. WARNER, Virginia PATRICK J. TOOMEY, Pennsylvania
JEFF MERKLEY, Oregon MARK KIRK, Illinois
KAY HAGAN, North Carolina JERRY MORAN, Kansas
JOE MANCHIN III, West Virginia TOM COBURN, Oklahoma
ELIZABETH WARREN, Massachusetts DEAN HELLER, Nevada
HEIDI HEITKAMP, North Dakota
Charles Yi, Staff Director
Gregg Richard, Republican Staff Director
Laura Swanson, Deputy Staff Director
Glen Sears, Deputy Policy Director
Erin Barry Fuher, Professional Staff Member
Kari Johnson, Legislative Assistant
Greg Dean, Republican Chief Counsel
Chad Davis, Republican Professional Staff Member
Hope Jarkowski, Republican SEC Detailee
Dawn Ratliff, Chief Clerk
Taylor Reed, Hearing Clerk
Shelvin Simmons, IT Director
Jim Crowell, Editor
(ii)
?
C O N T E N T S
----------
TUESDAY, DECEMBER 10, 2013
Page
Opening statement of Chairman Johnson............................ 1
Opening statements, comments, or prepared statements of:
Senator Crapo................................................ 2
WITNESSES
Kevin Palmer, Vice President, Strategic Credit Costing and
Structuring, Freddie Mac....................................... 4
Prepared statement........................................... 22
Laurel Davis, Vice President for Credit Risk Transfer, Fannie Mae 5
Prepared statement........................................... 26
Responses to written questions of:
Senator Crapo............................................ 47
Ted Durant, Vice President of Analytic Services, Mortgage
Guaranty Insurance Corporation................................. 7
Prepared statement........................................... 32
Sandeep Bordia, Head of Residential and Commercial Credit
Strategy, Barclays Capital..................................... 9
Prepared statement........................................... 40
Wanda DeLeo, Deputy Director, Division of Conservatorship,
Federal Housing Finance Agency
Prepared statement........................................... 43
Responses to written questions of:
Senator Reed............................................. 47
Additional Material Supplied for the Record
Prepared statement of Laurie S. Goodman, Director, Housing
Finance Policy Center, The Urban Institute..................... 50
(iii)
HOUSING FINANCE REFORM: FUNDAMENTALS OF TRANSFERRING CREDIT RISK IN A
FUTURE HOUSING FINANCE SYSTEM
----------
TUESDAY, DECEMBER 10, 2013
U.S. Senate,
Committee on Banking, Housing, and Urban Affairs,
Washington, DC.
The Committee met at 2:38 p.m., in room SD-538, Dirksen
Senate Office Building, Hon. Tim Johnson, Chairman of the
Committee, presiding.
OPENING STATEMENT OF CHAIRMAN TIM JOHNSON
Chairman Johnson. I call this hearing to order.
This hearing is the last in our in-depth series of policy
hearings to explore the benefits and challenges of reforming
the housing finance system. At the urging of Committee Members
and with the help of witnesses representing stakeholders across
the industry, consumer, and policy spectrum, Ranking Member
Crapo and I established an aggressive hearing schedule focusing
on what we agreed were essential pieces for consideration in a
new housing finance system. I would like to thank Ranking
Member Crapo and his staff for their good partnership and
coordination in undertaking this complicated effort. I would
also like to thank all the witnesses that have participated.
There are numerous players in the housing finance system that
have structured their businesses and household decisions around
the current system contributing to nearly 20 percent of the
economy. As we draft changes to the system, we must keep that
in mind.
Witnesses in previous hearings overwhelmingly agreed that
any new housing finance system should include an explicit
Government guarantee with private capital taking the first-loss
position. However, as we explored during the hearing regarding
the transition, we must be certain that there will be enough
interest from private capital to prevent the reduction of
liquidity and mortgage credit. Today's hearing will examine
several ways that private capital can take on additional credit
risk in front of a Government guarantee and the features that
are required to attract private capital.
When considering options for expanding private capital's
role in the secondary mortgage market, I believe the structures
should be compatible with the TBA market in order to ensure
continued availability of the 30-year fixed-rate mortgage. It
is not clear that certain transactions envisioned by S.1217 are
TBA compatible, but the recent credit risk transfer
transactions conducted by Fannie Mae and Freddie Mac may
provide one blueprint that is. I look forward to learning more
about those transactions today and how we can improve on S.1217
in this respect.
I am also concerned that in a system with multiple private
capital structures as options, not all forms of private capital
will provide equal protection for taxpayers. As we saw during
the recent housing boom and subsequent crisis, private capital
will participate during a boom and flee in a downturn. If we
are going to construct a new system that is even more dependent
on private capital, we must work to ensure that private capital
is stable and appropriately allocated and that any new
structure functions well so that responsible homebuyers are not
priced out of the market.
Last, for any future system to function and maintain the
trust of consumers, investors, and taxpayers, there needs to be
clear guidelines regarding how loan modification requests are
evaluated and how principal and interest are paid to investors
in the event of a modification or a borrower default. I would
like to thank our witnesses for being here today, and I look
forward to your thoughts regarding how the different private
capital structures could perform in various economic
circumstances and what a new system would need to be able to
maintain the TBA market.
With that, I will turn to Senator Crapo.
STATEMENT OF SENATOR MIKE CRAPO
Senator Crapo. Thank you very much, Mr. Chairman. We made
it to our last hearing, at least to this point, right?
As the Committee continues to consider broad reform of the
housing finance system, I am encouraged by the progress that we
have made over the last few months. Since August, the Committee
has held nine hearings on specific components that could make
up a new housing finance system, and I appreciate the
Chairman's initiative and perseverance and your thoughtfulness
in working through these complex matters. I remain strongly
committed to working with the Chairman and all of my colleagues
toward a bipartisan solution in the near future.
During a prior hearing, the Committee examined the
construct for a potential Government guarantee of certain
mortgage-backed securities. Today the Committee will take a
closer look into the mechanics of allocating private capital
ahead of such a properly tailored Government guarantee.
As in prior hearings, I am going to reinforce that if we
consider housing reform options that include a Government
guarantee, we must ensure that there is robust private capital
taking the first-loss position. We must also ensure that the
first-loss positions are in front of mortgages with high-
quality underwriting.
If there is not proper underwriting and allocation of
private capital ahead of the Government guarantee, we could
find ourselves in another taxpayer bailout scenario. That would
be unacceptable.
I welcome the perspective of our witnesses on the potential
risk transfer options that could be used to attract private
capital to this first-loss position, including the benefits and
tradeoffs of each option. In particular, I am interested to
hear your views on which risk transfer mechanisms could bring
in an appropriate amount of private capital while still
prioritizing taxpayer protection to the fullest extent
possible.
S.1217 contemplates several mechanisms for the security
level risk transfer: bond guarantors, one of two capital
markets approaches, as well as any risk-sharing agreements
undertaken by the Federal Housing Finance Agency.
The bill provides that bond guarantors would be approved by
the Government to hold a 10-percent first-loss position against
the covered mortgage-backed securities which they guarantee. To
ensure that we have responsible, sustainable mortgage-backed
securities products, the bond guarantor should stand behind 100
percent of the principal value of the security. This means that
the bond guarantor must exhaust all of its financial resources
and become insolvent prior to the triggering of a Government
guarantee. How, if at all, should legislation address allowing
a bond guarantor to engage in risk-sharing transactions for the
risk it takes on in absorbing the 10 percent?
With respect to capital market transactions, one option is
the creation of a senior-subordinated first-loss piece. Under
this capital markets approach, S.1217 provides that the
Government backstop would attach once the 10-percent
subordinated piece is exhausted. Some have expressed concern
regarding how the senior-subordinated structure contemplated in
S.1217 would interact with the current TBA market structure.
Could this senior-subordinated structure become TBA deliverable
if both pieces are made up of standardized loans?
A second capital markets option contemplated in S.1217 is a
credit-linked note structure which allows investors to receive
payment based on a pool's losses. Are there tradeoffs to the
credit-linked note structure that could make it more or less
attractive to private investors?
I look forward to the witnesses' views on whether these
three options could coexist in the marketplace as well as other
options that could be attractive to private sector capital. The
FHFA has started the experiment with risk transfer mechanisms
this year. The Freddie Mac STACR deal and Fannie Mae's NMI and
C-Deals are a starting point for gauging investor appetite in
taking the first-loss position. Can these deals be replicated
in the future? How quickly and to what extent can these
transactions be developed? What lessons can we learn as we try
to duplicate these risk-sharing transactions?
I am also interested to hear from the experts before us
today on how future risk-sharing transactions could work with a
common utility for the securitization of covered mortgage-
backed securities. The hearings we have held over the past
several months have allowed us to gather all relevant
viewpoints and develop a strong, factual record.
Mr. Chairman, again I thank you for your perseverance in
moving ahead aggressively and for your leadership as we move
forward on this housing reform package.
Chairman Johnson. Thank you, Senator Crapo.
Are there any other Members who would like to give a brief
opening statement?
[No response.]
Chairman Johnson. I would like to remind my colleagues that
the record will be open for the next 7 days for additional
statements and other materials.
Before we begin, I would like to introduce our witnesses
that are here with us today:
First, Mr. Kevin Palmer, Vice President at Freddie Mac;
Ms. Laurel Davis, Vice President at Fannie Mae;
Mr. Ted Durant, Vice President of Analytic Services at the
Mortgage Guaranty Insurance Corporation;
And Mr. Sandeep Bordia, head of Residential and Commercial
Credit Strategy at Barclays Capital.
We welcome all of you here today and thank you for your
time. I would also note that Ms. Wanda DeLeo was scheduled to
testify today, but because of the weather is unable to attend.
Ms. DeLeo's testimony will be submitted for the record, and she
has agreed to answer questions for the record.
Chairman Johnson. Mr. Palmer, you may proceed.
STATEMENT OF KEVIN PALMER, VICE PRESIDENT, STRATEGIC CREDIT
COSTING AND STRUCTURING, FREDDIE MAC
Mr. Palmer. Well, thank you. Chairman Johnson, Ranking
Member Crapo, and Members of the Committee, thank you for
inviting me to appear today. I am Kevin Palmer. I head the
business unit that developed and executed Freddie Mac's credit
risk sharing transactions.
More than 2 years ago, Freddie Mac began looking at ways to
build tools to transfer credit risk from our books to private
investors. Given the size of our guarantee books, we needed to
create tools that could be scalable and sustainable. We also
believe it is important to have multiple types of structures to
provide flexibility to transfer risk at the lowest cost and
with the least amount of disruption to the mortgage market.
So we studied a variety of risk transfer options and
structures, looking carefully at both their economics and
whether they could be made to work operationally. This helped
us prepare for FHFA's credit risk sharing directives in its
2012 and 2013 scorecards.
In 2013, we executed two different risk transfer structures
in three transactions. We continue to explore other structures
that meet our overall program objectives. At a high level,
these objectives are: first, to reduce taxpayer exposure to
credit risk of our mortgage purchases; second, bring new credit
investors into the mortgage market; third, create innovative
products that are both expandable and attractive over time;
and, finally, preserve the cost efficiencies of the to-be-
announced, or TBA, market.
I am pleased to report that we have met these objectives.
Our first two transactions were offerings of structured agency
credit risk debt notes. We call them ``STACR.'' We transferred
a portion of the credit risk from more than 200,000 single-
family mortgages we recently purchases. Payments to investors
in STACR notes are determined by the performance of this group
of mortgages. We retained some of the risk exposure as well as
the first-loss position. This assured investors that our
interests and theirs are aligned. Retaining the first-loss
position also helped investors get comfortable with STACR and
helped make the structure more economically attractive.
Freddie Mac retained control of the servicing of the loans.
This ensures we can maintain high servicing standards and
provide strong loss mitigation support to at-risk borrowers. We
are pleased by the market response to our STACR offerings.
Diverse groups of about 50 investors participated in each
transaction. We announced our third risk-sharing transaction in
November: an insurance policy underwritten by a large global
reinsurance company. We see a lot of potential for these types
of deals with reinsurance companies.
Going forward, we plan on doing more STACR and reinsurance
deals. We are also looking at other structures for risk
sharing, including credit-linked notes, recourse agreements
with lenders, and senior-subordinate structures.
In summary, we are very encouraged by the strong investor
interest in our risk-sharing offerings. We believe we can
create products that are scalable and attractive to investors.
Let me conclude with three key lessons that should be
helpful to you as you consider these issues.
First, we do not yet know how much mortgage credit risk
sharing investors are willing to do in the long term,
particularly during market and economic downturns. Our
offerings were successful, in large part because conditions
today are highly favorable for investment in mortgage credit
risk. House prices are generally appreciating, and credit
quality is high by historical standards. But we also have seen
in recent history that investors can and will leave the
mortgage market when conditions are less favorable.
Second, we have encountered a number of regulatory, tax,
and accounting issues that either affected the interest of
certain investors in our offerings or influenced how we
structured those offerings. I detail these in my written
statement. We believe Congress should consider these and other
similar issues as it determines the role that credit risk
sharing instruments can play in any future housing system.
Finally, risk-sharing transactions should continue to be
designed to be compatible with the forward mortgage market,
which provides key benefits to both borrowers and lenders.
Thank you again for this opportunity to appear today, and I
look forward to any questions.
Chairman Johnson. Thank you.
Ms. Davis, you may proceed.
STATEMENT OF LAUREL DAVIS, VICE PRESIDENT FOR CREDIT RISK
TRANSFER, FANNIE MAE
Ms. Davis. Chairman Johnson, Ranking Member Crapo, and
Members of the Committee, thank you for the opportunity to
testify today. My name is Laurel Davis, and I am the Vice
President for Credit Risk Transfer at Fannie Mae. My testimony
today will address how we manage our credit risk and our recent
credit risk transfer transactions.
Fannie Mae assumes credit risk on the loans in the
securities we guarantee. That means if a loan we guarantee goes
into default or foreclosure, Fannie Mae takes the resulting
losses. Since the onset of the housing crisis, Fannie Mae has
made numerous improvements to how we analyze our credit risk
and to our underwriting and servicing standards to reduce such
risk. We also made significant improvements to how we monitor
the origination processes of our lender customers.
As we have upgraded our credit policies and monitoring, the
performance of loans that we acquire has improved dramatically.
Loans originated since 2009 have been performing well and are
approximately 75 percent of our total book of business today.
As a result of our efforts and an improved housing market, we
have recorded seven straight quarters of profit and, as of
December 31st, we will have paid almost $114 billion in
dividends to the Treasury Department versus total draws of
approximately $116 billion.
Prior to bringing our credit risk transfer transactions to
market, we spent significant time with investors and mortgage
insurance, or MI, companies to educate them on our credit and
servicing policies. A strong understanding of these policies
was important to the success of these transactions. This was
important because investors understood that Fannie Mae was
acting as their counterparty and as an intermediary between
originators with whom we do business and investors in the
securities. This intermediary role makes these types of
transactions possible. It allowed the 77 investors in the
securities to rely on our infrastructure and standards rather
than understanding the standards of more than 1,000 lenders
with whom we do business. Fannie Mae also serves as an ongoing
and active credit risk manager on behalf of itself and the
investors.
We have brought two credit risk transfer transactions to
market in the second half of this year. One was a securities
transaction and the other a mortgage insurance contract to
provide deeper MI coverage. These transactions involved the
sale of mezzanine risk and sought to reduce our exposure to
unexpected losses on loans we have acquired. They also provide
an additional avenue for private capital to enter the mortgage
market.
We structured both transactions to meet a number of goals,
including having no impact on how a loan is serviced, to ensure
that borrowers have access to the full range of refinance and
foreclosure prevention options, to require no changes to lender
operations, and to preserve a TBA execution. The TBA market is
a well-functioning and liquid market that allows borrowers to
lock in mortgage rates in advance and originators to hedge the
associated risk.
The securities transaction was the first offering in our
Connecticut Avenue Securities Series, or our C-Deals. In this
transaction, investors purchased a portion of credit risk on a
reference pool of recently originated loans. The reference pool
in this first C-Deal transaction included approximately 112,000
loans totaling $27 billion in unpaid principal balance, which
were acquired in the third quarter of 2012. These loans
accounted for approximately 12 percent of the acquisitions in
that quarter.
As the loans in the reference pool are paid off by
homeowners, Fannie Mae will pay down the principal balance of
the C-Deal securities. If there are credit events on loans in
the reference pool such as default or foreclosure, investors
may experience losses in their investment.
Fannie Mae retained a first-loss portion in an amount that
represents a multiple of our expected losses. We sold two
tranches of mezzanine risk and retained the senior risk in the
structure. We also retained a portion of the mezzanine risk in
an effort, at least initially, to keep additional skin in the
game.
As I noted earlier, we also entered into a mortgage
insurance agreement with National Mortgage Insurance, or NMI.
This was a back-end MI contract on a group of loans that
totaled approximately $5 billion in unpaid principal balance.
The loans included in this transaction had loan-to-value
ratios, or LTVs, of 70 to 80 percent. In this transaction,
Fannie Mae's liability has been reduced to 50-percent LTV on
the covered loans. Under the contract, Fannie Mae will take the
first loss up to 20 basis point or $10.3 million. This amount
is over two times our expected losses on the covered loans. NMI
will be liable for the next $90 million in losses, and Fannie
Mae will then be responsible for any additional losses on the
loans. The first and mezzanine risk pieces were set at a level
that exceeds our projected losses in a stress scenario similar
to the 2006-12 timeframe.
We are very pleased with the reaction to these transactions
from market participants. We are currently working on bringing
another C-Deal transaction to market in January, and we will
continue to look for opportunities to execute other risk-
sharing transactions.
Thank you again for this opportunity to testify before the
Committee, and I look forward to answering any questions you
may have.
Chairman Johnson. Thank you.
Mr. Durant, you may proceed.
STATEMENT OF TED DURANT, VICE PRESIDENT OF ANALYTIC SERVICES,
MORTGAGE GUARANTY INSURANCE CORPORATION
Mr. Durant. Thank you, Chairman Johnson, Ranking Member
Crapo, and Members of the Committee. MGIC created the modern
private mortgage insurance industry in 1957, and we created the
modern bond insurance industry in 1971. We have experienced the
best and the worst of times in housing finance, and we hope
that our experience will be helpful as you contemplate reform.
We believe that having survived the recent housing crisis
and saved taxpayers $40 billion, the private mortgage insurance
model has proven its value and earned its inclusion in S.1217
as a fundamental component of a new housing finance system. We
would like to discuss today how private MI integrates with
other risk transfer mechanisms to create a housing finance
system in which mortgages are both affordable and widely
available through the cycle and taxpayer risk is truly remote.
S.1217 identifies several potential credit risk transfer
mechanisms and the entities that might provide them. However,
we suggest improvement in five areas.
First, legislation should clarify the roles and the
responsibilities of the bond guarantors and the mortgage
insurers. Bond guarantors should operate at a remote risk
level, guaranteeing the timely payment of principal and
interest to bond holders should the issuer fail. Mortgage
insurance operates primarily at the loan level, taking a first-
loss position on higher-risk loans and, importantly, providing
critical oversight of the underwriting and servicing of those
loans.
In addition, mortgage insurers offer pool insurance to
provide a cost-effective means of filling any gaps between
loan-level insurance and bond insurance. Mortgage insurance and
bond insurance should be kept distinct from each other and
viewed as complementary not competing forms of credit risk
transfer. As long as the appropriate distinctions and rules are
maintained, we expect investors to find both businesses
attractive.
Second, legislation should recognize the private capital
benefit that is provided by mortgage insurance. A subordinated
securitization structure provides a deceptively simple measure
of private sector capital at risk. Insurers provide substantial
loss-absorbing resources with a combination of capital,
reserves, and renewal premium, or guarantee fees. The formula
for counting up the private sector capital at risk must include
the resources provided by primary and pool MI.
Third, legislation should establish a clear preference for
using properly regulated and capitalized entities which are
dedicated to the business of identifying, assuming, and
managing credit risk through the economic cycles. The illusion
of a ``best execution'' cost advantage of structured
transactions is, in reality, the mechanism that creates the
boom-bust cycle, providing too much credit in a boom and no
credit in a bust.
To ensure a stable source of capital and proper regulation
of mortgage and bond insurers, we recommend the continued use
of State insurance regulators, reserving for FMIC the role of
counterparty risk management and monitoring. Regulation of bond
insurers or mortgage insurers by FMIC raises significant
Federal-State questions, adds further complexity to the
management of FMIC, and concentrates oversight in a single
point of failure.
Fourth, legislation should clarify the manner in which the
Government guarantee is implemented. Stepping in when a
subordinated tranche expires and stepping in when a guarantor
fails are very different. Individual securities are far more
likely to reach a specified loss level than thousands of
securities guaranteed by an entity. Stepping in at
predetermined loss level, for example, on a vintage basis will
create the optics of a bailout, but it will also create a
greater likelihood that the guarantors are able to continue in
business through a crisis. Such an approach, however, must take
into consideration both mortgage insurers and bond guarantors.
Fifth, legislation must comprehensively address housing
finance reform. We at MGIC are less concerned about the
absolute level of the capital requirement than we are about
unequal capital requirements that favor one form of financing
over another. Setting higher capital requirements and
attachment points makes the taxpayer risk more remote, but it
also translates into higher costs for borrowers and increased
motivation for lenders to fund loans through alternative
channels.
The current system divides the mortgage world into
Government and conventional lending silos. Reform efforts in
one silo merely encourage the shift of loan production into the
other silo. Reform should provide for consistent, uniform rules
that apply regardless of the source of funding for the loans.
Thank you for the opportunity to testify today. I look
forward to your questions.
Chairman Johnson. Thank you.
Mr. Bordia, you may proceed.
STATEMENT OF SANDEEP BORDIA, HEAD OF RESIDENTIAL AND COMMERCIAL
CREDIT STRATEGY, BARCLAYS CAPITAL
Mr. Bordia. Good afternoon, Chairman Johnson, Ranking
Member Crapo, and other Members of the Committee. My group at
Barclays covers research on mortgage markets, including
research on housing finance. I appreciate the opportunity to
discuss the fundamentals of transferring credit risk from the
U.S. taxpayer to the private markets.
To begin with, let me talk briefly about the credit-linked
deals recently sold by the GSEs. So far, three deals have been
priced. In each of these deals, the GSEs retained the risk in
the first 0.3-percent loss position and sold the credit risk on
the 0.3- to 3-percent loss position.
The structures were well received by the market, with
several dozen investors participating. To be successful, any
solution used to transfer mortgage credit risk to the private
market should have certain basic features. The solution should
preserve the well-functioning TBA market, should be simple, and
use existing financial technology. The credit-linked note
structure satisfies all these three conditions.
The initial response from the market is also positive.
However, a few more things need to happen for this program to
be successful in the long run.
Number one, since there are fixed costs for investors to
set up systems to analyze and track performance of these deals,
we need to feel assured that these deals are not one-offs and
the program is here to stay. So issuance should be programmatic
with limited experimentation in structures.
Number two, expanding the type of collateral on which
credit risk is sold is critical. The initial deals covered only
the cleanest portion of GSE originations that is not fully
representative of the collateral quality that any future entity
would be expected to guarantee over time.
In terms of the market appetite to absorb the risk, while
the initial three deals have been oversubscribed, the amount of
credit risk sold so far is minuscule in comparison to what the
GSEs have on their guarantee books. I believe that the market
can absorb $5 to $10 billion of such supply next year and even
greater numbers in later years. However, for the program to get
to a stage where it can absorb much of the mortgage credit risk
with GSEs, it would realistically take several years.
As for the attachment point for the Government entity, it
is a function of the policy goal and also the collateral
quality on which the credit risk is being sold. The attachment
point would be higher if the policy goal is to prevent taxpayer
losses even in extreme draconian scenarios. And it should also
be higher for worse credit pools. In my view, a constant
attachment point for all kinds of collateral, as discussed in
Senate bill 1217, may not be most optimal.
There are two other approaches that are being considered
for transferring credit risk from a Government-sponsored
entity. The first is to use a securitization style senior-sub
structure, and the second is to use well-capitalized bond
guarantors to cover losses.
A senior-sub structure is less preferable to a credit-
linked note such as it would entail a difficult transition from
a well-functioning TBA market. A senior-sub structure could
also increase the warehousing costs for originators, which
would be particularly problematic for smaller originators.
The other option is to use bond guarantors as providers of
first loss. On the positive side, a bond guarantor solution
would likely provide more stable funding for mortgage credit
than other options. However, the bond guarantor structure also
has two major drawbacks.
Number one, this form of insurance may result in some
counterparty credit risk to the taxpayer. The STACR/CAS deals
provide the GSEs with up-front cash available to protect
taxpayers from losses. Bond guarantors would not have to pay
anything up front and, as such, may not work as a safeguard
under certain conditions.
The second drawback is that the bond guarantee structures
would be less transparent and provide slower market feedback
than credit-linked notes.
In conclusion, while we favor the credit-linked structure,
given the size of credit risk transfer required over the long
run, it might be preferable to have multiple exit options
including through bond guarantors. We would caution policy
makers to closely watch the pace of any big transition.
Chairman Johnson, Ranking Member Crapo, and other Members
of the Committee, I thank you for your time and attention and
the opportunity to testify before the Committee. I look forward
to answering any questions you may have.
Chairman Johnson. Thank you. Thank you all for your
testimony.
As we begin questions, I will ask the clerk to put 5
minutes on the clock for each Member.
Ms. Davis, what elements of your risk-sharing were most
important to preserve the TBA market? Were there structures
that you considered but eliminated because they were
incompatible with the TBA market?
Ms. Davis. Thank you. Yes, we have been working on this
looking at various alternatives for credit risk transfer over
the past year and a half, almost 2 years, and we did look at a
variety of different structures as we decided what to do.
We considered a cash senior-sub style deal as well as the
credit-linked note. Ultimately, at least initially, we decided
against the cash senior-sub deal because of the very point that
you make. It would not have been compatible with the TBA
market.
With the credit-linked note, we can allow the loans to go
into TBA. That market provides the funding for the loans so
that the interest rate risk is transferred off to investors.
The loans are funded through a very deep, liquid market. We can
then sell the credit risk to a different set of investors, and
it does tend to be a different set of investors who are
interested in purchasing the interest rate risk versus the
credit risk on the loans. And so by using the credit-linked
note structure, we are able to keep that front end of the
market functioning. Borrowers can lock rates. Lenders can hedge
their pipeline. And yet we are able to still tap into the
credit markets and offload the credit risk. So that is why we
chose this particular structure.
Chairman Johnson. Mr. Palmer, the risk-sharing deals that
have been executed rely on large reference pools of loans with
at least 20-percent downpayments and 9 months of seasoning.
Also, the enterprises retain both the first-loss and a vertical
strip of the risk exposure.
Should legislation include flexibility regarding the
features of risk-sharing deals in a new system?
Mr. Palmer. Thank you. Yes, it is important, I think as
indicated in my oral statement, to have flexibility in risk-
sharing transactions. As you said, the risk-sharing
transactions that we have done initially were focused on 60 to
80 LTV loans. I think the focus in 2013 was building structures
where you could transfer credit risk, and using those same
structures, we now have an ability to transfer, you know,
different types of credit risk, which will be a focus going
forward.
Flexibility on the types of structures is important. We are
in a certain unique environment today with very good
underwriting standards, with home prices appreciating, and what
works today may not work or be optimal in other types of
environments, and retaining that flexibility we believe is very
important.
Chairman Johnson. Mr. Bordia, your testimony suggests that
capital levels proposed in S.1217 may be difficult to achieve.
What are the risks of setting capital at such high levels?
Mr. Bordia. I think there are a couple of issues that we
are trying to resolve when we have a 10-percent capital
assigned for all different kinds of collateral types.
Number one, if you are looking at really good quality
collateral, then the expected losses might only be 20, 30, or
40 basis points, and at that point, if you want the entity who
is insuring those mortgages to retain 10-percent capital, then
the return on that particular capital is going to be really
small.
So I think in some ways we are trying to strike a balance
between protecting the taxpayer, and if you want to do that,
then you want to have a lot of capital cushion. But at the same
time, it will also lead to lower returns and, therefore, it is
going to be harder to raise capital, that amount of capital.
So just to give an example, if you look at the total size
of the mortgages which the GSEs are guaranteeing today, it is
around $4.5 to $5 trillion. A 10-percent requirement on that
would roughly be around $400 to $500 billion in capital, which
is a very, very big number. If you look at the total amount of
money that was raised in IPOs in the U.S. this year, it was
around $50 billion. If you look at just financial services
firms, it was around $10 billion.
So, yes, I think it is certainly possible over time that
you create a market where you have a significant amount of
capital that has been raised. It is a very, very difficult
proposition.
Chairman Johnson. Mr. Durant, when considering the
different methods of risk transfer for mortgage credit, what
are the key issues to keep in mind?
Mr. Durant. We would say separating out the distinction
between protecting loans at the loan level, issuing credit
enhancement with the origination of the loans versus the
guaranteeing of the security-level enhancement. You need both.
Loans down to, say, a 60 LTV level, those are very safe loans.
You do not have to be as worried about loan-level credit
enhancement. But when you get to lower downpayments and other
risk factors on loans, you certainly do want to make sure that
the loans are, at the time they are originated, looked over by
entities like mortgage insurance companies who review the
underwriting and ensure that the credit enhancement is in place
prior to that loan entering into the securitization world.
Chairman Johnson. Senator Crapo.
Senator Crapo. Thank you, Mr. Chairman.
Mr. Bordia, in your testimony you discuss the need to have
a wide variety of options for moving mortgage credit risk into
the private market as opposed to limited options. Could you
explain the reasoning for that a little better? I assume that
if we are too restrictive in the options that market
participants have to satisfy private capital requirements, we
would risk diminishing the appetite of the market to move
private capital into the market. But could you explain that
dynamic a little better?
Mr. Bordia. I think I will refer to some of the things that
I mentioned in answering the previous question. So if you are
trying to insure close to $5 trillion worth of mortgages, then
you need a lot of capital. We have seen about $1.8 billion of
sales in credit-linked notes from GSEs this year, and they were
massively oversubscribed. But to assume that you can go next
year and sell $15, $20, or $30 billion of those kind of notes I
think it is practically not possible.
So the main reason when I say that, you know, we need to
look at multiple options is because it is a limited amount of
capital chasing different kind of assets on the fixed-income
side. We have about $850 billion of nonagency still
outstanding, and those nonagency securities have about $60 or
$70 billion in paydowns. And to the extent that all of those
investors have mortgage credit expertise, you could assume that
some part of that $60 or $70 billion comes into the
marketplace. But if you try to do anything more than that, it
is going to be complicated.
Then there is also another set of investors on the equity
side which can basically--where you can also raise money and do
something using the bond guarantors.
Senator Crapo. Thank you. And, Mr. Palmer, you indicated in
your testimony that there are a number of regulatory, tax, and
accounting, I guess I will call them, ``barriers''--I do not
know if you used that word, but it sounded to me like that is
what you were saying--to effectively creating the necessary
instruments and risk management procedures that we need to
engage in. Could you elaborate on that a little bit?
Mr. Palmer. Sure. We spent quite a bit of time working on
the structures that we came to market with this year, and I
think over the course of the 2 years that we worked on it, we
definitely ran into a number of regulatory issues, either
proposed or new rules that have come out. A couple key ones
that I want to mention that I refer to also in my written
testimony is, you know, we referred to the transactions that
have been done as credit-linked notes. They were actually, in
fact, not credit-linked notes. They were debt offerings of the
GSEs, and that was because of a new ruling that came down under
the CFTC in 2012 that would have deemed these type of credit-
linked notes as commodity pools. You know, we have engaged with
FHFA and with Fannie Mae in conversations with the CFTC to
ensure that we can be fully compliant with those rules. And we
hope to ultimately move to a more traditional credit-linked
note.
The importance there is that right now the structures that
we are doing, investors are taking counterparty risk to the
GSEs. Moving to a credit-linked note, you capture all of the
cash up front in a bankruptcy remote trust, which protects both
Freddie Mac and the investor from counterparty risk. So we
think it is an important thing to move to in this credit-linked
note structure.
Second, we think it is important for these securities to
have, I will call it, ``equal treatment'' from a tax and a
regulatory as others. So there have been some discussions here
about senior-subordinate securitizations. There are some pros
and cons there. One of the key cons is the lack of
compatibility relative to the TBA market.
Investors today that invest in a senior-subordinate
security are able to get--they treat it as a REMIC, which has
more favorable tax treatments under that rule, and having a
credit-linked note that is structured very similar to a senior-
subordinate, having similar tax treatment for investors, both
domestic REITs are the common investors as well as foreign
investors, is important to know.
Senator Crapo. All right. Thank you. You identify some
problems in getting through the complexity here, but it is
important that we get it right.
Mr. Durant, the last part of your testimony this afternoon
related to the necessity that reform be comprehensive in the
sense that, if I understood you right, we need to be sure that
all loans are subject to the same capital requirements. Could
you expand a little bit on your comment there?
Mr. Durant. Sure. Clearly you have to think about capital
that is required for loans that are held on balance sheets at
banks versus within S.1217 is a 10-percent subordination level
equivalent to a bond guarantor being required to be capitalized
to a 10-percent level. Clearly, those are not the same things.
More importantly, I think we also have to also think about
FHA, VA, and Ginnie Mae and how does all of this interact with
FHA financing in particular. The changes we make to the rules
around a Government-guaranteed section replacing Freddie and
Fannie clearly are going to potentially drive loans back into
FHA-VA lending.
Senator Crapo. All right. Thank you.
Chairman Johnson. Senator Tester.
Senator Tester. Thank you, Mr. Chairman. I want to thank
both you and the Ranking Member for your work in holding these
hearings. I think that they have been helpful, and I think that
as a result there is some pretty significant progress, so I
want to thank you both.
This is for Mr. Palmer and Ms. Davis. What guidance or
advice regarding how to structure credit risk transactions
would you provide to the future FMIC Director as conceived in
S.1217, the Corker-Warner bill, based on your experiences in
structuring such transactions at the enterprise that you have
been through?
Mr. Palmer. Yes, I think the guidance that I would provide
is, one, and it has been talked about here--that we need
multiple types of options. Just within the two GSEs, there's $5
trillion of guaranteed mortgages on our books, or maybe just
under that. To be able to transfer that amount of risk to the
markets I think you need to have multiple types of tools, as
well as in a variety of different economic markets being able
to use some tools more than others.
I think, second, it is important to think about
holistically the various different regulatory rules as it
relates to housing reform to ensure that it balances the need
to be able to transfer credit risk to private markets as well
as kind of the appropriate governance and controls for a
sustainable market.
And then, last, it is important to preserve a forward
market that we have discussed, the TBA market, which is what we
have today, that provides key benefits to borrowers and
lenders.
Senator Tester. Ms. Davis, anything to add?
Ms. Davis. Sure. I definitely agree with Mr. Palmer in
terms of what we learned about the TBA market and the fact that
if you just think about the scalability of the types of
programs that we are looking at here, the fact that you can tap
into TBA for the funding of the loans and think about, you
know, as I mentioned in my testimony, we sold risk on $27
billion worth of loans, but the securities amount that we
issued was $675 million. So we were able to actually issue a
much smaller amount of securities in order to cover that large
amount of loans because of the fact that the loans had already
previously been funded through TBA.
I think the second thing that I would add is one of the
things that we learned definitely was the importance of the
role of an intermediary in the transactions. As we went out on
our road shows, investors were very interested in understanding
how we originate the loans, what our credit policies are, what
the QC standards are on the loans, how we will service the
loans. And having that transparency around our practices and
having that role of somebody setting those standards--and
investors did a lot of credit work on these deals. We spent a
lot of time educating the rating agencies. We spent a lot of
time educating investors on what is in our guides, selling and
servicing guides, you know, how that aspect of the market
works. We released performance data on 18 million loans, and
investors used that to create models and understand the credit.
And I think we found that having that intermediary role was
very important to investors.
Senator Tester. OK. Thank you. And anybody who wants to
answer this can. What lessons should we draw from the response
of market participants to the enterprises' credit risk sharing
transactions? Go ahead, Ms. Davis.
Ms. Davis. I will jump in and start. I think we learned
certainly that in the current market, which Mr. Palmer pointed
out is certainly very favorable to credit risk transfer, we
learned that, you know, investors are interested in taking on
mezzanine credit risk on a very high credit quality pool of
loans, and we definitely had a very strong response. I think,
you know, the one thing that we would keep in mind is that the
current market is definitely very favorable.
Mr. Bordia. I would add that I think initially, when the
first deal came out, there was some level of skepticism in the
market. And as it became clear that it is going to be more
programmatic, investor interest has actually gone up. If you
look at the execution, it has been significantly better than
the first deal for the second and third deals. So I think
obviously to the extent that you can make sure that everyone
understands it is going to be a big program and you will
continue to see these kind of securities come into the
marketplace, it will certainly help.
Senator Tester. OK. As conceived in the Corker-Warner bill,
the FMIC Director would be tasked with approving credit risk
sharing transaction structures. Under this construct, we
believe that the bond guarantors will take on the bulk of the
credit risk required, but also provide the FMIC Director with
the flexibility to approve structures based on capital markets'
execution. How do you feel about this flexibility? And do you
think it is appropriate? Go ahead, Mr. Palmer.
Mr. Palmer. You know, I do think that the flexibility is
important. I think, you know, we are in the very early stages
of credit risk transfer. Freddie Mac has done two different
types of risk transfer structures, Fannie Mae also two
different types, three types between the two of us. And we are
still in the learning stages, and I think we will continue to
be there, and continuing to have that flexibility is important.
Senator Tester. OK. Thank you all very much.
Thank you, Mr. Chairman.
Chairman Johnson. Senator Corker.
Senator Corker. Thank you, Mr. Chairman, and I thank each
of you for being here and for your interest in the topic. I am
sure I am going to be redundant. I stepped in from another
setting, and I apologize.
In the 1217 bill that many of us have talked about and been
involved in and many of the questions end up being centered
around, we tried to lay out a situation that does create
flexibility, as Senator Tester just mentioned, and looks at
various ways of having private capital through a bond
guarantor, through credit-linked notes, through A and B pieces,
and trying not to be too prescriptive so that all of those are
available.
Is it your sense--and I would like to ask all four of you
this--that we are better off giving that flexibility and
allowing the FMIC Director, should a bill like this pass,
something similar, is it better for us to have that flexibility
and for us to be able to test each of those as we move along?
Or is it better for us to be more prescriptive and eliminate
some of those possibilities?
Mr. Palmer. I do think it is very important to have that
flexibility. I think today, as I said in my oral statement, it
is a very good market for taking on mortgage credit risk. House
prices are generally appreciating. The underwriting quality of
the loans is very strong. I think as we have seen in previous
markets, sometimes the capital markets go away, and we need to
have other options available to be able to transfer credit risk
in a variety of different markets.
Senator Corker. OK.
Ms. Davis. Yes, I think we would definitely agree that the
flexibility is important. Just relating back to the specific
transactions that we did, we certainly looked at a variety of
different structural options even within the credit-linked note
structure, and just specific to these deals, early on we looked
at sort of different points of risk to sell and should we sell,
you know, 3 percent, 4 percent, 5 percent, the right level. And
so a lot of it for us, the analysis came down to this
particular pool of loans and, you know, what did we expect the
risk to be on this particular pool of loans. And so we felt
that, you know, this was the level that was appropriate for
this particular pool.
So I think just the exercise that we went through on this
transaction showed us that you are going to have kind of
different structures that you are going to want to use in
response to different market conditions.
Mr. Durant. We, too, think the flexibility is important,
subject, of course, to the comments I made earlier about
separating out the loan-level credit risk for the higher-risk
loans, having that coverage placed as the loans are originated
versus the flexibility of developing credit risk sharing at the
security level, once you have made sure that at the loan level
the credit enhancement is already there.
Mr. Bordia. Well, I agree with pretty much everyone. I
think flexibility is important. If you are looking at really
good quality collateral, then there is no point asking someone
to hold 10-percent capital. But then there are also collateral
types where the average expected losses are going to be north
of 10 percent, and if you have just 10-percent capital cushion
for that, it is not going to be enough. So I think flexibility
is important. I would also add that while you want to keep the
flexibility, you want to be somewhat prescriptive. You do not
want to give a lot of leeway in terms of what people can hold
or cannot hold.
Senator Corker. And speaking of that, I have one more
question, but you a minute ago sounded somewhat negative on the
ability to have the available capital together. If I
understood--and I was doing something else at the time, but I
think you were talking about fixed income. But if you were
seeking private equity, too, and you had clear rules and you
had time, is there any question in your mind that with clear
rules and time and looking at the combination of all markets,
you would have the kind of capital available to have 10-percent
capital at risk in advance of any Government guarantee?
Mr. Bordia. I think over time I think it is certainly
possible. If you are looking at a collateral pool which has
half a point of loss and you ask them--and you need 10-percent
capital, the market will trade whatever, the first 2 or 3
percent as equity, and you should be able to raise the 3
percent to 10 percent at much more attractive levels. So I
think to answer your question in short, it is certainly
possible, but it will take a pretty long time.
Senator Corker. And let me just ask one more question, if I
could. I assume that the bond guarantors, if, you know, they
are playing--if they are in this arena, they themselves also
behind their risk are going to be accessing capital markets.
They are going to be using, I think, as you all described a
minute ago, they are going to be using credit-linked notes and
subordinated components to actually lay off their 10-percent
risk. Is that correct?
Ms. Davis. Yeah, I think we would agree with that. You
know, it is a little hard to extrapolate, obviously, from just
three transactions. But I think you can look at the model that
we used with these particular security structures where you
could have a guarantor who then taps into the capital markets
to lay off some of that credit risk when the markets are good.
You know, I think the nature of those types of markets is they
are cyclical, right? So if you have a guarantor who can then
access those markets when they are open, take advantage of, you
know, perhaps better pricing for credit when the markets are
good and then that becomes a component of their risk
management.
Senator Corker. Well, thank you. And I think it seems like
you all are shaking your head up and down, and so without
belaboring the Committee, Mr. Chairman, thank you, and I thank
each of you for your testimony.
Chairman Johnson. Senator Warren.
Senator Warren. Thank you, Mr. Chairman, Ranking Member
Crapo, for doing this.
We have been talking today about two basic ways of
transferring credit risk: a structured transaction, here the
senior-sub, or an investor takes the credit risk up front and
bond insurance, which pays out on the back end if a pool of
loans starts to go bad. And it seems like each one of these has
some pluses and some minuses as we think them through.
Obviously the nice thing about the senior-sub structure is
that the money is there up front to absorb any losses. But from
what I understand, that structure is incompatible with the TBA
market. So if an investor is going to take on the credit risk
of a pool of mortgages, they want to know exactly what the
mortgages are that are in the pool. And, of course, that cannot
happen in a TBA market.
Is there anybody who disagrees with that?
[Witnesses shaking heads.]
Senator Warren. OK, good. So we are all in the same place
on that. Good.
Well, I do not think the Government should be standing
behind structured transactions if those transactions do not
work in a TBA market. The TBA market allows borrowers and
lenders to lock in a rate in advance, critically important for
families, for community banks, for credit unions, for access to
mortgages.
Now, bond insurance works with a TBA market, but since it
pays on the back end instead of on the front end, the risk is
that the bond insurers will not actually have the money to pay
off if a whole deal goes south, if a whole tranche goes south.
So if the Government is going to depend on bond insurers to
stand before the Government guarantee, then the Government is
going to have to be very diligent in overseeing those bond
insurers to make sure they have enough high-quality capital to
cover the risks that they have taken on.
So do any of you have any ideas about the kind of
regulatory and oversight burden that this is going to place on
the Government? Mr. Durant.
Mr. Durant. Sure. So we, of course, look to our existing
regulatory environment. That is what we do today. And obviously
people point to it and say, ``But the bond guarantors active in
the residential finance market largely failed in this crisis.''
Senator Warren. Yes, they do say that. For a reason.
Mr. Durant. And, you know, one of the things we look at as
why did that happen is, as I talked about, the separation of
the loan-level risk, the individual loan risk, from the overall
security risk. Bond insurance was originally designed to
guarantee municipal bonds. In 1971, my predecessors at MGIC,
they created that business to do that. It was adapted for use
in mortgage-backed security financing, and people did not think
a lot about the different loan-level risks that are present on
individual borrowers versus a municipal bond that is backed by
taxpayer funds. Very different kinds of things.
So, first of all, this separation of the loan-level risk
from the security risk we think is a very important component
of ensuring that the bond guarantor model is going to survive.
Now, as I talked about in my testimony, right now bond
insurers, their insurance is regulated by the States. That is
the way things work, and unless you want to make changes to
McCarran-Ferguson, we kind of have to work within that
construct.
That said, I think particularly in the mortgage insurance
industry, we have a very good example of how State regulation
combined with Federal oversight that OFHEO and now FHFA and
Freddie Mac and Fannie Mae has worked quite well. The fact is
five of the eight mortgage insurers are still here, and, you
know, I think that model has shown that it does work.
Senator Warren. So, Mr. Durant, your position is there is
adequate regulation already in this market, notwithstanding
what has happened over the past----
Mr. Durant. I think----
Senator Warren. Notwithstanding the facts?
Mr. Durant. ----the changes that are being made are
addressing a lot of those concerns. So within the mortgage
insurance----
Senator Warren. So you are saying there needs to be more
regulation.
Mr. Durant. Different regulation, yes.
Senator Warren. All right. Mr. Bordia, did you want to add
anything?
Mr. Bordia. Yes, I think there are just a couple of things
that I wanted to add.
So, one, to the extent that you have bond guarantors
guaranteeing some of these securities, either do you want the
capital that is standing behind sit in some separate reserve
account or something like that? Because if insurers or bond
guarantors are in five or six different lines of businesses and
something bad happens somewhere else, then it is going to be a
problem. The money is not going to be around to take care of
the losses that come from Fannie Mae or Freddie Mac or any such
future entity.
Senator Warren. So we need some kind of regulatory
structure around this, and this is----
Mr. Bordia. Yes. Either you ensure that the cash is sitting
somewhere and so you raise like 2- , 3-percent capital, then
you raise a lot of debt, but that, for example, can only be
used to cover losses from these entities; or you just set aside
money right up front to take care of these losses.
Senator Warren. So that brings me then to the final
question and how these tie together, and that is, is there a
viable alternative to the two approaches, structured deals and
bond insurance? What about the proposals from the Center for
Responsible Lending and the New York Fed for one or more
mutuals that issue and guarantee mortgage-backed securities?
Any thoughts on that, a third way to go about this that does
not have this kind of regulatory structure but still supports
the TBA market?
Mr. Durant. I think those still will require a regulatory
environment of some kind, and really, for example, I am more
familiar with the New York Fed proposal. I think the idea of
cooperatively owned and operated entities the way they have
suggested it actually fits in quite well in the S.1217
structure. Issuers guaranteeing their own securities is done
every day. It is called Ginnie Mae. And so we already know what
that looks like and how it needs to be managed.
Senator Warren. Although part of what we get, of course, in
a mutual is that, in effect, they regulate each other. They
become self-regulating instead of the----
Mr. Durant. If they are set up correctly, right.
Senator Warren. ----``catch me if you can'' of other forms
of regulation. So thank you. I appreciate it, Mr. Chairman. You
know, I think this is a really central question when it comes
to housing finance. Who is going to take on the credit risk of
all these mortgages? And it is really critical we get this
right, so thank you.
Chairman Johnson. Senator Warner.
Senator Warner. Thank you, Mr. Chairman. First of all,
echoing what some of my other colleagues have said, thanking
you and the Ranking Member for continuing to aggressively move
these hearings forward, and we have gone through snowstorms and
shutdowns, and obviously the focus of this Committee is to take
on certain international issues as well. I know Senator Corker
has been involved, and others. So I am very grateful that we
have kept an aggressive schedule, and my hope is that you and
the Ranking Member take some of the work that was started and
reshape this that we can very, very early on in the New Year
get a piece of legislation marked up and out of this Committee.
Mr. Bordia, I think you responded a bit to--I think you
responded affirmatively to Senator Corker that, you know,
because of this variety of tools that would be out there, we
could get to that 10-percent level, and I would simply point
out as well that there are other proposals out there that get
rid of any Government role at all that would require
exponentially higher amounts of private capital, would you not
agree, if you get rid of the Government role entirely, as some
have proposed?
Mr. Bordia. Again, I am not sure, I think, what the right
capital required is. If you look at the total GSE portfolios, I
am pretty certain losses on that $5 trillion portfolio are not
$500 billion, and they have clearly gone through 2006 and 2007.
So I am not sure, I think, what the right number is. What I was
trying to say is that it might not be a good idea to have the
same amount of capital, and even if you require $100 or $200
billion of capital that has to come from the private markets,
it will take a lot of time.
Senator Warner. I believe what you have also said and I
think what some of us believe is if there has been some
overshooting and an extra buffer to make sure, again, that the
taxpayer does not get hit with another $188 billion bill the
way we did last time and now some people coming back and
saying, you know, that risk exposure the taxpayer took should
be returned at a 1:1 rate and that somehow makes everything
whole, as a former VC, that would not be a return I would be
happy with. But if we do have an abundance of caution here,
there will be an ability, I think, to tranche part of that
capital out, and as you said earlier----
Mr. Bordia. Yes. You cannot raise all of the demand as
equity capital.
Senator Warner. Right. It would be priced in different
ways. I think Senator Warren has raised, you know, some good
components about this, the question about how we get this
insurance component piece right, and, you know, I would point
out that in S.1217 we do have a mutual. We would still--the
concept of that mutual to service the portion of the market to
make sure there is quality of pricing, we still envision the
guarantee piece, the private capital loss piece being perhaps
separate from that, or separate from that mutual, but we think
we have got that in the toolbox.
I guess one of the things, Mr. Durant, that we have
pushed--and I think now you have had some concern about--is you
pointed out in your testimony that you thought it was
appropriate that the insurance function get separated from the
loan level, from the security level, and that if we are--you
know, going forward, and particularly if an entity is in a
variety of these lines of business, we are going to have to
have, you know, separate levels of characterization, separate
structures so we do not have this overlapping problem where, if
the string starts to get pulled on one of these lines of
business, it does not bring down the whole house. Do you want
to--and I do believe that it will require--and we have put some
requirements in S.1217 to make sure--I know there is some
pushback from you all on this, but that you do not have that
kind of overlapping coverage.
Mr. Durant. Yeah, I think our concern would be within a
parent company holding structure, why wouldn't you allow a bond
guarantor to also, you know, be a sibling to a mortgage
insurer? As long as they are separately capitalized entities,
we think that is the right way to do those things. Merging the
loan-level mortgage insurance exposure with the security-level
bond exposure into a singly capitalized entity or, as Sandeep
said, having that entity also be exposed to other kinds of
insurance risks clearly is not going to accomplish what you
want to accomplish here. We think the monoline requirement is a
very important aspect of ensuring that capital is brought into
these entities for the express purpose of funding mortgages
through the very long cycle, and you want to keep the capital
within those entities dedicated to that task.
Senator Warner. And we do envision the FMIC would be
playing, again, a very important role in conjunction with your
State insurers to make sure that gets----
Mr. Durant. Absolutely. As the ultimate counterparty, they
are the ones who are saying, look, we are the ones holding the
bag for you guys, you----
Senator Warner. My time is about up. Let me make one other
point. I just want to follow up on something that Senator Crapo
raised in his line of questioning, which was how do we make
sure--and, again, Mr. Palmer and Ms. Davis, you may want to
comment on this. If we are going--we all recognize--and one of
the things we have tried to build in as a transition time a
ramp-up period so that we can get more of this private capital
in place, and how we can ensure that we do not put unnecessary
regulatory, accounting, or other burdens to make this
transition. We have talked about the concern that some of the
entities had about registration with CFTC. I know there are
some concerns in terms of tax treatment around REITs. I know
there are also accounting issues.
One of the things I would ask for the record would be
whether from the experience you have had in the transactions so
far and looking forward as you expand that universe, what are
those items that we could perhaps look at to minimize the flow
of private capital coming in to play this very, very important
role in terms of protecting the taxpayer. And, again, with the
discretion of the Chair, if they would let them answer that
question, then I will be finished.
Mr. Palmer. Yes, first, we would love to be able to work
with you and your staff on these issues. I think we definitely
have learned a lot over the last 2 years, and I think as we
continue to grow and expand these programs, I anticipate that
we will continue to learn more and that we can help articulate
better some of these issues that should be addressed.
Senator Warner. Thank you, Mr. Chairman.
Chairman Johnson. Thank you again to all of our witnesses
for being here with us today and helping the Committee better
understand the options for increasing private capital's role in
the housing finance system.
Chairman Johnson. This hearing is adjourned.
[Whereupon, at 3:45 p.m., the hearing was adjourned.]
[Prepared statements, responses to written questions, and
additional material supplied for the record follow:]
PREPARED STATEMENT OF KEVIN PALMER
Vice President, Strategic Credit Costing and Structuring, Freddie Mac
December 10, 2013
Chairman Johnson, Ranking Member Crapo, and Members of the
Committee, thank you for inviting me to appear today. I am Kevin
Palmer, Vice President of Strategic Credit Costing and Structuring at
Freddie Mac. I head the business unit with primary responsibility for
the development and execution of the credit risk sharing transactions
that are the subject of today's hearing.
In my statement, I summarize the risk-sharing transactions we have
undertaken to date, discuss the lessons we have learned from these
transactions, and outline our future risk sharing plans.
I would like to highlight three points from my statement that I
believe will be useful as you consider these issues:
First, we have been very encouraged by strong investor
interest in our credit risk sharing offerings to date. However,
it is important to note that conditions today are highly
favorable for investment in mortgage credit risk. We have seen
from recent history that investors can and will leave the
mortgage market when conditions are less favorable. So I
caution that it will take some time to determine the level of
long-term sustainable investor interest in mortgage credit risk
sharing, particularly during market and economic downturns.
Second, we have encountered a number of regulatory, tax and
accounting issues that have either affected investor interest
in participating in our credit risk sharing transactions or
influenced our choices of particular structures for those
transactions. We believe Congress should consider these and
other similar issues as it determines the role that credit risk
sharing instruments can play in any future housing finance
system.
Finally, we designed our credit risk sharing transactions
to ensure they are cost effective and preserve forward markets.
Freddie Mac's Progress Under Conservatorship
The Freddie Mac of today is not the company that existed prior to
conservatorship. We have a new management team, including new chief
executive officer, chief financial officer, general counsel, chief risk
officer, chief compliance officer, chief administrative officer, head
of human resources, and chief information officer. In addition, we have
a new head of every business line: single-family, multifamily and
investments. Most of these leaders are new to Freddie Mac, while a few
are in new roles since conservatorship. At the same time, we have
retained many employees with significant experience in and knowledge of
the mortgage finance industry.
Freddie Mac is highly mindful and appreciative of the taxpayer
support we have received. We are focused on using this support wisely
and effectively to provide liquidity to the home mortgage market, help
at-risk borrowers avoid foreclosures, and fulfill the objectives of the
conservatorship. Let me offer a few examples:
We helped more than 11 million families buy, refinance, or rent a
home since 2009: Our mortgage purchases enabled 7.5 million families to
refinance into lower interest rate home mortgages. For loans we
refinanced during 2013, families are saving an average of $3,400 per
year. We also funded home purchases for 1.9 million families and rental
housing for 1.5 million families.
We helped 913,000 at-risk families avoid foreclosure since 2009: By
preventing avoidable foreclosures, we not only help at-risk families
but also stabilize and revitalize neighborhoods. More than 475,000 of
these families received loan modifications, saving each family an
average of $5,220 per year. We also are implementing a Federal Housing
Finance Agency (FHFA) directive to substantially improve how servicers
work with delinquent borrowers.
We implemented stronger credit standards, resulting in a
significantly improved book of business: Our focus is on helping
borrowers own homes that they can afford and keep. As of September 30,
2013, mortgages we purchased after 2008 comprise 73 percent of our
single-family credit guarantee book of business but only 8.6 percent of
our credit losses. Even with our preconservatorship book of business,
Freddie Mac's single-family serious delinquency rate--2.58 percent as
of September 30--is less than half the mortgage industry average of
5.88 percent (as of June 30).
By the end of 2013, we will have paid $71.345 billion in dividends
to taxpayers: This slightly exceeds the $71.336 billion in U.S.
Treasury funds we have drawn to date. As explained in our November 7,
2013, financial release, Freddie Mac's payment of dividends does not
reduce the balance of prior draws of Treasury funds received, and
Treasury retains a liquidation preference of $72.3 billion on Freddie
Mac's senior preferred stock.
We could not have achieved these results without the support we
have received from FHFA, Treasury and the American taxpayer. But these
results also attest to the hard work, commitment, and expertise of our
employees.
Credit Risk Sharing Initiatives
Freddie Mac began analyzing mortgage credit risk sharing structures
in early 2011 as a way of removing some of the risk on our books. This
included studying both the economics and the operational issues
involved in various risk transfer options and structures. This work
helped prepare us for FHFA's directives in its 2012 and 2013 scorecards
to begin undertaking credit risk sharing transactions. The 2013
scorecard set a target for us to undertake transactions involving
single-family mortgages with at least $30 billion in unpaid principal
balances. FHFA specified that we must conduct multiple types of risk
sharing transactions to meet this target. Such transactions could
include expanded mortgage insurance, credit-linked securities, senior/
subordinated securities, and other structures. In addition to reducing
our credit risk exposure, the risk sharing goal is also intended to
bring more private capital into the market and demonstrate the
viability of multiple types of risk transfer transactions involving
single-family mortgages.
Freddie Mac has undertaken three such transactions during 2013. Our
first transaction, announced in July, was an offering of Structured
Agency Credit Risk (STACR) Debt Notes. We just settled a second STACR
offering on November 12. Also on November 12, we entered into a
reinsurance transaction. The aggregate unpaid principal balances of the
mortgages involved in these three transactions will include more than
$30 billion that we believe qualifies toward FHFA's scorecard.
Common to these three transactions are several objectives:
Reduce Freddie Mac's, and therefore taxpayers', exposure to
the credit risk of our mortgage purchases by transferring a
portion of that risk to private investors.
Bring new credit investors into the mortgage market.
Create products that are scalable and sustainable over
time.
Preserve the cost efficiencies of a forward market.
STACR Debt Notes
STACR Debt Notes allow Freddie Mac to transfer credit risk from
recently acquired single-family mortgages to credit investors who
invest in the notes. Interest and principal payments to investors are
determined by the delinquency and principal payment experience on that
group of newly guaranteed mortgages, known as a reference pool.
Although structured as debt notes that are unsecured general
obligations issued by Freddie Mac, the transaction is similar to a
credit linked note, which allows us to take advantage of the cost
efficiencies of the TBA market.
We structured STACR Debt Notes to attract investors by providing a
large and highly diversified pool of loans in the reference pool.
Diversification is attractive to credit risk investors because it
reduces idiosyncratic risk stemming from concentration in specific
geographical areas, in originator quality, and servicer practices. The
reference pools for the first two offerings together included more than
200,000 high-quality loans, which are diversified based on
geographical, originator, servicer, and other risk factors,
representing more than $50 billion in unpaid principal balance. These
pools consist of a subset of 30-year fixed-rate single-family mortgages
acquired by Freddie Mac in two recent quarters. Most other securities
that transfer mortgage credit risk, by comparison, are based on a
smaller pool of mortgages, generally less than 1,000 loans.
Freddie Mac remains the master servicer in the STACR transactions,
retaining control of the servicing of the loans in the reference pools.
This is beneficial because the loans will be subject to Freddie Mac
servicing guidelines, allowing us to maintain our strong loss
mitigation support to borrowers. The structure provides for a defined
loss transaction--when a borrower is 180-days delinquent (behind by 6
months) the bond holder takes the defined loss. The structure also
allows Freddie Mac to manage the assets of the pool after 180 days.
Freddie Mac retains some risk exposure (at least 5 percent of the
losses), assuring investors of our aligned interests. This ``skin-in-
the-game'' is important for servicing and loss mitigation control. To
further demonstrate our alignment of interest, Freddie Mac also
retained the first-loss risk position in the two STACR offerings.
Retaining that first-loss position helped investors get comfortable
with this new type of credit risk sharing instrument, and helped make
the structure more economically attractive.
Our first two STACR offerings received positive market responses.
About 50 broadly diversified investors participated in each offering,
including mutual funds, hedge funds, REITs, pension funds, banks,
insurance companies, and credit unions.
Additionally, our STACR transactions had little or no impact on the
TBA market. The TBA market provides the means for lenders to sell
conforming loans into the secondary market before they are originated.
This enables lenders to better manage the risks of 30-year fixed-rate
loans and allows borrowers to lock in mortgage rates up to 90 days in
advance of closing.
What We Have Learned From Our STACR Offerings
Based on our initial STACR offerings, we can identify several
issues and challenges:
Limits to investor appetite: While our initial offerings received
positive market responses, this does not guarantee that future
offerings will receive equal levels of investor interest. Investor
appetite for a particular asset class at any given time depends on a
variety of factors, including broad economic and market conditions and
returns offered by other asset classes. Our first two STACR offerings
took place during very favorable conditions for investing in mortgage
credit risk. For example, house prices are generally appreciating and
credit quality is high by historical standards. As we have seen from
recent history, investors will leave the mortgage market when risks and
returns are less favorable. While we believe we are well on the road to
creating an attractive and scalable investment product, it will take
some time to determine the level of long-term sustainable investor
interest in STACR Debt Notes as an asset class, particularly during
market and economic downturns.
Credit ratings: Our first transaction was not rated by a rating
agency. In the course of structuring it, we found investors had
differing views over the need for a rating. In the end, we decided
against obtaining a rating because doing so would have slowed our
ability to complete two transactions this year. This somewhat limited
investor participation and impacted the pricing on the first
transaction. In the second transaction, one of the two tranches was
rated Investment Grade by Moody's and Fitch. The pricing on this
transaction was substantially improved. While we attribute most of this
improvement to greater market acceptance and familiarity with STACR
Debt Notes, obtaining a credit rating also helped.
Tax treatment: Current tax laws affect investor interest in STACR
transactions. Real Estate Investment Trusts (REITs), for example, must
primarily invest in real estate assets, including interests in
mortgages. Because the STACR transactions were general obligation debt
issuances and not secured by interests in mortgages or real estate
assets, they did not qualify as real estate assets for REIT purposes.
While STACR Debt Notes could be held by REITs, there are restrictions
on the amounts. Also, Real Estate Mortgage Investment Conduits (REMICs)
cannot hold STACR Debt Notes as collateral because the notes are not
secured by real property or interests in mortgages and are not
interests in mortgages secured by real property.
Accounting treatment: Investors in STACR Debt Notes will mark their
investments to market under accounting rules, and this will discourage
some investment in them. Some large investors, such as insurance
companies, are not interested in assets that are marked-to-market
because this would create additional income statement volatility.
Freddie Mac also faces increased income statement volatility from mark-
to-market treatment of our STACR issuances.
Regulatory hurdles: A Commodity Futures Trading Commission (CFTC)
regulation played a role in our decision to structure STACR Debt Notes
as an unsecured general obligation instead of a credit linked note.
That regulation requires Freddie Mac to register with the CFTC as a
Commodity Pool Operator and a Commodity Trading Advisor or to seek a no
action letter from CFTC in order for us to issue credit linked notes.
Registering with the CFTC likely would have required us to create a
subsidiary company--a complicated matter given our being in
conservatorship. Further, if we were to seek a no action letter as
issuer, our understanding is that some investors would be required to
register with the CFTC or require a no action letter in order to
purchase credit linked notes we would issue. CFTC's rule did not
anticipate this type of offering, and changes to the rule would help.
In the meantime, Freddie Mac, in concert with FHFA, continues to work
with CFTC to ensure our full compliance.
Our experiences with these tax, accounting and regulatory issues
suggest that policy makers, in the course of legislating housing
finance reform, should carefully consider how these and other similar
issues can affect the ability and willingness of private investors to
assume mortgage credit risk.
Despite these challenges, we believe Freddie Mac's STACR
transactions to date have met our objectives and help us meet FHFA's
scorecard objectives for 2013.
Reinsurance Transaction
Freddie Mac announced on November 12 that we had entered into a
reinsurance-based credit risk sharing transaction. We obtained an
insurance policy underwritten by Arch Reinsurance Ltd. to cover up to
$77.4 million of credit losses for a portion of the credit risk Freddie
Mac retained from the reference pool in the first STACR transaction.
The transaction with Arch enabled both parties to leverage this
reference pool, and the associated disclosures and due diligence.
This new insurance coverage is intended to attract new sources of
private capital from nonmortgage guaranty insurers and reinsurers
interested in assuming a portion of the credit risk on specified
portions of our single-family mortgage loan portfolio. Reinsurance
companies are large diversified companies that specialize in managing a
variety of risks. Freddie Mac regards reinsurance companies as a
promising new source of capital for mortgage credit risk transfer.
Our experience with conducting this first reinsurance transaction
has led us to conclude that there is interest at this time for U.S.
mortgage credit exposure among the reinsurance community. Accordingly,
we see potential to build a risk sharing product targeted at
reinsurance companies that meets our objectives of transferring credit
risk, bringing new investors into the market, creating repeatable and
scalable products, and preserving the cost efficiencies of the TBA
market. Of course, the level of long-term sustainable interest by
reinsurance companies in mortgage credit risk sharing transactions,
particularly during market and economic downturns, remains to be seen.
Risk Sharing Plans Going Forward
In addition to conducting additional STACR and reinsurance
transactions, we are looking at two other options for credit risk
transfer. The first is risk retention by mortgage originators who sell
mortgages to us through recourse agreements. The key challenge to
recourse transactions is that sellers retain these obligations on their
balance sheets. This has regulatory capital consequences for regulated
financial institutions.
A second option we are exploring is a senior-subordinate
securitization. While doing this type of securitization would not allow
for TBA securitization on these loans under current rules, this
structure is common, particularly for jumbo and other nonconforming
loans.
Conclusion
As Congress determines the future structure of the housing finance
system, Freddie Mac will remain focused on providing liquidity to the
home mortgage market, helping at-risk borrowers avoid foreclosures and
protecting taxpayers' investment in the company. This includes working
with FHFA to develop new and innovative approaches and products to
transfer credit risk from taxpayers to private investors.
Thank you again for this opportunity to appear today.
______
PREPARED STATEMENT OF LAUREL DAVIS
Vice President for Credit Risk Transfer, Fannie Mae
December 10, 2013
Chairman Johnson, Ranking Member Crapo, and Members of the Senate
Banking Committee, thank you for the opportunity to testify today. My
name is Laurel Davis and I am the Vice President for Credit Risk
Transfer at Fannie Mae.
I appreciate the opportunity to share with the Committee
information on the credit risk transfer transactions that Fannie Mae
conducted this year. My testimony today will address how those
transactions were structured and brought to market and the results of
the transactions.
Background
Before I address the specific transactions, I think it would be
helpful to provide additional background on how Fannie Mae manages
credit risk, the credit risk that we currently hold and how we have
sought to reduce this risk substantially during conservatorship. This
is important because what we learned from these particular transactions
was that investors are willing to purchase mezzanine credit risk on a
high quality pool of loans if they receive a yield that meets their
investment targets and where the credit is actively managed by an
intermediary.
In order to assess the risk they were purchasing, potential
investors in our Connecticut Avenue Securities transaction (C-Deal) and
our counterparty in the mortgage insurance (MI) transaction received
significant information about our credit policies, our monitoring of
lender operations, their exposure to the sellers' representations and
warranties, and our reviews of loan origination quality. Investor
comfort with our processes and our ability to enforce representations
and warranties facilitated the investor demand in the transactions.
While this testimony will not address our servicing standards and
oversight of servicers, those factors play a strong role in reducing
our risk of loss and are important considerations for investors
evaluating the credit risk on Fannie Mae loans.
Because we accept the credit risk on securities we guarantee, we
have a rigorous process for managing this risk. We do so through both
the pricing of loans based on the risk such loans entail and the
establishment of underwriting standards that lenders with whom we do
business must follow. Our Selling Guide, which is extensive, is our
legal contract with lenders. It sets forth the underwriting standards
to which lenders are required to adhere.
Our standards are not static. We revise them continuously based on
our analysis of the performance and quality of our acquisitions and our
existing book of loans, changes in market conditions and new issues
that might arise. We also review the loan origination processes of our
lender customers to ensure that their controls are working properly.
During conservatorship, Fannie Mae has made numerous changes to our
analysis of credit risk and to our underwriting and eligibility
requirements to reduce our credit risk. Some of the significant changes
include:
Creating external tools and internal risk models to improve
assessment of collateral value;
Creating a process and risk models to assess the quality of
new loan acquisitions, track defect rates and enforce
contractual rights soon after delivery to mitigate risk;
Standardizing our credit policy by eliminating most
negotiated credit terms with specific lenders;
Tightening our underwriting and eligibility requirements
for higher risk products, including interest only loans and
adjustable-rate mortgage loans;
Implementing a minimum credit score of 620; and
Eliminating contract terms that would allow delivery of
Alt-A loans or other reduced-documentation loans.
As a result of our efforts, and improvements in market conditions,
our serious delinquency rate has fallen dramatically (see Illustration
A). This rate peaked in February 2010 at 5.59 percent and has since
fallen to 2.55 percent as of the end of the Q3 of this year. Even at
its highest point, our serious delinquency rate was substantially lower
than loans in private label securities or held on bank balance sheets.
In addition, the loans we have acquired since 2009 have performed
well. The serious delinquency rate for loans acquired since January
2009 is 0.32 percent. These new, well-performing loans now make up
approximately 75 percent of our total book of business.
The performance of these loans and improving conditions in the
housing market are two of the primary reasons for our recent financial
performance. We have recorded seven straight quarters of profit and, as
of December 31, we will have paid almost $114 billion in dividends to
the Treasury Department versus total draws of approximately $116
billion.
Credit Risk Sharing Transactions
Our credit risk sharing initiatives are aimed at reducing our
retained credit risk, thereby reducing our footprint in the mortgage
market and providing a way for greater private investment in mortgage
credit risk.
It should be noted that Fannie Mae's Charter requires that there be
private risk ahead of Fannie Mae's guarantee for high loan to value
(LTV) ratio loans or loans with less than 20-percent downpayment. This
has not changed during conservatorship. For all loans with LTV ratios
greater than 80 percent, our charter requires that Fannie Mae seek
credit enhancement. This is predominantly done through the required
purchase of mortgage insurance through regulated insurers capitalized
by private capital. Our standard coverage requirements, however,
require greater protection than just the first 20 percent of the
property value. Loans with LTVs in excess of 80 percent generally have
coverage that protects the company down to 68 percent to 75 percent of
the loan amount.
With that background, I would like to turn now to the two
transactions we conducted this year to transfer additional credit risk
to other market participants--our C-Deal transaction and our mortgage
insurance risk transfer deal. The transactions have a few key
similarities. First, in both transactions, Fannie Mae retained a first-
loss credit risk position vis-a-vis investors, while selling mezzanine
risk to investors to cover ``unexpected losses'' after that first-loss
risk piece is exhausted. Mezzanine risk is located between the first
loss and the top loss risk levels. For both transactions, the first-
loss piece we retained covered at least 2 times what we modeled as our
expected losses on the loans underlying the transactions. Fannie Mae
also retained the risk of catastrophic loss, which was sized at greater
than what was experienced during the recent housing crisis.
Second, both transactions were aided substantially by the fact that
an intermediary stood between investors and originators, in this case,
Fannie Mae. Investors did not need to underwrite the credit themselves,
ensure that the underlying loans are properly serviced or make certain
that representations and warranties with originators are enforced. In
these particular transactions, this intermediary role allowed the 77
investors involved in our credit-risk note transaction to rely on
Fannie Mae's credit policies, underwriting standards, lender oversight
requirements, and servicing standards rather than understanding the
standards of more than 1000 lenders with whom we do business. Moreover,
Fannie Mae serves as an ongoing and active credit risk manager on
behalf of itself and the investors.
While we have purchased pool mortgage insurance policies for two
decades, the C-Deal transaction was a successful first attempt to
transfer portions of our credit risk to private securities investors.
We intend to engage in additional transactions to learn more about
investor appetite for credit risk.
These two transactions were positive first steps in transferring
credit risk, but it is early in the process and therefore difficult to
extrapolate the extent to which broad investor demand exists for
securities with residential mortgage credit risk or what yield might be
required.
Connecticut Avenue Securities (C-Deals)
On October 15, 2013, Fannie Mae priced its inaugural credit-risk
note transaction under its Connecticut Avenue Securities series, also
known as C-Deals. This first transaction settled on October 24, 2013.
Fannie Mae's C-Deals were structured to meet certain program goals:
First, to provide an additional avenue to manage the credit
risk of our guaranty business, in addition to our active
management of credit risk as discussed above;
Second, to create a program that is sustainable and
scalable for Fannie Mae;
Third, to explore the most cost efficient means for
transferring credit risk;
Fourth, to not interfere with how lenders currently sell
loans into the secondary market;
Finally, to have no impact on how a loan is serviced.
Loans included in this risk sharing transaction will be serviced in
the same manner as all other loans in our book. Servicers have no
knowledge of which loans are in a C-Deal reference pools and which are
not.
By design, Fannie Mae's C-Deal is structured very similarly to
Freddie Mac's two STACR offerings, the first of which closed in July
and the second in November. There are slight differences between Fannie
and Freddie's deals. Some differences were due to the response to
market feedback, and others were due to how Fannie Mae evaluated the
cost/benefit trade-off of particular deal features.
The C-Deal notes are debt issuances of Fannie Mae. One of the main
differences between C-deal series debt and Fannie Mae's standard debt
is that investors in C-Deals may experience a full or partial loss of
their initial principal investment, depending upon the credit
performance of the mortgage loans in the related reference pool.
Another difference is that the repayment of C-Deal notes is tied to the
credit and prepayment performance of a reference pool of loans. The
reference pool for our first transaction included approximately 112,000
single-family loans with an outstanding unpaid principal balance of $27
billion, which represented about 12 percent of our total acquisitions
for Q3 2012.
To arrive at the pool, we applied certain selection criteria to the
entire population of loans acquired in Q3 2012 to create an eligible
population of loans. For a loan to be included, it had to be: (1) a 30-
year fixed-rate mortgage; (2) not a HARP loan; (3) with an LTV between
60-80 percent; and (4) current in terms of payments since acquisition.
From the eligible population, we used a random selection process to
derive the reference pool. By only referencing the loans, they remain
in the MBS pools, thereby avoiding any disruption to the TBA market.
If the loans in the reference pool experience credit defaults, the
investors in the C-Deals may bear losses. Credit defaults occur in the
C-Deal when a loan in the reference pool reaches 180 days of
delinquency, a short sale, a third party sale, a deed-in-lieu, or an
REO (Real Estate Owned) disposition occurs prior to 180 days of
delinquency.
In the first transaction, Fannie Mae retained the first-loss
position and holds both the catastrophic risk and a vertical slice of
the transaction (see Illustration B). The first-loss piece of the
structure is intended to cover a multiple of our expected losses on the
underlying loans. We decided to hold the first-loss piece for a number
of reasons. First, any securities that represented a first-loss
position may not have been considered ``debt'' for tax purposes and
could carry significant tax consequences to potential investors.
Second, given that this was a new program, we believed that retaining
the first loss would make the transaction easier for investors to
understand, model and price. Lastly, it was unclear if private
investors would be willing to purchase the first-loss position at
pricing that made economic sense for Fannie Mae. However, Fannie Mae
may choose to sell the first loss in subsequent transactions if the
economics are appropriate and the associated regulatory issues are
resolved.
In addition, we sold two classes of mezzanine risk to the market in
order to shed the risk of unexpected losses on the underlying loans.
Fannie Mae retained the catastrophic piece in the structure, which is a
multiple of a stress scenario based on the recent financial crisis
experience. Finally, we kept a roughly 6-percent vertical slice of the
mezzanine risk sold to the market. This was done to align our interests
with investors and give them confidence that we will diligently service
the loans in the reference pool so as to limit losses to both investors
and Fannie Mae.
The mezzanine risk that we sold was comprised of $675 million of
notes split evenly between a senior and junior class. The senior class
of notes, otherwise known as M-1 notes in the marketplace, received an
investment grade rating of BBB- from Fitch Rating Agency. These notes
were priced at 1-month LIBOR + 200 basis points. The investment grade
rating on the M-1 notes opened up investor participation to a wider
variety of accounts, and we believe this will help promote secondary
market liquidity.
The junior class of notes, otherwise known as M-2 notes in the
marketplace, priced at 1-month LIBOR + 525 basis points. Fannie Mae did
not pursue a rating on the M-2 notes. We did not receive strong
feedback from investors that a rating on the M-2 notes would be
particularly important. Both notes were issued with 10-year final
maturities.
A diverse group of 77 investors participated in the offering,
including asset managers, mutual funds, pension funds, hedge funds,
insurance companies, banks and Real Estate Investment Trusts (REITs).
Fannie Mae has disclosed details of our credit risk sharing
activities on our Web site at fanniemae.com. Loan level data, such as
interest rate, LTV and original debt to income ratio, was provided on
the reference pool as part of the initial disclosure. This loan level
data, as well as ongoing performance on the transaction, will be
updated monthly.
We considered other transaction structures, including a senior-
subordinate cash transaction, or ``cash senior/sub'', and a credit-
linked note transaction. As it relates to a cash senior/sub, we closely
examined the use of this structure to transfer credit risk. In a cash
senior/sub structure, loans must be deposited into a trust and
therefore could not be in TBA securities. There are several reasons why
we decided not to pursue this structure. First, compared to the
structure we used, the cost of doing a cash senior/sub transaction
would have been greater. Second, a cash senior/sub structure could
present scalability issues. In a cash senior/sub structure, the loans
themselves are sold and thus there is a transfer of both interest rate
and credit risk. By contrast, in a credit-risk note structure, only
credit risk is sold, since the interest rate risk was previously
conveyed in the TBA markets. Accordingly, if Fannie Mae had used a cash
senior/sub structure to transfer credit risk for the same amount of
loans as occurred in the first transaction, we would have had to sell
$27 billion of securities backed by mortgage loans, as compared to
selling only $675 million in credit securities, given that the loans
had already been funded through the MBS market. Lastly, a cash senior/
sub structure could introduce a number of operational inefficiencies
for lenders compared to how they now conduct business.
With a credit-linked note structure, there are a variety of
outstanding regulatory issues. As alluded to in the press, these issues
include the impact of Commodity Futures Trading Commission (CFTC)
regulations and whether Fannie Mae and investors would need to register
with, and be regulated by, the CFTC. In addition, there are certain
issues under proposed conflicts of interest rules being considered by
the Securities and Exchange Commission, as well as potential tax issues
for certain investors under Internal Revenue Service regulations. These
regulatory concerns, and potentially other issues, will need to be
resolved prior to this type of structure being a viable option.
We gained several insights from the C-Deal transaction. First, as
noted above, in this particular transaction, having an intermediary
that served as an active credit manager was important to potential C-
Deal investors. Our comprehensive approach to credit risk management
helped to build market reception for the C-Deals well before the
transactions took place.
Second, it is essential to be transparent and provide detailed
information to investors. The rollout and launch of the C-Deals were
designed to provide transparency to the marketplace on our requirements
and processes. Over the course of 2 years, Fannie Mae held extensive
discussions with investors and engaged in a road show to assess the
market appetite and structure preferences. It was also critical that we
provided historical loan-level credit performance data to investors on
over 18 million loans acquired by Fannie Mae over the past 10 years so
that investors could make their own assessment of expected loan
performance.
Lastly, we learned that in these particular market conditions,
investors will buy mezzanine risk on a high quality pool of loans if
they receive a yield that meets their investment targets.
It is too early in the process to reach further conclusions from
these transactions. Fannie Mae's next transaction is tentatively
scheduled for January 2014. This will be a debt issuance deal that
references a pool of loans and will be very similar to Fannie Mae's
October deal. The reference pool will be comprised of single-family
loans acquired in Q4 2012 with the same criteria used in our first
deal.
Mortgage Insurance (MI) Risk Transfer Deal
After a competitive bidding process, Fannie Mae entered into a
transaction with National Mortgage Insurance Corporation (``National
MI'' or ``NMI'') to provide credit risk coverage on over $5 billion in
single-family mortgages. The agreement was reached on July 15, 2013,
and coverage went into effect September 1, 2013.
The MI risk transfer deal covers 2 percent of the loans acquired by
Fannie Mae in Q4 2012, each of which had an original LTV of 70-80
percent. These loans were not HARP refinances nor had any credit
enhancements. None of these loans were covered by mortgage insurance
prior to the commitment with NMI.
The coverage was provided in the form of a ``pool insurance''
policy, a form of insurance that we have utilized since the mid-1990s,
to enhance the credit of certain segments of our acquisitions. The pool
insurance policy that we negotiated with NMI will result in Fannie
Mae's loan-level exposure on the covered loans being reduced to
approximately 50 percent of the original property value, subject to a
pool deductible amount and an aggregate pool loss limit, as explained
below. The pool insurance policy sunsets after 10 years.
Similar to the C-Deal transaction, Fannie Mae will be responsible
for the first losses on the pool. We are insuring for ``unexpected
losses'' through the establishment of an aggregate pool deductible. The
deductible was set at 20 basis points of the initial balance of the
pool, or approximately $10.3 million of initial losses for which Fannie
Mae will be responsible.
The insurance policy will cover the next $90 million of claimable
losses. The aggregate pool loss limit was set to 2 percent of the
initial balance of the pool. At an approximate $5.17 billion initial
pool balance, the aggregate pool loss limit is approximately $103
million, including the deductible layer. Thus, once aggregate claimable
losses on the pool of loans exceed approximately $103 million, the
policy would terminate. Fannie Mae would be responsible for losses in
excess of the pool loss limit. The aggregate pool loss limit was set to
a level that exceeds our projected losses in a stress scenario
comparable to the recent experience (2006-2012).
This pool insurance credit enhancement has several advantageous
features. First, it will preserve the ability for lenders and Fannie
Mae to pool mortgage loans into a highly liquid TBA market. This will
enable the current efficient origination process, allow borrowers to
lock financing in advance and lenders to hedge that interest rate risk,
and lower mortgage rates for borrowers because of the liquidity of this
TBA market. A second advantage is that the mortgage insurance policy
form for the transaction preserves Fannie Mae's ability to pursue all
appropriate and needed loss mitigation which Fannie Mae deems
acceptable, e.g., loan modifications and short sales. Thus, servicers
follow standard Fannie Mae servicing protocols and service these loans
as they would other loans, irrespective of the credit enhancement.
We solicited insurance bids from seven MI companies, which are
currently approved by Fannie Mae to provide charter-compliant coverage
for loans that we acquire. In addition to the terms of the pool policy
structure, we stipulated that MI companies participating in these
transactions would need to meet certain counterparty requirements,
including a minimum level of statutory capital relative to their
outstanding risk in force. Six of the MI companies provided insurance
bids in response to our request. Only three of those MI companies would
have met the transaction counterparty requirements without needing to
raise additional capital.
We chose to commit the transaction to NMI for several reasons. NMI
had the materially lowest pricing of all six bidding MI companies,
agreed to cover all loans in the targeted pool, met our counterparty
capital requirements, and agreed to the terms of the new pool policy
form that we requested, including coverage certainty provisions that
provide for rescission relief.
One key observation was that although this coverage structure could
be repeated in the future, the mortgage insurance industry is currently
capital constrained. Pricing might need to rise considerably in
connection with possible future transactions, unless the MI industry is
able to raise new capital.
In conclusion, I appreciate this opportunity to present testimony
before the Committee and look forward to answering any questions you
may have.
PREPARED STATEMENT OF TED DURANT
Vice President of Analytic Services, Mortgage Guaranty Insurance
Corporation
December 10, 2013
MGIC History
Originally formed in 1957 by Milwaukee real estate attorney Max
Karl, MGIC was established to provide an innovative, affordable
alternative for families wanting to buy a home with less than a 20-
percent downpayment. In 1965 MGIC became the Nation's first publicly
traded mortgage insurer. Throughout its history MGIC has been at the
center of what has evolved into today's highly efficient secondary
market. In addition to providing mortgage insurance, MGIC created the
Nation's first private secondary market facility that brought buyers
and sellers together, was the first to insure mortgage backed pass-
through securities, and was the first to form a conventional mortgage
securities conduit. In 1971, MGIC created the first modern bond
insurance company. With a focus on sustainable home ownership, MGIC has
grown to provide a critical component of our country's residential
mortgage finance system, protecting mortgage lenders and investors from
credit losses.
The Roles of Mortgage Insurance and Bond Insurance in a New Housing
Finance System
We believe that, having survived the recent housing finance crisis
and saved taxpayers from $40 billion of additional losses, the private
mortgage insurance model has proven its value and should be a
fundamental component of a new housing finance system. The definition
of an Eligible Mortgage in S.1217 recognizes the benefits of loan-level
credit enhancement by requiring minimum coverage levels for low
downpayment loans. However, mortgage insurers can provide an important
additional role with the provision of pool-level mortgage insurance.
The role of bond insurer is different but is also an important
component of the new housing finance system. Bond insurance was adapted
for use in mortgage-backed securities (MBS) from the municipal bond
market. Importantly, bond insurers do not, like the GSEs, purchase and
aggregate loans for securitization. The role of the bond insurer is to
guarantee timely payment of principal and interest to the bondholders
in the event of failure of the issuer. In order to provide this
guaranty, bond insurers require that the risk of the issuer failing be
remote. Bond guarantors generally paid insufficient attention to loan-
level credit enhancement on the MBS they guaranteed leading up to the
housing crisis. Consequently, most of them failed. Nevertheless, the
role of guaranteeing timely payment of principal and interest is an
important one and, with the first-loss risk sufficiently transferred to
other entities, the bond guarantor model can work.
The new housing finance system envisioned in S.1217 can be improved
by:
1. Clarifying the distinction between mortgage insurance and bond
insurance, allowing for both loan-level and pool-level mortgage
insurance, and limiting bond insurance to its traditional role
of guaranteeing timely payment of principal and interest to
bondholders in the event of failure of the issuer, relying on
other, first-loss credit enhancement to ensure that the bond
guarantor is in a remote risk position.
2. Recognizing the loss absorbing resources of mortgage insurers in
the calculation of private capital at risk in front of the
FMIC.
3. Establishing a preference for entities over securities as a means
of ensuring a stable supply of capital through the cycle, and
relying more on existing regulators of those entities as a
means of clarifying the role of FMIC and avoiding a single
point of regulatory failure.
4. Clarifying the point of attachment of the Government guaranty,
and taking into consideration the application of the guaranty
to all forms of credit enhancement.
5. Broadening the bill to include comprehensive housing finance
reform that establishes consistent, uniform rules that apply
regardless of the source of funding for the loans.
Would the Bond Guarantor Business Be Attractive?
With those recommended improvements in place, we expect that there
will be interest from investors, including companies who write mortgage
insurance, to capitalize bond guarantors. However, we expect that the
insurance would be provided by separately capitalized and regulated
companies who might be jointly owned within a holding company
structure. In serving very different purposes, mortgage insurance and
bond insurance should be viewed as complementary, not competing forms
of mortgage credit risk enhancement. Nevertheless, they should be
managed, capitalized, and regulated separately.
The GSEs combined mortgage insurance, bond insurance, mortgage
aggregation, and securitization into two excessively powerful entities.
As a consequence they were both our largest beneficiaries and our
largest competitors. We would be hesitant about competing against
entities that are renamed GSEs, who are chartered to provide a
combination of mortgage insurance and bond insurance. With years of
data and experience as de facto regulators of the MI companies, and
currently being in the process of determining new eligibility
requirements for MI companies, rehabilitated GSEs set up to be mortgage
insurers would have an insurmountable advantage over existing or other
new mortgage insurers. Setting up the former GSEs as bond insurers
would likely limit investor appetite for creating competition in that
business. A healthy housing finance system that minimizes cost to the
consumers will require many bond guarantors and many mortgage insurers.
Attachment of the Government Guaranty
S.1217 creates two different ways in which the FMIC guaranty could
be triggered. In the case of structured finance, such as a senior/
subordinate structure, the guaranty would be triggered by the failure
of any individual security reaching the subordination level of losses.
In the case of a bond guarantor, the guaranty would be triggered by
failure of the entity.
Individual securities will be backed by a limited number of loans,
possibly all from the same lender and concentrated geographically. They
will certainly be originated within a narrow window of time. Thus,
individual securities will have a great deal of variation in their
performance, and their likelihood of reaching the 10-percent level will
be much greater than a collection of those securities. Using a vintage-
level loss trigger eliminates some of the potential lender and
geographic risk, but still has a higher likelihood of reaching the
trigger level than a collection of securities issued over many years.
As the point of attachment gets farther away from individual
securities, the attachment level needed to provide the same level of
protection to taxpayers decreases.
In general, the higher the attachment level, the greater the amount
of private capital that will be required and, consequently, the higher
the fees will need to be to provide the private sector guaranty. This,
of course, translates directly to higher costs to the borrower.
However, our appetite for participation as mortgage insurers or bond
guarantors depends not so much on the level of attachment as it does
the equality of the attachment level and capital requirements among all
competing forms of mortgage finance. A requirement for a 10-percent
subordination level for individual securities and a 10-percent capital
level for bond guarantors would make the bond guaranty business
uncompetitive until the next housing crisis, when investors in
subordinate tranches will again abandon the market.
Regulation of Bond Guarantors and Mortgage Insurers
Bond guarantors and mortgage insurers, being engaged in insurance
activities, are regulated by the States. Regulation of bond insurers or
mortgage insurers by FMIC raises significant State-Federal questions,
adds further complexity to the management of FMIC, and concentrates
oversight in a single point of failure.
We believe there are strong arguments in favor of maintaining the
existing system of State regulation and Federal oversight. Aside from
the political challenges of changing the State-Federal landscape with
respect to insurance, there are good reasons to separate the
responsibilities of regulation and prudential oversight from the
responsibilities of counterparty risk management. In the housing
finance system envisioned by S.1217, the bond guarantors would hold the
counterparty risk of the mortgage insurers, and FMIC would hold the
counterparty risk of the bond guarantors. Thus, FMIC is the ultimate
counterparty for both bond guarantors and mortgage insurers. As such,
it makes sense for FMIC to be responsible for issuing eligibility
requirements and monitoring compliance. Giving them full authority for
approval and prudential regulation, however, concentrates too much
responsibility in one entity that may have conflicting priorities. The
recent financial crisis demonstrated the importance of having multiple
points of oversight of mortgage insurers, with the majority of
companies surviving and continuing to fully pay valid claims.
Considerations for Choosing Risk Transfer Tools
Higher Risk Loans Require Loan-Level Guarantees
A fundamental principle in selecting a form of risk transfer is
that, the higher the level of the risk of the loans, the closer the
risk transfer should be to the loan level. Any loan with a significant
level of risk of loss should require loan-level credit enhancement
placed at origination by an entity that is involved in the
underwriting, origination, and servicing of the loan.
Safe loans, to borrowers with substantial downpayments and income,
steady jobs, and strong credit histories, do not require much
individual attention. There is very little credit risk to transfer and
the entities that acquire the risk can safely do so after the
origination of those loans. This lower risk segment of lending is where
pool MI and bond insurance are appropriate. This is also the segment in
which Fannie Mae and Freddie Mac have undertaken their recent credit
risk transfer transactions.
However, as the level of risk increases, it becomes progressively
more important to pay attention to the quality of underwriting and
verification of the offsetting factors that will help a borrower
overcome weak points in their qualifications. Entities that take the
credit risk on these loans must participate in and make their credit
decisions during the underwriting and origination process. This is a
distinguishing feature of loan-level private MI and an important source
of protection for the U.S. taxpayer. The use of subordinated tranches
and security-level guarantees for the securitization of subprime
mortgages, for example, produced disastrous results. The guarantors and
the investors in the bonds were too isolated from the underwriting and
origination of the loans to understand and manage the true risk they
presented. Loan-level mortgage insurance has been proven to reduce the
default risk on high LTV loans, demonstrating its effectiveness and
justifying its longstanding inclusion in bank capital requirements, GSE
charter requirements, QRM statutory language, and the S.1217 definition
of Eligible Mortgage.
Differentiation Must Be Maintained Between Mortgage Insurance and Bond
Insurance
Bond insurance and mortgage insurance serve two different purposes.
Mortgage insurance covers losses to the lender in the event a borrower
defaults. Bond insurance covers timely payment of principal and
interest to bond investors in the event an issuer defaults. The
guaranty of timely principal and interest requires substantial,
immediate liquidity in the event of an issuer default, so bond insurers
rely on other forms of credit enhancement to ensure that the likelihood
of a claim is remote. Mortgage insurance, on the other hand, involves
frequent claims at the loan level, but the time between borrower
default and the resolution of the claim is substantial (usually well
over a year), so liquidity is not as important as overall capital.
Those are different business models that require separately capitalized
entities for proper risk management.
Importantly, bond insurers should be kept in a remote risk position
through a combination of loan attributes and additional credit
enhancement. As long as that is the case, bond insurance and mortgage
insurance should be thought of as complementary, not competing, forms
of credit enhancement.
Entity-Based Enhancement Is More Stable Than Security-Based Enhancement
Another fundamental goal of housing finance reform should be to
ensure the proper supply of capital for mortgages through the economic
cycle. People like to refer to two states of the world--``Normal'' and
``Stress'' (See, for example, the presentation by James Stock to the
Urban Institute 11/13/13, available at http://www.urban.org/
UploadedPDF/412947-Cyclical-Stabilization-and-the-Structure-of-
Mortgage-Finance.pdf). From the perspective of a mortgage insurance
company, the period 2003-2007 is not normal, and we should not be
trying to get ourselves back to that state. If there is going to be
Government intervention in mortgage markets, the purpose must be both
to ensure sufficient liquidity in a stress and to prevent excessive
liquidity in ``normal'' times. This is only feasible if mortgage credit
risk is managed by entities that dedicate their capital, both human and
financial, to being in business through the cycle. The illusion of a
``best execution'' cost advantage of structured transactions is, in
reality, the mechanism that creates the boom-bust cycle, providing too
much credit in a boom, and no credit in a bust. Entities in the
business of creating structured transactions are motivated to make the
next deal, creating a very short-term focus on transaction volume.
Insurers, in contrast, are motivated to build and maintain a book of
insurance in force that is sized to the amount of capital they have
available. This capital level does not change quickly, creating a
significantly more stable level of funding capacity through the cycle.
Capital Requirements Must Relate Consistently to Risk Absorbed
While it is important for there to be a number of tools available
for mortgage credit risk transfer, it is also important for regulators
to ensure that capital treatment across the tools is consistent with
the risks they bear and the benefits they bring. It is a deceptively
simple matter to calculate the amount of ``private sector capital'' in
a senior/subordinate securitization structure. The level of
subordination marks a clear dividing point, and the subordinated bonds
represent true cash available to absorb losses at the initiation of the
transaction. In practice, the level of subordination can be highly
affected by prepayments. Complicated interest maintenance mechanisms
have to be in place to ensure the sufficiency of the subordination
level as the security is paid down. Shifting prepaid principal can
significantly alter the prepayment characteristics and, consequently,
the valuation of the senior bonds.
The equivalent protection offered by insurers is best measured as
loss absorbing capacity, which includes capital, reserves, and premium.
Capital and reserves are readily converted to an equivalent of cash.
While the forecast of premium is subject to some uncertainty, in
practice the forces that cause increased losses also generally cause
increased premium (also known as guarantee fees) through longer loan
lives. Projection of premiums and, thus, the calculation of how much
loss absorbing capacity is provided by an insurer should be the kind of
task easily performed by a competent regulator.
In addition to correctly calculating the capital requirements for
each form of risk transfer, it is vitally important that all of the
loss absorbing capacity be included in the calculation of private
capital at risk in front of the taxpayers. As written, S.1217 does not
appear to allow for credit for mortgage insurance, for example, to be
included in the calculation of whether there is 10-percent capital. The
result will be a significant understatement of the private capital at
risk. This will increase borrowing costs and create the disincentive
for use of anything other than minimal levels of MI, regardless of the
actual economics and amount of risk transferred.
Incentives and Moral Hazard
The phrase ``skin in the game'' is overused, but it describes an
important aspect of designing a sound housing finance systems. All the
participants must have some incentive to properly manage risk. Insurers
employ a variety of tools to manage the risk of moral hazard, in which
insurance beneficiaries have the incentive to behave in such a way as
to increase the risk to the insurer. Deductibles and coinsurance are
two commonly used tools. In mortgage insurance, limited depth of
coverage on primary loan-level insurance provides servicers with the
incentive to take proper care of delinquent borrowers and minimize the
loss severity on defaulting loans. Risk transfer tools should always be
designed to ensure that the potential for moral hazard is explicitly
managed. Retention of some amount of risk is frequently and
appropriately used to accomplish that.
Accounting True Sale and Consolidation
Discussion of risk retention in mortgage securitization must also
include consideration of the accounting issues of true sale and
consolidation. One of the primary values of securitization is to create
a source of funding that allows the lender to remove the loans from
their balance sheet. This occurs when the securitization transaction is
considered to be a true sale of the loans to the securitization entity.
Accounting rules, specifically FAS 166 and 167, describe the
circumstances under which a true sale of the loans has occurred, and
whether the loans must be consolidated back to the lender's balance
sheet even if it is considered a true sale. While moral hazard
considerations make it desirable for lenders to retain risk, true sale
and consolidation issues could cause those risk retention features to
make the mechanism unusable. Critical factors for ensuring the
securitization successfully transfers the risk include control of the
underwriting criteria, control of the servicing, beneficial interest in
the securities, and exposure to risk. In Ginnie Mae securitization, the
control of underwriting and servicing criteria by the insurers (FHA,
VA, RHS) and the position of the Government as the ultimate bearer of
risk make the Federal Government the consolidating entity, despite the
fact that the lenders, as Ginnie Mae issuer/servicers, retain a portion
of the risk. In GSE securitization, there is no question that a lender
selling a loan to the GSE constitutes a true sale and there is no
consolidation risk back to the lender. However, as private entities,
the GSEs should have to consolidate all the loans underlying their
guaranteed bonds back to their balance sheets. Under S.1217, it is not
yet clear whether the system envisioned would result in the Government
being the consolidating entity, or whether private entities would have
to consolidate. The resolution of that question will have a significant
impact on the feasibility of the system.
Background: The Fundamentals of Mortgage Risk
Mortgage loans are secured lending, meaning that the borrower has
pledged her ownership of her house as collateral in case she is unable
(or unwilling) to repay the loan. The risk to the lender, then, is
determined by both the likelihood of the borrower failing to make her
payments and, should that happen, the risk that the value of the
property will not be sufficient to pay off the remaining debt. The
likelihood of the borrower failing to pay off the loan is referred to
as default incidence by insurers and probability of default (PD) by
bankers. The amount that the lender loses, which is the difference
between the remaining debt and the value of the property, is referred
to as loss severity by insurers and loss given default (LGD) by
bankers.
Default incidence is driven by borrower circumstances, including
the amount of equity the borrower has in the property. When the
borrower purchases the home, the amount of equity is the downpayment.
Over time, the property may gain or lose value, the borrower may pay
down the loan, or refinance for a greater amount (cash out), or take
out a second mortgage. All of those events change the borrower's equity
in the home. If a borrower takes out additional financing and the home
loses value, the borrower may find himself in a position of negative
equity, also referred to as being underwater. A borrower with negative
equity, who might otherwise be able to afford to make his mortgage
payments, might choose to stop making those payments, in what is called
a strategic default. Under normal market conditions, most borrowers
default because of adverse changes in their personal circumstances,
such as job loss, death or disablement, or divorce.
Loss severity is driven by the amount that can be recovered by
selling the property, relative to the outstanding debt. In addition,
expenses associated with foreclosure, including legal fees, accrued
interest, and real estate maintenance and sale expenses, increase loss
severity. The longer it takes to complete the process, the greater the
loss severity. This results both from the increased expenses and
interest, and the deterioration of the property as homes in foreclosure
are typically not maintained properly.
Loss mitigation is the reduction of loss severity through a variety
of actions by the loan servicer to, first, keep the borrower in the
home and, second, minimize the amount of time it takes to resolve the
default. Keeping the borrower in the home often results in an improved
outcome for the lender. Techniques for doing this include forbearance,
in which some amount of the debt is delayed in repayment, and
modification, in which the term of the loan may be extended, the
interest rate reduced, or some portion of the debt forgiven. Another
loss mitigation approach is a short sale, in which the borrower and
lender agree to sell the property for a loss, and the lender then
either forgives the remaining debt or the borrower may agree to pay off
some portion of the remaining debt as an unsecured loan.
Fraud and misrepresentation are an additional risk in mortgage
lending that became more widely recognized in the recent financial
crisis. They are more accurately described as operational risk, not
credit risk, but they have a significant impact on credit risk.
Mortgage lenders, investors, and insurers all rely on representations
and warranties (reps and warrants) from other entities as to the truth
of the information on a mortgage application. Borrowers make
representations to lenders. Lenders make reps and warrants to investors
and insurers. Misrepresentation of facts, either unintentionally or
fraudulently, may significantly alter the credit risk of a mortgage
loan. For example, if a borrower makes $40,000 per year and the
application shows $480,000 per year, the borrower will have a
substantially greater likelihood of default than what would be expected
from the application. Note that this mistake could have occurred
because (a) the borrower lied, (b) the lender changed the information
without the borrower's knowledge, or (c) an annual income amount was
accidentally treated as monthly and multiplied by 12.
The consequences of fraud and misrepresentation have been
widespread, particularly in loans originated from 2005-2008. Mortgage
insurers that found material fraud and intentional misrepresentation
have rescinded coverage, meaning they canceled the coverage and
returned all premiums paid, refusing to pay insurance claims on those
loans. Mortgage investors like Freddie Mac and Fannie Mae have required
lenders to repurchase billions of dollars of loans. Financial
guarantors, having paid significant claim losses on guaranteed pools,
have sued and obtained billions of dollars in recoveries from issuers.
FHA has recovered billions of dollars from their lenders through
indemnification requests and through Government fraud suits involving
treble damages.
First-Loss Exposure and the Credit Risk Stack
First-loss exposure is a significant concept in secured lending.
Most defaults involve a recovery of some amount from sale of the
property or continued payment of the modified loan by the borrower.
Someone in a first-loss position takes losses regardless of how much is
recovered, assuming the recovery is not sufficient to pay off the whole
loan. For example, if a loan has a balance of $100,000, and the lender
recovers $70,000 from the sale of the home, the loss is $30,000. If the
lender has mortgage insurance that covers the first 25 percent of
losses, the insurer pays the first 25 percent x $100,000 = $25,000,
leaving the lender with the remaining $5,000 of loss. The remaining
loss is referred to as residual risk after the first-loss position.
Importantly, the first dollar of loss is more at risk than the next
dollar. This follows from the uncertainty of how much will be recovered
from the sale of the property, and the fact that each additional dollar
of recovery is less likely. Put another way, the first dollar of loss
will almost certainly be lost in a default, but the last dollar will
almost certainly be recovered.
This concept illustrates why downpayment is such an important
consideration in mortgage lending. The borrower's downpayment
represents the true first-loss position in the transaction. Losses to
the lender (and any insurers) only come after the borrower's equity is
used up. Greater borrower equity directly lowers the expected severity
in the event of default. And, as discussed earlier, greater borrower
equity, all else equal, lowers the likelihood of default, as well.
The credit stack is often used to illustrate the exposure to loss,
as shown in Figure 1. The various entities exposed to risk are shown in
a vertical stack, with the first-loss position at the bottom, and more
remote positions toward the top. If the borrower has equity, they may
be shown at the bottom. In this example, the borrower makes a 10-
percent downpayment, the mortgage insurer covers 25 percent of the loan
amount, and the investor (Freddie or Fannie) has the residual risk. The
amount covered by the mortgage insurer is referred to as the depth of
coverage. The farther you get from the bottom of the stack, the more
remote is your risk.
Structured securitization involves a similar concept, but it
operates on a pool of loans, rather than an individual loan. In a
senior-subordinate securitization, the monthly loan payments flow
through a waterfall, in which the senior bondholders are paid first and
the subordinate bondholders receive any remaining payments. In this
case, the subordinated holders bear 100 percent of the severity of each
loss, up to the point at which they have lost their remaining
principal. At that point, the senior bondholders begin bearing 100
percent of the severity of each loss. Figure 2 shows a securitization
credit stack, which typically does not include the borrower equity. In
the example, there is a 10-percent subordination level, in which two
subordinate bonds equal 10 percent of the total debt and senior bonds
equal 90 percent of the total debt. Just as in the case of the
individual loan, the higher you go in the stack, the more remote the
likelihood of a loss.
Forms of Mortgage Credit Risk Transfer
Mortgage credit risk transfer is typically done in two ways,
through entities and through structured transactions. These two are not
mutually exclusive, and in private securitization often both have been
used. Entity-based forms of risk transfer include mortgage insurance,
financial guaranty (bond insurance), and reinsurance. Structured
transactions include surplus notes, senior/subordinated (tranched)
securitization, and synthetic derivatives.
Entity-Based Forms of Credit Risk Transfer
Mortgage Insurance can be provided by Government insurers (FHA, VA,
RHS, PIH, Housing Finance Agencies) and by private mortgage insurers.
Mortgage insurance typically covers individual loans, though it may
also be used on pools of loans. Mortgage insurance is almost always in
a first-loss position or is used in combination with other mortgage
insurance that is in a first-loss position. Mortgage insurers control
their risk on loan-level insurance through limited depth of coverage,
which limits the severity risk but not the incidence risk. In other
words, the depth of coverage limits the losses on any individual loan,
but does not limit the number of loans on which losses may be paid.
Losses paid on one loan do not reduce the insurer's obligation to cover
the remaining loans. Pool insurance typically reverses that, covering
100 percent of the losses on each loan, but limiting the total losses
and, therefore, the total incidence. Once the coverage limit has been
reached, remaining loans are uncovered.
Standard private insurance coverage depth today is 30 percent for
loans with a 5-percent downpayment, 25 percent for loans with 10
percent, and 12 percent for loans with 15 percent. FHA insurance covers
100 percent of the losses, although their interest reimbursement
typically does not cover all of the accrued interest advanced by the
servicer. VA insurance generally covers 25 percent. RHS insurance
generally covers 90 percent.
Mortgage insurance generally requires that the servicer acquire
title to the property through foreclosure or complete a short sale in
order to file a claim. The insurer then adjusts the claim to ensure the
expenses are appropriate and the loss is calculated properly. The
insurer also may investigate the loan documents for evidence of fraud
or misrepresentation. After the insurer has made any required
adjustments and assuming they do not find fraud or misrepresentation,
they pay the servicer.
Financial Guaranty (Bond Insurance) can be provided by Government
entities (Ginnie Mae, for example) or private financial guaranty firms.
Private bond insurers, like mortgage insurers, are regulated by State
insurance departments. Bond insurance is distinguished from mortgage
insurance in that: it always operates at the pool level, never at the
loan level; it is always placed in conjunction with a securitization
transaction; it generally operates at a zero expected loss level, i.e.,
a remote level in which some other form of risk transfer is in the
first-loss position; and it covers the risk of default by the issuer of
the mortgage-backed security, not by the individual borrowers. Because
of that last factor, bond insurers begin making payments to bondholders
immediately on default of the issuer. If the bond insurer later finds
material fraud and misrepresentation, they must sue the issuer to
recover those losses. Bond insurance generally does not have any limit
on losses. This combination of features makes it very important that
bond insurers place their guarantees on pools of loans that are very
safe or sufficiently credit enhanced to make the bond insurer's risk
very remote.
Reinsurance can be provided by Government entities (e.g., TRIA for
terrorism risk, FCIC for crop insurance) or by private insurers. In the
private sector, there are global firms that specialize in providing
reinsurance across a variety of sectors and risks. They seek to
diversify across uncorrelated risks, so that their likelihood of facing
claims on multiple exposures at the same time is minimized. Like
financial guarantors, reinsurers generally operate at remote layers of
risk, with some other entity (typically the entity they are reinsuring)
taking the first-loss position. Government reinsurance is typically
used to cover true catastrophic risk such as terrorist attacks, floods,
or crop failure.
Mortgage insurers and financial guarantors have similar regulatory
rules that are different from other forms of insurance and from other
forms of mortgage banking. The primary feature of their regulation is
the capital and contingency reserve requirement. Each company must hold
a minimum amount of capital, relative to the risk insured. In addition
to that capital and case based reserves, which are specific reserves
for delinquent loans, the company must hold a contingency reserve. The
contingency reserve requirement is typically to hold 50 percent of
earned premiums for a period of 10 years. Funds may only be released
from the reserve in the event that losses exceed a significant level.
As a result, mortgage insurers and bond insurers have a unique
countercyclical capital requirement, forcing them to accumulate claims
paying resources in excess of their minimum capital requirement during
profitable periods, which may be drawn upon during periods of
significant stress.
Claims paying resources, or loss absorbing resources for an insurer
are the sum of their capital, their reserves (including the contingency
reserve), and the premium they receive from coverage renewal. These
resources form the ``private capital at risk'' when a mortgage insurer
covers a loan.
Structured Forms of Risk Transfer
Surplus Notes are a type of debt used by insurance companies to
transfer risk to the debt holders. They typically involve a variable
rate of interest that depends on the loss performance of the insured
risk. As losses to the insurer increase, the payments to the surplus
note holders decrease, offsetting the losses to the insurer.
Senior/Subordinate (Tranched) Securitization, as described earlier,
strips risk from the underlying loans and transfers most of that risk
to the subordinate bondholders, leaving the senior bondholders in a
more risk-remote position. Like pool insurance, once the subordinate
layer is used up, the remaining loans are no longer protected.
Synthetic derivatives transfer risk to investors through securities
whose performance depends on the performance of a reference pool of
loans. They are similar to surplus notes, in that they feature debt
securities that provide a variable rate of return based on the
performance of the reference pool. There is not an exact correlation
between actual losses and the performance of the pool, however. Recent
examples of this approach, like the Freddie Mac STACR transaction,
transfer losses to the investors at a fixed severity level when loans
reach a specified level of delinquency. As a result, investors are
insulated from the consequences, both positive and negative, of loss
mitigation and loss severity risk.
______
PREPARED STATEMENT OF SANDEEP BORDIA
Head of Residential and Commercial Credit Strategy Barclay's Capital
December 10, 2013
Good Morning, Chairman Johnson, Ranking Member Crapo, and other
Members of the Committee. My name is Sandeep Bordia and I am the head
of residential and commercial credit strategy at Barclays in New York.
My group covers research on mortgage credit markets in the U.S. and
Europe, including research on housing finance. I appreciate the
opportunity to discuss the fundamentals of transferring credit risk
from the U.S. taxpayer to the private markets.
In my remarks, I will start by describing the STACR and CAS credit-
linked deals (Freddie Mac and Fannie Mae risk-transfer deals),
including what has worked for these structures and what can be improved
going forward. I will also talk about the buyer base, the market's
appetite to absorb such issuance and how that would change if the
attachment point of the Government guarantee is higher. Finally, I will
compare and contrast the credit-linked note approach to two other
proposed structures: (a) the senior-sub structure; and (b) the bond
guarantor approach.
STACR/CAS Deals Overview
To begin with, let me talk briefly about the STACR/CAS deals
recently sold by the GSEs. So far, three deals have been priced, two
from Freddie Mac and one from Fannie Mae (for a total of $1.8 billion
in credit issuance). In each of these deals, the GSEs have retained the
risk on a 0.3-percent first-loss position and sold the credit risk on
the 0.3- percent to 3-percent loss piece. This means that the GSEs will
absorb losses on the first 0.3 percent of notional on the underlying
reference pool of loans for these transactions. Further, at the risk of
oversimplifying, the buyers of the issued securities will absorb losses
to the extent that they range from 0.3 percent to 3 percent of the
notional. In each case, the GSEs have also retained some small amount
of this 0.3-percent to 3-percent slice of risk while retaining the
right to reduce their ownership to as low as 5 percent by sales in the
secondary market. Appendix A shows a snapshot of the three deals. The
0.3-percent to 3-percent risk slice sold is broken into two parts, one
more senior than the other to better target the risk appetites of
various classes of investors. In all the three deals, the risk of
losses above 3 percent is borne by the GSEs and by extension, the
taxpayer.
The structures were very well received by the market with all three
deals oversubscribed many times over. The buyer base was fairly broad
with several dozen investors participating. Money managers dominated
purchases of the more senior of the two tranches on offer. Hedge funds,
money managers and REITs invested in the junior of the two tranches.
Insurance company involvement was somewhat limited due to uncertainty
around capital requirements on these tranches under the National
Association of Insurance Commissioners (NAIC) model-based approach.
Many investors were comfortable with the credit profiles and also used
financial leverage to buy these bonds.
What Has Worked for the Credit-Linked Note Structures So Far?
In our past published research, we have argued that to be
successful, any solution used to transfer mortgage credit risk to the
private market should have certain basic features. The solution should
preserve the well-functioning To-Be-Announced (TBA) market for
disseminating the interest rate risk on mortgages and allow mortgage
originators to hedge out their origination pipelines. The solution
should also be simple (to the extent possible), use existing financial
technology and be programmatic so as to attract a wide range of
investors.
In our opinion, the credit-linked note structure satisfies most of
these conditions. It allows the preservation of a liquid, well-
functioning TBA market, is simple for market participants to
understand, uses existing financial technology and is scalable into a
standardized program.
What Else Needs To Happen for This Program To Be Successful?
In our view, a few more things need to happen for this program to
be successful in the long run.
One, for GSEs (or any new entity) to be able to access a
well-functioning liquid credit market on a regular basis,
involvement from a broad range of investors is required. Since
there are fixed costs for investors to set up internal systems
to analyze and track performance of these deals, broader
participation requires a programmatic approach to issuance. In
other words, investors need to be confident that the deals are
not one-offs and the program is here to stay. We would also
caution against excessive experimentation with the structures
that may create a more fragmented marketplace and reduce
liquidity.
Two, expanding the type of collateral on which the credit
risk is sold is critical. The initial deals covered only the
cleanest portion of GSE originations that is not fully
representative of the collateral quality that GSEs or any such
entity would be expected to guarantee over time.
Three, in the long run, reducing the time between agency
MBS issuance and credit-risk transfer would help. The GSEs are
effectively warehousing the credit risk during that time
period. As such, shortening the window would reduce potential
taxpayer exposure. In addition, a shorter time window would
also allow for more timely market-based feedback into guarantee
fee pricing for future production. It might make sense to sell
even the 0-percent to 0.3-percent first-loss tranche as the
time between agency MBS issuance and credit-risk transfer
shrinks.
Market Appetite To Absorb the Risk
While the initial three deals have been heavily oversubscribed, the
amount of credit risk sold so far is miniscule in comparison to what
the GSEs have on their guarantee books. To put numbers in perspective,
a 3-percent to 4-percent loss tranche on a $5 trillion book would
translate into $150-200 billion of credit-linked notes (compared to the
$1.8 billion that was sold this year). We believe that the market can
absorb $5 to $10 billion next year without much disruption, and even
greater numbers in 2015 and later. For the program to get to a stage
where it can absorb much of the mortgage credit risk with GSEs, it
would realistically take several years of continued ramp up.
One big source of potential demand would be investors in legacy
nonagency MBS. There is currently about $850 billion (face value)
outstanding in the nonagency market. This is paying down at the rate of
$60 to $70 billion annually. Given strong mortgage credit expertise
among many of these investors, some of the paydowns they are receiving
would likely be reinvested in these securities. We could also see
additional interest from money managers and REIT-like entities.
What Is the Right Attachment Point for the Government Entity To Absorb
Losses?
Among other things, the attachment point for the Government entity
to absorb losses is a function of the policy goal and also the
collateral quality on which the credit risk is being sold. The
attachment point would be higher if the policy goal is to prevent
taxpayer losses even in extreme draconian scenarios. A 3-percent
attachment point might be reasonable for pools where the market expects
very low losses but would not be enough where base expectations are
close to or even higher than 3 percent. Generally speaking, a worse
quality pool of underlying mortgages would require a higher attachment
point and/or higher risk premiums for the credit-risk-transfer
securities.
For example, consider loans originated in Q3 2012, with an average
loan-to-value of below 80 percent. Since then, home prices have risen
another 10 percent to 15 percent around the country. As such, the
current loan-to-value ratio makes these mortgages even safer and a 3-
percent attachment point might be reasonable. In contrast, a 3-percent
attachment point on newer production with greater LTVs and no
accumulated home price appreciation might not be enough. This is
especially so because we learnt through the crisis that in a bad
economic environment, poor credit quality loans have losses that are
several multiples of the losses of good quality loans.
How Do We Think About the 10-Percent Tranche Proposed by S.1217?
As I mentioned earlier, a constant attachment point for all kinds
of collateral might not be reasonable, in our view. In a scenario where
we look at a 10-percent first-loss piece, the first thing to consider
is whether all of this would even be considered as a first-loss piece
by the market. So, while a 10-percent slice of a $5 trillion market
would equal $500 billion in mezzanine/subordinate bonds, not all of it
may be considered as deep credit investments and some may even receive
high investment-grade credit ratings. In other words, while more
credit-linked securities would need to be sold in the market, this
should mean that the buyer base could be expanded from what we have
seen on the STACR/CAS deals to include more risk-averse money managers
and insurance companies.
One number to consider is that, even at its peak, the total amount
of subordinate and mezzanine bonds outstanding in the nonagency market
in 2005-2007 was slightly below $400 billion. So, while it is certainly
possible for the private market to absorb $500 billion in supply, it is
by no means a done deal and would take a relatively long time to
achieve.
Other Approaches to Credit Risk Transfer
There are two other approaches that are being considered for
transferring credit risk from a Government-supported entity. The first
is to use a securitization style vehicle in the form of a senior-sub
structure. The second is to use well-capitalized bond guarantors to
cover losses.
Senior-Sub Structure Less Preferable
As we have recommended in the past, we prefer credit-linked notes
to senior-sub structure as they allow us to preserve the well
functioning TBA market as is. A senior-sub structure could also
increase the warehousing costs for originators if they were forced to
hold both the interest rate and credit risk until they accumulated
enough loans to issue a senior-sub deal. This could be particularly
problematic for smaller originators who may have to accumulate loans
for months before they could do a reasonably sized deal. In theory, it
would be possible to create a new TBA-like market just for the seniors
but it might orphan the existing TBA market, would likely be a
difficult transition and may have lower liquidity than the current set-
up.
Bond Guarantors as Providers of First Loss
Alternatives to selling credit risk in transactions like STACR/CAS
include using bond guarantors as providers of first loss. On the
positive side, this exit solution will likely provide more stable
funding for mortgage credit than securitization options (credit-linked
and senior-sub structures). The securitization option is likely to be
more procyclical, especially because of the availability of leverage to
investors in buying those securities. However, the bond-guarantor
structure also has two major drawbacks compared to the STACR/CAS
structures, in our view.
First, this form of insurance may result in some
counterparty credit risk. The STACR/CAS deals provide the GSEs
with cash equal to the face value of the first-loss piece sold.
This cash can be set aside to provide the GSEs with an actual
cash capital cushion in case losses exceed the threshold that
the GSEs have chosen. In the insurance/bond guarantee
transaction, the insurer does not have to pay this cash up
front but only if losses exceed a certain level. While S.1217
requires bond guarantors to hold capital equal to at least 10
percent of the guaranteed balance, this only works as a
safeguard if the bond guarantors' only business is to provide
insurance on these MBS. If the guarantor is involved in other
lines of business, unless the capital is held in a separate
account for the benefit of the enterprises or their successor,
the taxpayer still takes on some counterparty credit risk to
the guarantor. For example, if in certain extreme situations,
the losses on the guarantors' other lines of business exceed
the capital set aside for those business lines, there is some
risk that the insurers have to pay out using the capital
otherwise required to be held to cover mortgage losses. This
could potentially lead to a situation where some part of the 10
percent is not covered and the taxpayer is exposed to the risk.
Stronger oversight and regulations separating the capital held
for guaranteeing MBS could potentially mitigate this risk, but
would not eliminate it completely.
Second, the bond guarantee structures would not be as
transparent in pricing as the STACR/CAS deals since there would
be no secondary market to provide liquidity/pricing information
on an ongoing basis. The secondary market would provide more
immediate feedback to guarantee fee pricing than an insurance/
bond guarantee transaction could. A fully functional secondary
market in these credit tranches also provides useful
information that could allow a fully private market to price
credit risk in a more transparent manner and could help in
fostering a fully private market.
Conclusion
Overall, while we favor the credit-linked structure, given the size
of credit risk transfer required over the long run, it might be
preferable to have multiple exit options including through bond
guarantors. While I believe that there are various paths to achieve the
goal of transferring credit risk to the private market, I would caution
policy makers to closely watch the pace of any such transition. The
availability of mortgage credit remains extremely important to the
housing market and the economy as a whole and any sudden shocks to the
system that reduce this availability could have far-reaching
consequences.
Chairman Johnson, Ranking Member Crapo, and other Members of the
Committee, I thank you for your time and attention and the opportunity
to testify before the Committee.
PREPARED STATEMENT OF WANDA DELEO
Deputy Director, Division of Conservatorship, Federal Housing Finance
Agency
December 10, 2013
Chairman Johnson, Ranking Member Crapo, and Members of the
Committee, my name is Wanda DeLeo and I am the Deputy Director of the
Office of Strategic Initiatives at the Federal Housing Finance Agency
(FHFA). Thank you for the opportunity to appear before you today to
discuss the credit risk transfer activities we have asked Fannie Mae
and Freddie Mac, or the Enterprises as I will refer to them, to
participate in, particularly securities market sales of credit-linked
debt instruments. I'd like to start by recognizing the important work
this Committee has undertaken to redesign the Nation's housing finance
structure, including specifically the current work of the Chairman and
Ranking Member, the efforts of Senators Corker and Warner, and those of
their cosponsors, as well. We remain eager to help in any way we can.
More than 5 years into conservatorship, the Enterprises continue to
provide funding for roughly two-thirds of all new mortgages. Combined
with direct Government guarantees through FHA and VA, this amounts to
roughly 90 percent of new loans being supported by the Federal
Government. Enterprise losses since the financial crisis in 2008
required the Treasury to inject $187.5 billion of capital into those
companies. While the new loans they insure or guarantee are of much
higher quality than those that led to most of the losses, it is prudent
to seek alternative funding mechanisms that place less potential burden
on taxpayers. Our credit risk transfer program is designed to do
exactly that.
Improved housing market conditions, coupled with policy changes and
strong efforts of staff of both Enterprises to address still serious
deficiencies in their business operations, have enabled a welcome
return to profitability. But that should not blind us to the very real
costs associated with the Enterprises' failures. The dividends they
have paid to the Treasury reflect not a return of capital, but payment
for the extraordinary risk the Government was forced to take in view of
the potential at the time for economic disaster. The current earnings
are only possible because of the Treasury investment; no one even today
would be purchasing Enterprise debt in the absence of it.
It is in keeping with FHFA's responsibilities as conservator to
minimize taxpayer risks while helping to ensure the secondary mortgage
market continues to serve its functions. At the same time, we are
seeking to develop standards, norms, experience, and private investment
capacities that can continue into the future of a new secondary market
structure. Credit risk transfers can help us simultaneously in all
three of our broad conservatorship goals: build, contract, and
maintain. Accordingly, we have set a target for each of the Enterprises
to conduct multiple types of risk sharing transactions involving single
family mortgages with a total of at least $30 billion of unpaid
principal balances in 2013. We specified that the transactions be
economically sensible, operationally well-controlled, transparent to
the marketplace, and involve a meaningful transference of risk.
Further, we informed the Enterprises that our evaluation for assessing
their performance on FHFA's conservatorship scorecard objectives will
also consider the utility of the transactions to furthering the long-
term strategic goal of risk transfer. We will make final judgments
later this year, but clearly the transactions completed this year have
accomplished a great deal.
The Enterprises have initially focused on two broad categories of
credit risk sharing transactions. One transaction category is prefunded
capital markets transactions, which include Freddie Mac's Structured
Agency Credit Risk securities (STACRs) and Fannie Mae's Connecticut
Avenue Securities (C-deals). In these transactions, investors buy debt
securities that offer relatively higher returns if the credit
performance of loans in a reference pool is good, but may lose
principal when credit performance deteriorates. There is no
counterparty risk for the Enterprises because when investors buy the
securities, they are putting up cash that covers their maximum losses.
This approach offers efficient, competitive, market pricing of risk. It
also spreads risk across many investors with varying degrees of
leverage, and with varying degrees of risk concentration in mortgages.
Less risk concentration and less leverage has the potential to reduce
systemic risk relative to past and current practices that channel the
bulk of the risk into a very small number of highly leveraged
institutions, such as the Enterprises. A possible downside is that
overreliance on this approach may leave the market for risk more prone
to price change in response to changing market conditions.
The other transaction category for this year's Enterprise
transactions is insurance or guarantee agreements. In these, a mortgage
insurer, reinsurer, or other guarantor pays claims in the event of
loss. These deals can take advantage of such firms' mortgage expertise
and dedicated capital, and they may be less quick to leave the market
during a temporary market disturbance, especially one not directly
related to housing markets. However, this approach involves more
counterparty risk, more vulnerability to housing market weakness when
the counterparties are not diversified, and a more limited set of
bidders for the risk.
In both types of transactions, the Enterprises essentially use a
portion of their guarantee fee income from the reference pool to
purchase credit protection, either through higher interest rates paid
on the capital market transactions, or though premiums paid to
insurance companies. FHFA worked closely with the Enterprises on each
of this year's transactions, and in each case was confident that
conservatorship goals would be served. Reaching this point required
strong efforts by many over an extended period of time, and I want to
recognize the excellent work of the staffs of Fannie Mae and Freddie
Mac, including those sitting beside me today.
2013 Securities Transactions
This year, each Enterprise has sold debt securities that transfer
to private investors a portion of the credit risk of a large reference
pool of single-family mortgages that the Enterprise had previously
securitized. Freddie Mac has completed two STACR transactions to date,
and Fannie Mae has completed one C-deal. Each transaction provides
credit protection to the issuing Enterprise by reducing the principal
on the debt securities as credit performance of the reference pool
deteriorates.
Freddie Mac's transactions occurred in July and earlier this month.
In the July offering, the Enterprise sold $500 million in STACR notes,
resulting in credit protection on $18.5 billion of collateral
consisting of mortgages funded in the third quarter of 2012. In the
November offering, Freddie Mac sold an additional $630 million in STACR
notes, resulting in credit protection on $23.3 billion of collateral
that the Enterprise had funded in the first quarter of 2013. The STACR
notes are unsecured general obligations of Freddie Mac.
The credit event that results in losses on the STACR notes is
determined to occur if a loan becomes 180-days delinquent or there is a
third-party sale, a short sale, a deed-in-lieu at foreclosure, or a
sale of real-estate owned (REO) before 180-days delinquency. When such
a credit event occurs, a credit is calculated based on a tiered loss
severity schedule, where the severity increases with the cumulative
unpaid principal balance (UPB) of the underlying loans that experience
credit events. If calculated credit losses exceed 0.3 percent of the
UPB of the reference collateral pool, the principal of the STACR notes
is written down by the amount of the excess, until the calculated
losses exceed 3 percent and the remaining value of the STACR notes is
eliminated. In these initial transactions, Freddie Mac retained the
risk for the first 0.3 percent of UPB and any losses beyond 3 percent
in large part because of cost effectiveness considerations. Covering a
wider range of losses may be appropriate in the future.
In each STACR transaction, Freddie Mac sold notes that provide
protection on about four-fifths of the underlying loan pool to
investors, retaining the risk on the balance of the reference pool. The
Enterprise can elect to seek protection on some of the retained risk,
but has committed to maintain a minimum 5-percent interest in each
tranche of each deal. The risk-retention requirement is designed to
align the interests of Freddie Mac and investors that have bought the
STACRs.
The STACR notes have a final maturity of 10 years. With a fixed
loss severity and final maturity, Freddie Mac is exposed to some basis
risk on calculated credit losses on the reference pool from 0.3 percent
to 3 percent. In addition, the Enterprise retains exposure to the first
0.3 percent of calculated credit losses and to calculated losses beyond
3 percent.
In October Fannie Mae issued debt securities with a similar
structure. Specifically, Fannie Mae sold $675 million worth of
Connecticut Avenue Securities, resulting in credit protection on $25
billion of mortgages securitized in the third quarter of 2012. A
material difference compared to the STACR transactions was in the
tiered loss severity schedule. Further, one tranche of the Fannie Mae
security received an investment-grade rating from one credit rating
agency, and that was also achieved in Freddie Mac's second issue this
month.
Legal Issues Associated With the Security Structures
Both the STACRs and C-deal were issued as senior debt of Fannie Mae
or Freddie Mac. In each case, investors can rely on those Enterprises'
special credit standing, including the backing of the Treasury through
the preferred stock purchase agreements, for comfort that the payments
on the securities to investors will occur as specified in the terms of
the notes.
Part of the purpose of these transactions, though, is to develop
standardized credit risk investments that could be sold in the future
by securitizers other than the Enterprises. The Enterprises ultimately
hope to issue credit-linked notes through bankruptcy remote trusts that
would have the same economics for investors, but a different legal
structure that would not rely on an Enterprises' credit standing, but
rather on the trust holding and managing the proceeds from the note
issuance. Issues that arose include questions about whether issuers or
purchasers of the trust certificates would be commodity pool operators
under the Commodity Exchange Act, and whether issuers could be subject
to the conflict of interest rules under the Securities Act of 1933. We
are working with other agencies to resolve these questions, but
statutory clarifications might be helpful, and we are working with
Committee staff on possible solutions. We would also note that all of
these structures would be ineligible for REMIC tax treatment because
they would be considered synthetic structures. If they could obtain
similar treatment, the investor base for the securities would be
significantly expanded.
2013 Insurance Transactions
Both Enterprises have completed insurance transactions this year,
as well. In October Fannie Mae executed a pool insurance policy with
National Mortgage Insurance (National MI). The policy transfers a
substantial portion of the credit risk on a pool of single-family
mortgages securitized by the Enterprise in the fourth quarter of 2012.
The aggregate initial UPB of the loans in the pool was nearly $5.2
billion, and each mortgage had an initial LTV ratio of between 70
percent and 80 percent. Under the policy, Fannie Mae is responsible for
actual credit losses on the pool up to 0.2 percent and above 2 percent
of the initial aggregate UPB. National MI is exposed to credit losses
above 0.2 percent and less than or equal to 2 percent of the initial
aggregate UPB, but its exposure on each loan is limited to 50 percent
of its initial UPB. Thus, the policy has an aggregate loss limit of
about $103.4 million with a deductible of about $10.3 million. National
MI will pay claims based on actual credit losses determined after an
REO sale, short sale, or third-party disposition of the property. To
limit its counterparty risk, Fannie Mae has required National MI to
maintain a risk-to-capital ratio not to exceed 15:1 through 2015.
Thereafter National MI will maintain capital levels required by Fannie
Mae's then-applicable requirements.
In November Freddie Mac executed a transaction that transferred to
Arch Reinsurance, a global reinsurer, a portion of the residual credit
risk that the Enterprise had retained on the reference pool of
mortgages underlying the first STACR transaction. Specifically, Freddie
Mac had retained the credit risk associated with approximately $4.0
billion (about 18 percent) of the UPB of the reference pool. Under the
reinsurance transaction, Freddie Mac transferred the risk on $2.9
billion of that UPB to Arch, leaving the Enterprise with retained risk
on just over 5 percent of the total UPB, as required by the terms of
the STACR transaction. Because Arch insures diversified risks, its
financial health likely is less tightly tied to housing markets and
mortgage performance, so it may, other things equal, be better able to
pay mortgage claims in a severe housing stress environment.
Looking Forward
The Enterprises have executed transactions that transfer single-
family mortgage credit risk to capital-market investors and to firms in
the insurance industry. Each type of risk-transfer model has inherent
strengths and weaknesses. From an Enterprise perspective, the sale of
securities to capital-market investors provides up-front funding of
credit risk without posing any counterparty risk, while transferring
credit risk to an insurer leaves an Enterprise exposed to the claims-
paying ability of its counterparty.
From an overall housing finance system perspective, the leverage of
participating investors in capital markets transactions may not be
regulated, so there may be significant variation in the amount of
equity capital deployed to bear credit risk. Further, capital markets
funding sources maybe more volatile as a source of funding for mortgage
credit risk over the credit cycle. Transferring risk to the insurance
sector could be a more stable source of funding mortgage credit risk
over the cycle to the extent the financial strength and leverage can be
closely monitored either by the market or through regulatory
requirements.
Potential differences in the leverage of investors under the two
models also have implications for their relative cost. FHFA and the
Enterprises will continue to assess those strengths and weaknesses as
we explore both models of credit-risk transfer in parallel. Pricing on
all of the transactions this year has been attractive, suggesting that
each may be scalable to a significant degree. An increased volume of
issues next year will provide additional information about the depth of
demand.
A potentially powerful means of risk transfer is use of senior/
subordinate security structures. While none are expected this year, the
Enterprises have made progress in considering how such structures might
best work. In the process, they are grappling with many of the problems
faced by private label securities issuers of the recent past such as:
due diligence, representations and warranties, dispute resolution, and
the role of trustees. This approach has an advantage in that markets
have a good deal of familiarity with it, but the experience has been
less than satisfactory in many cases, particularly involving private-
label mortgage-backed securities. If good solutions can be found for
past problems, this approach may be easier than some others for non-
Enterprise issuers to adopt. A disadvantage to transferring losses on a
small pool of mortgages in a cash transaction, rather than on a large
reference pool in a synthetic transaction or insurance agreement, is
that credit evaluation costs can be considerably higher, as investors
must consider the idiosyncratic risks of a particular small pool,
rather than those of a cohort diversified by geography, lender, and
sheer size. Considering ways to develop more standardization and
liquidity in this market could help to address some of these issues.
The transactions considered so far have been Enterprise-centric in
that they depend heavily on the Enterprises' existing business
practices and the familiarity of loan sellers and investors with those
practices. To increase the potential generality of risk-sharing
approaches and reduce the dependence on the Enterprises, it may be
useful to explore the potential for loan sellers to arrange for credit
enhancements, such as those provided by securities or insurance before
the loans are sent to an Enterprise, rather than leaving it to the
Enterprise. Similarly, servicing and loss mitigation could possibly be
outsourced to firms specializing in those activities. Such changes
would not happen soon or quickly, but they merit consideration over
time.
Conclusion
The Enterprises have made major steps in risk transfer this year.
If sufficiently scalable, these transactions provide mechanisms to free
taxpayers from shouldering almost all the burden of mortgage credit
risk and place that risk in the private sector. We will, with the
Enterprises, continue to explore new techniques or variations on those
already tried to find the most workable solutions and those that show
the best promise of reducing the Enterprises' footprint, consistent
with maintaining efficient and effective mortgage markets. Thank you
and I am happy to answer any questions you may have.
RESPONSES TO WRITTEN QUESTIONS OF SENATOR CRAPO
FROM LAUREL DAVIS
Q.1. Many witnesses called for providing flexibility to the
FMIC in permitting different structures for the transfer of
credit risk. Can you elaborate as to why flexibility needs to
be provided to guarantors in their transfer of credit risk?
A.1. As the question above states, the issue of flexibility in
credit risk transfer transactions was raised several times
during the recent hearing. In responding to that question, I
was only addressing the desirability to permit flexibility for
guarantors when structuring particular transactions to transfer
credit risk to other credit investors. I was not intending to
address the more general question of whether a statute should
provide flexibility in how or in what amount private capital
should be required to attach prior to a Government guarantee.
As it relates to the need for flexibility for transfers of
credit risk by guarantors, to the extent that credit risk
transfer transactions proliferate, there are several reasons
why different structures may need to be used. They will need to
be conducted in different market environments, with different
investor demands that include potentially different pools of
collateral, and different investor bases (e.g., insurer's,
REITs, and money managers) which may have different tax,
regulatory and accounting needs. In order for the transactions
to be economically efficient for guarantors and attractive to
potential investors in credit risk, these differences may
require the creation of transactions with diverse structures
and will likely require different attachment points.
Accordingly, if Congress intends to draft legislation that
either requires or permits guarantors of mortgage credit risk
to transfer such risk, it would be highly desirable for such
guarantors to be granted significant flexibility in how they
structure such transactions.
Another aspect that should be considered related to
flexibility is the prevailing regulatory environment in which
guarantors will operate. Guarantors may need flexibility in how
they structure transactions to ensure compliance with such
requirements, including capital regulation. Federal financial
regulators have already opined on these issues as it relates to
other financial institutions.
------
RESPONSES TO WRITTEN QUESTIONS OF SENATOR REED
FROM WANDA DELEO
Q.1. On what date will the entire Common Securitization
Platform be ready to perform all of its functions?
A.1. The Enterprises are currently developing the Common
Securitization Platform (CSP), and have built the core
functionality and the related infrastructure components.
Preliminary testing is underway. The CSP's design and its
development have necessarily evolved over time, and a
significant amount of work remains with regard to both the CSP
itself and the business entity that will own it. The
Enterprises are engaged in developing and implementing
operational and business processes for the CSP and the joint
venture entity, and they are developing their integration plans
critical to the success of the CSP. Fannie Mae and Freddie Mac
are conducting this work under FHFA's guidance and with
industry input. Consequently, plans for this project will
continue to evolve as the Enterprises take into account the
many factors that will drive project success. The project plans
will not be finalized until the Enterprises, under FHFA's
oversight, are in a position to do so. As a result, we do not
yet have a date by which the Common Securitization Platform
will be operational.
Q.2. When the Common Securitization Platform is finally ready
to perform all of its functions, how much money, all in, will
have been spent in total by FHFA, each GSE, and Common
Securitization Solutions, LLC, including contracting costs?
Does this cost include the cost of any adjustments and upgrades
that may be necessary so that Fannie and Freddie can take
advantage of the Common Securitization Platform? If not, what
is this additional cost expected to be?
A.2. As discussed above, the Common Securitization Platform
project plans, inclusive of the design, build, and testing of
the technology and the Enterprises' system and process changes,
are being finalized. As a result, we have neither final plans
nor specific budgets assigned to these still-in-development
projects. To date, the following funds have been spent:
CSP and CSS: $65 million (1/21/2012-12/31/2013)
Fannie Mae Integration: $20 million (1/1/2013-12/
31/2013)
Freddie Mac Integration: $7 million (1/1/2013-12/
31/2013)
Q.3. FHFA staff has stated that FHFA ``has not prepared a
formal valuation analysis regarding the platform,'' which I
find disturbing, especially since taxpayer funds are
essentially at stake here and are in the process of being
spent. Should we be worried by the fact that FHFA is making
financial decisions with taxpayer funds without any ``formal
valuation analysis regarding the platform?''
A.3. FHFA understands your concern but believes that the
approach to the project has been prudent and well considered.
The project is consistent and aligned with many other projects
undertaken by the Enterprises, at the direction of the agency,
to achieve uniformity in areas essential to achieving an
effective mortgage securitization system. The Servicer
Alignment Initiative, Common Appraisal Data Portal, and Uniform
Mortgage Data Program are some of the projects that have
established common and uniform standards and practices in the
Nation's housing finance system, providing benefits not just to
the Enterprises, but also to other market participants.
The decision to engage the Enterprises in this project is
neither solely nor even principally a financial decision,
although the financial costs associated with it are very
important and being monitored. Rather, the decision is rooted
in FHFA's legal obligations, both as conservator and regulator.
The decision is based on achieving market efficiencies and
providing policy makers with options as they determine the
future of the U.S. housing finance system. The agency has
determined that the building and operation of the CSP would
also achieve many supervisory goals and realize other
significant benefits.
Q.4. Fannie and Freddie are still two distinct legal entities,
and FHFA acts as conservator for each GSE. Given how valuable
the Common Securitization Platform would be to each GSE on its
own, how did FHFA, as conservator for each GSE, determine that
a 50/50 joint venture was the right decision for each GSE? In
preserving and conserving the assets of Fannie with a view
towards putting it in a sound and solvent condition, why would
FHFA, as conservator for Fannie, give Freddie a 50 percent
stake in such a valuable asset?
A.4. As Conservator, FHFA decided that it was most beneficial
to establish common securitization technology, which would be
available to Fannie Mae and Freddie Mac and ultimately to all
market participants, rather than have each Enterprise
separately undertake extensive and proprietary infrastructure
projects. FHFA believes that building the CSP functionality
once, through the joint and collaborative efforts of, and its
use by, both Enterprises, will be more cost-effective than
having each Enterprise independently rebuild its core
securitization and servicing systems. Neither of the
Enterprises' existing systems would allow for relatively quick,
effective and efficient access by the industry either in the
near or medium term. Furthermore, independent and proprietary
Enterprise systems would not allow for uniformity across the
mortgage finance industry, thereby exacerbating the current
disarray within the industry and complicating the already
difficult task before policy makers. FHFA believes that two
different systems rather than common technology could seriously
delay or complicate attempts to reform the Nation's housing
finance system. Independent technology provides policy makers
with greater options for reforming the system than would a
rebuilding of the Enterprises' individual systems. FHFA and the
Enterprises have established a process to ensure that each
Enterprise's contribution to the joint venture is equitable and
fair retroactively and prospectively.
Q.5. Please provide all formation documents prepared in
conjunction with the formation of Common Securitization
Solutions, LLC (CSS), including but not limited to the
operating agreement, all legal opinions, all resolutions from
the Board of Directors for each of Fannie Mae and Freddie Mac
duly authorizing the formation of CSS, and documentation of all
costs incurred thus far and expected costs associated with CSS.
A.5. We would be happy to provide you and your staff an
opportunity to review the documents noted above at the FHFA
offices. Please contact Peter Brereton, Associate Director for
Congressional Affairs, if you would like to schedule such a
review, and if you require additional information or have
additional questions.
Additional Material Supplied for the Record
PREPARED STATEMENT OF LAURIE S. GOODMAN, DIRECTOR, HOUSING FINANCE
POLICY CENTER, THE URBAN INSTITUTE
Mr. Chairman, Ranking Member Crapo, and Members of the Committee,
thank you very much for the opportunity to testify today. My name is
Laurie Goodman, and I am the director of the Housing Finance Policy
Center at the Urban Institute. This is a new center, dedicated to
providing data-driven analysis of policy issues relating to housing
finance and the housing market. Prior to joining the Urban Institute
this past summer, I spent almost 30 years as a mortgage-backed
securities research analyst and as head of securitized products
research/strategy at several firms, including Amherst Securities Group
LP and UBS.
Recently, both Freddie Mac and Fannie Mae have completed deals in
which they transferred some of the risk from their guarantor book of
business to private investors. As we contemplate a new housing finance
system in which private entities take the first loss, backed up by a
catastrophic Government guarantee, the obvious question arises: to what
extent are these deals applicable to a new housing finance structure?
The answer is that there are lessons that can be learned from the
recent transactions, but the lessons are not completely transferable to
a new structure. This discussion is divided into four sections. The
first looks at the Freddie and Fannie risk-sharing transactions and
their impact in the current environment, where efforts are being made
to reduce the Government footprint. The second section looks at the
role of risk-sharing type structures in a guarantor/bond insurance
framework. The third section looks at the role of risk-sharing type
structures in a capital markets framework. The final section contains
my conclusions. To quickly preview my conclusions:
Regulatory relief through changes in the CFTC commodity-
pool rules is necessary to promote the use of credit-linked
notes.
Capital regulation for future guarantors should include
stress tests, a base capital ratio of 5 percent, a risk-based
capital component, and capital relief for credit-risk
transfers, subject to a minimum absolute capital requirement.
The system must have a guarantor (insurer) execution and
not rely solely on the capital markets to lay off credit risk.
Freddie and Fannie Risk-Sharing Transactions
In the absence of Government-sponsored enterprise (GSE) reform
legislation, the Federal Housing Finance Agency (FHFA) has attempted to
bring private capital back into the mortgage market. They have employed
a number of mechanisms and have contemplated others. These fit into
three main categories:
The FHFA has attempted to decrease the share of
originations purchased by the GSEs. They have raised guarantee
fees to encourage lenders to use other execution channels, such
as holding loans in portfolio or opting for a private-label
securitization. Guarantee fees at Fannie Mae have increased
from 28 basis points (bps) in the first quarter of 2012 to 58.7
bps in the third quarter of 2013, more than a doubling in an
18-month period. This has not been sufficient to curb the
reliance on the GSEs, but future guarantee-fee increases of 10-
20 bps could tip the execution of the highest quality loans to
bank portfolios, which could, in turn, result in adverse
selection to the GSEs. Private-label securitizations are much
more expensive than either GSE or bank executions at the
present time, and a considerably larger guarantee fee increase
would be required for this execution channel to be used.
Reducing loan limits is another lever that the FHFA has
considered as a way to reduce the GSE share.
The FHFA has contemplated vehicles that allow for risk
transfer at the point of sale (up-front risk sharing). The GSEs
would be permitted to accept loans with more credit enhancement
in exchange for lower guarantee fees. This can be done though
deep mortgage insurance (MI), through lender recourse, or
conceptually by allowing the lenders to arrange their own
capital markets transactions, similar to the risk sharing deals
that have been recently completed by Fannie and Freddie. The
Mortgage Bankers Association has proposed greater use of up-
front risk sharing. \1\
---------------------------------------------------------------------------
\1\ See, Mortgage Bankers Association ``Key Steps on the Road to a
Sustainable Secondary Mortgage Market'', 9/18/2013.
The FHFA has also required Fannie and Freddie to lay off
risk that is already on their books (back-end risk sharing).
The GSEs have tried three different methods for laying off this
risk: (1) capital markets transactions, (2) purchasing mortgage
pool insurance, and (3) purchasing reinsurance. Fannie and
Freddie have each done deals in which the mortgage credit risk
has been laid off via capital markets transactions. There have
been three deals to date: Freddie Mac's Structured Agency
Credit Risk deals (STACR 2013-DN1 and -DN2), and Fannie Mae's
Connecticut Avenue Securities deal (CAS 2013-C01). In August,
Fannie Mae announced it had tapped National Mortgage Insurance
Corporation to insure a pool of $5 billion of mortgages already
on Fannie's books. And in November, Freddie Mac executed a
transaction that transferred to Arch Reinsurance, a global
reinsurer, a portion of the credit risk that it had retained on
the first STACR transaction. The capital markets transactions,
and their applicability to the future state, are the focus of
this panel, but a few comments on their applicability in the
---------------------------------------------------------------------------
current state is also in order.
Fannie and Freddie's three capital markets risk-sharing
transactions have been very successful. The first deal was priced too
cheaply, as one would expect from a new asset class. This, however, had
the effect of enticing investors who did not participate in the first
deal to take a look at subsequent structures. Since the first deal,
there has been a move to tighter spreads as the asset class has gained
acceptance. For example, the M-2 tranche of the first STACR deal,
priced in July, sold at 715 bps over 1-month LIBOR, while the second,
priced on November 12, sold at a 425 bps spread. Table 1 shows details
for the three transactions, including the spreads at which the
securities were sold. At this point there is very substantial private-
sector interest, which is critical as policy makers look to the private
markets to take more mortgage credit risk.
These risk-sharing transactions, in conjunction with the other
actions being contemplated and taken, are a very valuable way to
contract the Government's footprint in the mortgage market while the
GSEs are in conservatorship. They can be done administratively at the
direction of the FHFA, and require no legislative action. We expect to
see many more of these transactions now that both GSEs have established
programs.
A Few Details on the Risk-Sharing Transactions
Before we delve into the applicability of these transactions in a
new, reformed housing finance system, it is important to underscore a
few specifics about these transactions.
The transactions are synthetic; that is, they reference the
relevant credit risk. The transactions are structured as unsecured
general obligations of Freddie Mac (for the STACR deals) and Fannie Mae
(for the CAS deal). The return of principal on the notes is tied to the
credit risk of a pool of residential mortgage loans (the reference
pool) owned or guaranteed by Freddie Mac (Fannie Mae). Freddie Mac
(Fannie Mae) is entitled to reduce the principal balance of the notes,
at a tiered severity percentage, when the loans in the reference pool
became at least 180-days delinquent or when another credit event
occurs. This tiered severity ranges from 10 percent to 40 percent in
the Fannie deal, and 15 percent to 40 percent in the Freddie deals.
Prepayments are generally passed through to the note holders pro rata
as a return of principal.
The deals were done as synthetic transactions because of the desire
to mimic the credit-risk transfer in a senior/subordinate structure.
Using an actual senior/subordinated structure would not be economical
because the senior bonds would not be eligible to trade in the ``to be
announced'' (TBA) market and thus would lose a considerable amount of
liquidity.
The FHFA has publicly stated that they want to expand the types of
deals being done to include senior/subordinated transactions. These
types of transactions make sense for collateral that is not eligible
for delivery into the TBA market. One of the largest buckets of non-TBA
eligible collateral is pools of jumbo loans, which are priced lower
than corresponding TBA securities. \2\ I expect there to be a senior/
subordinated transaction backed by jumbo collateral at some point in
2014.
---------------------------------------------------------------------------
\2\ Jumbo loans are those over the base GSE limit of $417,000; in
high-cost areas, where the limit is tied to area median prices, it now
ranges up to $625,500, and has been as high as $729,750. A de minimus
amount of these loans can be included in TBA pools; the balance must be
pooled separately. The collateral for a senior/subordinated deal would
be composed entirely from the non-TBA eligible jumbo loan pools.
---------------------------------------------------------------------------
The structures take the form of debt obligations, not credit-linked
notes. During the first half of 2012, as the planning for the risk
sharing began, it was expected that the structures would assume the
form of a credit-linked note with an embedded swap. In a credit-linked
note, the security is issued by a special purpose company or trust.
This special purpose vehicle (SPV) takes the initial proceeds of the
offerings and invests them in a risk-free instrument, such as U.S.
Government debt. The SPV simultaneously enters into a credit default
swap; under the terms of the credit default swap, the SPV receives an
annual fee based on the remaining principal balance, and makes
payment(s) if credit event(s) occur. Thus, investors will receive the
interest on the cash investment plus the annual fee. They may receive
less than a par return on their cash, depending on whether the level of
credit events was sufficient to impact their principal. A credit-linked
note structure is a pure bet on the risk being transferred; the credit
risk of its sponsor is not a factor.
In September 2012, the CFTC broadened its definition of ``commodity
pool'' to cover many transactions that include swaps. If the issuance
had been done as a credit-linked note, it would have fallen under the
broadened commodity pool definition. Commodity pools are subject to
reporting and regulatory burdens that I believe are inappropriate for
an instrument of this nature. For example, a nonexempt pool operator
must not only register as a Commodity Pool Operator (CPO), but the CPO
must become a member of the National Futures Association (NFA). Its
personnel must register with the NFA and pass an NFA exam. There are
also numerous reporting requirements designed to capture information
from entities operated for the purpose of trading commodities; it is
unclear how many of these can be applied to securitizations, which are
passive vehicles containing illiquid assets. This includes periodic
reporting concerning the commodity pool's changes in net asset value,
trading strategies, and performance data. It would also require the
securitization to name a Commodity Trading Advisor (CTA), and the
choice of this entity in a securitization is unclear: it could be the
sponsor or the trustee. The commodity pool registration could trigger
Volcker Rule prohibitions, making it difficult for banks to own these
instruments.
By issuing the obligation as Freddie Mac or Fannie Mae debt, rather
than from a SPV, these problems were avoided. The only difference
between doing these transactions as GSE debt rather than as credit-
linked notes is that the investor was exposed to the credit risk of the
sponsor. The STACR and CAS transactions contain exactly the same
embedded swap as they would in an SPV structure.
Investors are happy to buy these transactions as a debt issue
because they believe Fannie and Freddie are backed by the full faith
and credit of the U.S. Government, and hence they are making a decision
solely on mortgage credit risk, rather than a joint decision on the
mortgage credit risk and the strength of the underlying entity. In a
market in which the entity laying off the risk was not fully
Government-backed, the structure used for the STACR and CAS
transactions would definitely be more expensive, and might not be
viable. Thus, as we move away from a Fannie/Freddie-dominated world, it
becomes critical that the CFTC issues some form of regulatory relief so
these transactions could be done as credit-linked notes.
The structures reference well-diversified pools of loans. Freddie
Mac's STACR 2013-DN1 deal (see Table 1) had the smallest reference
pool, at $22.5 billion. This reference pool included all loans acquired
by Freddie Mac between July 1, 2012, and September 30, 2012, that met
the following criteria: (1) full documentation, 30-year fully
amortizing fixed-rate first-lien loans on one- to four-unit properties;
(2) originated on or after April 1, 2012, and securitized in Freddie
Mac PC prior to January 31, 2013; and (3) original LTV between 60 and
80. \3\ The other transactions referenced similarly broad groups of
loans. The sheer size and diversification suggests that there is little
idiosyncratic risk in these pools.
---------------------------------------------------------------------------
\3\ There were additional minor restrictions. Loans that were ever
delinquent, found to contain underwriting defects, or had prepaid in
full were excluded.
---------------------------------------------------------------------------
The GSEs are retaining some risk on these deals, giving them ``skin
in the game.'' In the three transactions that have been done to date,
the GSEs have retained the first-loss position as well as part of the
risk of the subordinate M1 and M2 tranches. For example, in the first
STACR deal, total subordination was 3 percent. The first-loss position
(B-H) was 30 bps, and the M1 and M2 slices were 135 bps apiece. Freddie
retained the entire B-H tranche as well as about 17.8 percent of the M-
1 and M-2 slices (the M-1H and M-2H tranches). This is important
because the GSEs have substantial control over the servicing practices,
and this helps assure investors that these loans will be serviced no
differently than anything else in the GSEs' portfolios.
The timing of these transactions is discretionary. Since these
assets are already on the GSEs' books, the risks can be laid off at any
time. If one of the GSEs was thinking of doing a deal, but market
conditions changed, it could be pulled until market conditions
improved.
The Future State of the Mortgage Market
There is a growing consensus the GSE reform is necessary. There is
also a growing consensus on two principles:
the private sector must play a far greater role in bearing
mortgage credit risk; and
continued Government involvement is essential to ensuring
that mortgages remain available and affordable to qualified
homebuyers throughout the business cycle.
Thus a number of proposals--including S.1217 (Corker-Warner); the
Bipartisan Policy Council's Housing Commission (BPC HC), \4\ of which I
was a part; and a paper coauthored by two of my Urban colleagues, Ellen
Seidman and Sarah Rosen Wartell, and by Phillip Swagel and Mark Zandi
(SSWZ), \5\--are aligned in proposing that the future state of the
mortgage market should consist of mortgage originators and servicers
who make the loans, a securitization platform, and a system of private
credit enhancement. The securitizer must arrange for the private credit
enhancement prior to securitization. There would be a limited
catastrophic Government guarantee, paid for up front, which would be
triggered only after all private capital available to support the
mortgages had been exhausted.
---------------------------------------------------------------------------
\4\ See, Bipartisan Policy Center Housing Commission, ``Housing
America's Future: New Directions for National Policy'', February 2013.
\5\ See, Ellen Seidman, Phillip Swagel, Sarah Rosen Wartell, and
Mark Zandi, ``A Pragmatic Plan for Housing Finance Reform'', Moody's
Analytics, the Milken Institute, and the Urban Institute, June 19,
2013.
---------------------------------------------------------------------------
The proposals suggest that private credit enhancement could take
two different forms: a guarantor (bond insurer) framework and a capital
markets framework. Both Corker-Warner and the BPC HC allow for both
mechanisms in the reformed system, while SSWZ allows for the insurer
alone. In all cases, the guarantor (insurer) is able to lay off risk
through risk-sharing arrangements. I believe the form of the private
capital will dictate the use and importance of risk-sharing
arrangements in the future state. Moreover, the decision as to what
form that first-loss piece takes is quite important and is not obvious;
each approach has its strengths and weaknesses.
The Guarantor/Bond Insurance Framework
In a guarantor framework, the bond insurer (guarantor) is liable
for the credit enhancement up to the amount of its capital (as long as
it is solvent). We assume GSE reform legislation would permit an
insurer to voluntarily decide to lay off some of the risk on its
transactions, and use structures similar to that being used in the
STACR and CAS deals. The bond insurer would essentially play the role
that Fannie and Freddie play in the current environment. This would
allow bond insurers to employ these structures when it is cost-
effective to do so, so the timing would be flexible. There will be
times when the capital markets bid will be lower than what bond
insurers require, and guarantors will likely to try to lay off risk
under such circumstances. This can be expected to occur when debt
financing is trading cheaply relative to equity financing. Guarantors
can also choose to lay off only part of the risk. This access to
capital markets execution would also allow the bond insurers to do
price discovery.
It is very likely that the bond insurer would dictate minimum
servicing standards in order to minimize its losses, just as Fannie and
Freddie do today. Investors would probably require that the guarantor
retain some skin the game, to gain assurance that the mortgage loans in
which the risk has been laid off are not treated any differently than
those in which the risk has not been laid off.
However, risk sharing by the bond insurers under any of the
proposals would be different from the STACR and CAS deals in one
important respect. Investors are happy to take mortgage credit risk in
synthetic form as embedded in Fannie Mae and Freddie Mac securities,
because they believe the underlying entities are backed by the full
faith and credit of the U.S. Government. If a bond insurer under any of
the current proposals were selling the risk, the investor would be
making a joint bet on the entity and the mortgage credit risk. The
investors that evaluate mortgage credit risk are not necessarily the
entities that take corporate credit risk. The result would be poorer
execution. I believe the securities would trade at more favorable rates
if one could separate the risk of the underlying entity (the corporate
credit risk) from the mortgage credit risk, and allow for the issuance
of credit-linked notes. This would require that transactions used to
transfer mortgage credit risk be exempt from commodity pool rules.
It is possible that the bond insurers would put together pools that
are poorly diversified, but I believe the market will be unwilling to
accept this. That is, investors are likely to believe that the insurer
has better information than they do about any given loan, so if a pool
is not a large representative sample, there is some chance the loans
have been adversely selected, and the securities are likely to be
priced accordingly. However, as long as the bond insurer itself is
adequately diversified, and liable as long as it is solvent, then this
is a business decision for a future bond insurers and investors in the
securities, and not an issue for a future regulator.
Sizing Capital Requirements
There are additional issues that would need to be dealt with in a
future state that are not considerations for the GSEs in
conservatorship. One of the most important is whether the institutions
receive capital relief for risk-sharing transactions. I assume this
would be the case, as these transactions would clearly allow the
institutions to operate safely with less capital. However, the amount
of any relief is more difficult to size than one might think. For
example, if the amount of capital relief is fixed, a bond insurer may
choose to transfer only its safest loans to the capital markets. The
result would be that after the transaction, the bond insurer would be
holding less capital than is prudent against a riskier set of loans.
This problem could be partially solved by requiring a minimum absolute
capital requirement, with a variable amount of capital relief for risk-
sharing transactions.
There has been a considerable amount of conversation about how much
capital is enough. Corker-Warner suggests 10 percent. The required
capital should be sufficient to allow the institution to withstand
severe stress. Many, including myself, believe the recent crises should
be used to size capital requirements. The data that has been provided
in support of the STACR and CAS risk-sharing deals has been invaluable
to market participants in assessing mortgage credit risk, and can be
used to size capital requirements. The data covers the performance of
Fannie's and Freddie's books of business from about 2000 onward.
Although it is limited to full-documentation fixed-rate amortizing 30-
year mortgages, excluding special affordability products, the structure
of the loans is very similar to the business the GSEs are currently
doing and are likely to do going forward.
My colleagues and I at the Urban Institute have analyzed this data.
\6\ We have shown that the 2007 Freddie Mac book of business, because
of the subsequent nearly 35-percent drop in home prices nationwide,
experienced cumulative ``defaults'' of 10.9 percent, the highest of any
vintage year. A ``default,'' or ``credit event,'' as it is referred to
in the risk-sharing transactions, is defined as a mortgage that went
180-days delinquent or was liquidated prior to going 180-days
delinquent by short sale, REO sale, or deed-in-lieu. To translate
defaults to losses, we assume a severity of 40 percent, the highest
number used in the three risk-sharing transactions, and multiply the
severity by the default rate to size losses. \7\ Freddie's 2007 vintage
10.9-percent default rate translates into a 4.4-percent loss rate. For
the Fannie book of business, the cumulative credit event rate was 14.1
percent, which translates into a 5.6-percent loss rate. Thus, it is
clear that 5-percent capital would have allowed the GSEs to weather
this very adverse environment. This is a conservative estimate in that
it applies the worst vintage year to the entire business.
---------------------------------------------------------------------------
\6\ See, Laurie Goodman and Jun Zhu, ``The GSE Reform Debate: How
Much Capital Is Enough?'', Urban Institute, October 2013, for more
detail.
\7\ The 40-percent severity covers the probability that a loan
that goes 180-days delinquent goes on to liquidate multiplied by the
severity if the loan eventually liquidates.
---------------------------------------------------------------------------
However, this analysis cannot be conducted in the abstract. It
cannot be divorced from the question of how many insurers there would
be and how to ensure that each one is adequately diversified. For
example, for an insurer who insured loans in only one State, even a 10-
percent capital requirement might be insufficient. Similarly, if the
regulator gave capital ``credit'' for credit-risk transfers, it would
be conceptually possible for an insurer to lay off almost all the risk,
keeping a small piece of nondiversified risk, which should demand
significantly more capital than would have been needed to support the
risk in the initial diversified book.
A regulator is unable to be perfect, and unforeseen events occur.
Thus, I suggest that some type of stress testing should be implemented.
Certainly, the Federal Reserve's stress testing of systematically
important banks has been a huge success. Stress testing would identify
insurers that are nondiversified or have laid off risk in a manner than
leaves them exposed. If the insurer failed the stress test, it should
be required to take corrective action promptly, including cutting off
dividends and raising capital within a well-defined period of time. If
the insurer failed to take these actions, or was unable to raise more
capital, it would be shut down, with the insurance transferred to
another entity (similar to a servicing transfer in the current
environment). This transfer could require a Government subsidy.
Thus, when sizing the capital requirements, it seems clear that any
new system relying on insurance should include the following:
a 5-percent base capital requirement;
additional risk-based capital requirements to cover
inadequate diversification or an unusually risky book of
business (e.g., an unusually high concentration of high-LTV or
low-FICO loans on the part of the insurer);
capital relief for risk-sharing transactions, subject to a
minimum capital requirement; and
stress testing of insurers, with corrective action required
if there is a stress test failure.
Because sizing capital cannot be divorced from diversification, the
ideal system should have a moderate number of well-diversified insurers
that compete with each other. Too few institutions would limit
competition and raise ``too big to fail'' concerns; too many are likely
to be insufficiently diversified and operationally inefficient. I do
believe it is necessary to have minimum diversification requirements.
The design of a new system must also consider the role of
traditional MI providers. By statute, Fannie and Freddie are required
to lay off the risk on any mortgage greater than 80 LTV; that is, they
are not permitted to bear this risk. Thus, a mortgage insurance
industry has been established to take the risk on all mortgages over
80-percent LTV. Under Corker-Warner, the mortgage insurers would stay
largely as they are, though they would cover loans down to about 70-
percent LTV, and they could not also be bond guarantors. In my opinion,
if one is remaking the system, it should not be taken as a given that
the mortgage insurance industry remains as is. Both the bond insurers
and the mortgage insurance companies would be assuming the same
mortgage credit risk, hiring people with similar skills, and developing
models to evaluate the credit risk. It may make more sense to combine
the functions. However, if the functions were combined, it would be
prudent to require bond insurers to hold more capital than in a system
in which these functions remain separate.
In short, the guarantor model can easily employ the risk-sharing
techniques used in the STACR and CAS transactions, but the capital
credit that is given for doing so cannot be divorced from the question
of how capital requirements should be sized for the guarantors in a
future system.
The Capital Markets Framework
Under a capital markets framework for a new housing finance system,
each security would require credit enhancement at the time of the
securitization via a senior/subordinated bond market transaction or a
synthetic alternative like Freddie and Fannie are using. The
transactions would be more natural in nonsynthetic form. However, since
there is no insurance entity, the transaction could not be done as a
debt issuance; it would have to be a credit-linked note structure, with
the originator as the deal sponsor. This, in turn, creates the problem
that the deal is considered a commodity pool, an issue discussed
earlier.
Any capital markets solution must be combined with an insurance
solution for several reasons. First, a pure capital markets solution
relies on only one source of capital (the debt markets), with no
flexibility to also rely on equity capital to take mortgage credit
risk. Second, it requires the pricing of the credit enhancement to
occur simultaneously with the securitization. There may be times when
the pricing is quite unfavorable, and the originator would be forced to
take that pricing, leading to more volatile mortgage rates for
borrowers. Third, this structure is not kind to small lenders, who will
have trouble aggregating a large enough pool of loans to obtain the
required enhancement. That is, a small pool is almost by definition
nondiversified and would command unfavorable pricing. Additional
concerns include regulatory efficiency and effectiveness, the viability
of the mortgage market during times of market stress, and ensuring
broad-based credit availability. As a result, none of the GSE reform
proposals advocate exclusively a capital markets solution. Corker-
Warner allows for both capital markets and guarantor arrangements, as
does the BPC HC proposal. The SSWZ proposal does not allow for capital
markets execution except in allowing an insurer to lay off some risk.
If the capital markets solution is meant to coexist with the
guarantor solution, the capital markets structure must either (1) use
synthetic structures, or (2) if one were to use cash structures, the
market must agree the senior bonds in the senior/subordinate structures
are eligible for TBA delivery. This would require the market to
simultaneously accept, as TBA, mortgages that reflect the full cash
flow stream, as well as those in which the subordinated cash flows are
not included. I am not sure the market will allow for this degree of
flexibility.
Diversification
The biggest problem I see with the capital markets solution, which
is often overlooked, is that the interaction between the required
capital and diversification is quite complicated, and hard to get
right. Let us assume we fix the capital requirements at Corker-Warner's
10 percent. As discussed in the previous section, our work at Urban has
shown that under most circumstances a 5-percent capital requirement is
more than enough. A 5-percent capital requirement would have been
sufficient to cover, for their entire book of business, the losses
Fannie and Freddie incurred on the 2007 origination activity, a vintage
that experienced 35-percent home price declines. However, in a capital
markets execution, even if you require 10-percent capital, it is easy
to construct pools where inadequate diversification due to either size
or geography means that is not nearly enough. For example, we looked at
1,000 randomly selected pools of 2,500 Freddie Mac loans from the 2007
vintage and found the mean default rate would have produced losses in
the 4.0 percent to 4.5 percent area, with default rates on individual
pools up to 1 percent higher and lower. \8\ But a pool of only Arizona
loans had a much higher default rate than the Nation as a whole; the
mean loss rate, using 40-percent severity, was in the range of 9.0
percent to 9.5 percent. When we looked at 1,000 randomly selected pools
of 2,500 loans, the loss rates ranged from 8 percent to 10.75 percent.
When we looked at 1,000 randomly selected smaller pools of 100 loans,
the loss rates of the pools ranged from 4.4 percent to 16 percent.
Clearly, individual pools will be smaller than all the loans in a given
Fannie or Freddie vintage, and the Government should want more first-
loss capital to come before its guarantee when pools are small,
nongeographically diversified, or nondiversified in other ways. In
contrast, the STACR and CAS securities were large, well-diversified
pools, eliminating any nonsystematic results.
---------------------------------------------------------------------------
\8\ Laurie Goodman and Jun Zhu, ``The GSE Reform Debate: How Much
Capital is Enough?'', Urban Institute, October 2013. In this article,
we expressed the numbers as default rates, we have converted to
severities for the purpose of this exercise.
---------------------------------------------------------------------------
Thus, if a reformed system mandates a fixed amount of capital for
capital markets transactions, it must also mandate diversification
requirements. The new regulator would set the diversification
requirements, but these requirements will be very hard to calibrate.
Instead of a fixed capital standard, a risk-based approach could be
used. One can imagine a system in which the originators enter the loan-
by-loan composition of a proposed pool into a system provided by their
regulator, and the system tells them the capital necessary to support
that pool. It's the equivalent of buying tomatoes at the supermarket,
then bringing them to be weighed. It is cumbersome, and creates some
pricing uncertainty. It also means that different pools will have
different amounts of credit enhancement, and the redesigned TBA market
must be willing to accept this.
These diversification issues are further compounded for smaller
originators. It is not clear how they get the number of loans needed
for a capital markets offering. Even if they could get a critical mass
of loans, those loans are apt to represent insufficient
diversification. It is unlikely smaller lenders could utilize a capital
markets solution without an aggregator. If there is a parallel
guarantor execution vehicle, this issue is still difficult, although
less critical. In short, the interaction between required capital
(subordination) and diversifications is complicated, and there is no
silver bullet.
Conclusion
The STACR and CAS transactions have clear applicability in any new
housing finance system. If a guarantor structure is used, the guarantor
plays the role of Fannie and Freddie. The largest issue relating to
risk transfer in this model is what credit the guarantor will receive
for laying off these risks on the capital markets. I definitely believe
credit should be given, as transferring risk allows the guarantors to
operate safely, with less capital.
If the new housing finance regime allows for both a guarantor
execution and a capital markets execution, the big issue with risk
transfer is how to ensure adequate diversification to protect the
Government. If a fixed capital requirement is mandated, the regulator
must ensure adequate diversification on each individual pool. If a
risk-based capital requirement is used, it requires substantial
calibration on the part of the regulator, as well as some uncertainty
for the lender, as the lender does not know the capital requirement
until the pool is final.
More generally, three conclusions emerge from this analysis:
First, it is important to resolve the commodity pool issue so that
synthetic structures using credit-linked notes, which allow for
separation of the risk of the sponsoring entity from the credit risk on
the mortgages that are being transferred, can be used.
Second, in any future state, capital regulation for guarantors
should include stress tests, a base capital ratio of 5 percent, a risk-
based component, and capital relief for credit-risk transfer, subject
to a minimum absolute capital requirement.
Third, in the future state, the system must have a guarantor
(insurer) execution, and not rely solely on the capital debt markets to
lay off credit risk. This is necessary in order to promote the TBA
market, allow for the presence of small lenders, assure broad-based
credit availability of credit, and be resilient during periods of
market stress.