FRBNY Economic Policy Review / July 2012 35
The Role of Bank Credit
Enhancements
in Securitization
1.Introduction
oes the advance of securitization—a key element in the
evolution from banking to “shadow banking” (Pozsar et al.
2010)
1
—signal the decline of traditional banking? Not
necessarily, for banks play a vital role in the securitization
process at a number of stages, including the provision of credit
enhancements.
2
Credit enhancements are protection, in the
form of financial support, to cover losses on securitized assets
in adverse conditions (Standard and Poor’s 2008). They are
in effect the “magic elixir” that enables bankers to convert
pools of even poorly rated loans or mortgages into highly rated
securities. Some enhancements, such as standby letters of
credit, are very much in the spirit of traditional banking and
are thus far from the world of shadow banking.
This article looks at enhancements provided by banks in the
securitization market. We start with a set of new facts on the
evolution of enhancement volume provided by U.S. bank
holding companies (BHCs). We highlight the importance of
bank-provided enhancements in the securitization market by
comparing their market share with that of financial guaranties
sold by insurance companies, one of the main sellers of credit
protection in the securitization market. Contrary to the notion
1
According to Federal Reserve Chairman Bernanke (2012), “Examples of
important components of the shadow banking system include securitization
vehicles.”
2
See Cetorelli and Peristiani (2012) for analysis of banks’ role in other steps
in the securitization process.
that banks were being eclipsed by other institutions in the
shadow banking system, we find that banks have held their
own against insurance firms in the enhancement business.
In fact, insurers are forthright about the competition they
face from banks:
Our financial guaranty insurance and reinsurance
businesses also compete with other forms of credit
enhancement, including letters of credit, guaranties and
credit default swaps provided, in most cases, by banks,
derivative products companies, and other financial
institutions or governmental agencies, some of which have
greater financial resources than we do, may not be facing
the same market perceptions regarding their stability that
we are facing and/or have been assigned the highest credit
ratings awarded by one or more of the major rating agencies
(Radian Groups 2007, form 10-K, p. 46).
Given the steady presence of bank-provided enhance-
ments in the securitization market, we next study exactly
what role enhancements play in banks’ securitization process.
The level of credit enhancements necessary to achieve a given
rating is determined by a fairly mechanical procedure that
reflects the rater’s estimated loss function on the underlying
collateral in the securitization (Ashcraft and Schuermann
2008). If estimated losses are high, then—all else equal—
more enhancements are called for to achieve a given rating.
Those mechanics suggest a negative relationship between
Benjamin H. Mandel is a former assistant economist and Donald Morgan an
assistant vice president at the Federal Reserve Bank of New York; Chenyang
Wei is a senior economist at the Federal Reserve Bank of Philadelphia.
Correspondence: don.morgan@ny.frb.org
The authors thank Nicola Cetorelli, Ken Garbade, Stavros Peristiani, and
James Vickery for helpful comments and Peter Hull for outstanding research
assistance. The views expressed are those of the authors and do not necessarily
reflect the position of the Federal Reserve Bank of New York or the Federal
Reserve System.
Benjamin H. Mandel, Donald Morgan, and Chenyang Wei
D
36 The Role of Bank Credit Enhancements in Securitization
the level of enhancements on a deal and the performance of
securitized assets. Note that in this scenario, enhancements
serve as a buffer against observable risk (as embodied in the
estimated loss function).
We are interested in the idea that enhancements might also
be used to solve part of the asymmetric information problems
that may plague the securitization process. If banks are better
informed than outside investors about the quality of the assets
they are securitizing, as they almost certainly are, banks that are
securitizing higher-quality assets may use enhancements as a
signal of their quality. In other words, by their willingness to
keep “skin in the game” to retain some risk, banks can signal
their faith in the quality of their assets. Such signaling implies
a positive relationship between the level of enhancements and
the performance of securitized assets, just the opposite of the
buffer explanation. Obviously, enhancements could, and
probably do, serve both as a buffer against observable risk
and a signal against unobservable (to outsiders) quality.
However, since the buffer role is almost self-evidently true,
we are interested in whether we can detect any evidence for
the role of securitization enhancements as a signal.
Others have also considered the hypothesis that
enhancements might play a signaling role. Downing, Jaffee,
and Wallace (2009) observe that asymmetric information
about prepayment risk in the government-sponsored-
enterprise (GSE) mortgage-backed-security market should
motivate the use of signaling devices.
3
Albertazzi et al. (2011)
note the potential centrality of asymmetric information to
the securitization process and conjecture that a securitizing
sponsor can keep a junior (equity) tranche “as a signaling”
device of its (unobservable) quality or as an expression of a
commitment to continue monitoring. James (2010) comments
that if asset-backed securities include a moral hazard (or
“lemons”) discount due to asymmetric information, issuers
have an incentive to retain some risk “as a way of
demonstrating higher underwriting standards.”
4
A variant of the question we are asking about credit
enhancements showed up in earlier literature on the role
of collateral in traditional (on-the-books) bank lending.
A theoretical literature in the 1980s predicted that in the
context of asymmetric information, safer borrowers were more
likely to pledge collateral to distinguish themselves from riskier
ones (Besanko and Thakor 1987; Chan and Kanatas 1985).
However, an empirical study by Berger and Udell (1990) found
strong evidence against the signaling hypothesis: that is,
3
Because the mortgage-backed securities that the authors study are
guaranteed, prepayment risk is the only risk investors need to worry about.
4
In a paper that is somewhat related to ours, Erel, Nadauld, and Stulz (2011,
p. 37) investigate why banks hold highly rated tranches of securitizations,
and conclude that their doing so may partly serve as “a credible signal of deal
quality to potential investors.”
collateral was associated with riskier borrowers and loans.
In other words, when it comes to loans on the books, collateral
seems to serve more as a buffer against observable risk than
as a signal of unobservable quality.
We found only one other paper that looks at the relationship
between enhancements and the performance of securitized
assets. Using loan-level data, Ashcraft, Vickery, and
Goldsmith-Pinkham (2010) find that delinquency on
underlying subprime and Alt-A mortgage pools is positively
associated with the amount of AAA subordination.
5
Those
results are consistent with the hypothesis that subordination is
used as a buffer against observable credit risk. Interestingly,
however, the authors find that BBB subordination is negatively
associated with mortgage performance on Alt-A deals, which
they consider more opaque (hard to rate). The latter result
seems consistent with the signaling hypothesis: the issuer of an
opaque security submits to a high degree of subordination to
signal its confidence in the quality of the assets it is selling.
We investigate our question from two angles. First, we look
directly at the relationship between the performance of
securitized assets and total enhancements in a panel analysis
where we regress the fraction of securitized assets that are
severely delinquent (delinquent for ninety or more days or
charged off) on total enhancements per unit of securitized
assets. We estimate the regression for seven categories of credit:
residential real estate loans, home equity loans, credit card
loans, auto loans, other consumer loans, all other loans, and
total securitizations. We are not able to detect any evidence
for the signaling hypothesis; when we find a significant
relationship between delinquency on securitized assets and
enhancements, the relationship is positive, consistent with
the buffer hypothesis.
In the second part of our article, we test the hypotheses
from the perspective of market participants. Specifically,
we investigate how stock investors and the option market
reacted when BHCs detailed for the first time their
securitization activity in their 2001:Q2 regulatory reports,
which include enhancements and aggregate loan performance
(delinquencies) of the assets that BHCs securitized. We
calculate the cumulative abnormal stock return around that
date for each BHC that had positive securitization activity.
We find first that abnormal returns are highly positively
correlated with the extent of securitization activity at a
BHC. That comes as no surprise, since securitization was
presumably viewed at the time as positive net-present-value
(NPV) activity. More interestingly, we find that the
relationship between total credit enhancements and
5
The amount of subordination at a given rating is the fraction of bonds that
absorb losses before the bond in question. If 90 percent of the bonds in a deal
are senior AAA bonds and 10 percent are junior, subordination of the AAA
bonds is 10 percent.
FRBNY Economic Policy Review / July 2012 37
cumulative abnormal returns depends on the delinquency rate
on securitized assets; when the rate is below some threshold,
cumulative abnormal returns are positively correlated with
total credit enhancements. This result suggests that when the
delinquency rate is relatively low, enhancements serve as a
signal of quality (hence, the high cumulative abnormal
return). However, when the rate is above that threshold,
the relationship between enhancements and cumulative
abnormal returns becomes negative. This finding suggests
that when the delinquency rate is relatively high—meaning
that securitized assets are demonstrably risky—enhancements
serve as a buffer against observable risk.
We also examine how securitization activity and
enhancements are related to BHC risk, as measured by
the implied volatility of BHC stock prices. We find that
securitization activity is positively correlated with implied
volatility, suggesting that markets view securitization as a risky
activity. We also find that total enhancements are positively
related to implied volatility. This result implies that just as
traditional originate-and-hold banking exposed bank
shareholders to risk, so does banks’ provision of credit
enhancements.
2. Background on Bank-Provided
Credit Enhancements
While credit enhancements can take many forms, Schedule
HC-S, on which BHCs report on their securitization activity,
includes fields for three types of enhancements.
6
The first is
credit-enhancing, interest-only strips. Schedule HC-S
instructions define these strips as:
an on-balance-sheet asset that, in form or in substance,
1) represents the contractual right to receive some or
all of the interest due on the transferred assets; and
2) exposes the bank to credit risk that exceeds its pro-rata
share claim on the underlying assets whether through
subordination provisions or other credit-enhancing
techniques.
Elsewhere, the HC-S instructions note that the field for
credit-enhancing, interest-only strips can include excess spread
accounts.
7
Excess spread is the monthly revenue remaining on
6
To be clear, our article focuses on the three types of enhancements reported
by bank holding companies on Schedule HC-S. For a more general discussion
of enhancements, see Ashcraft and Schuermann (2008).
7
Levitin (2011, p. 16) asserts that, in the context of credit card securitization,
excess spread accounts are also referred to as credit-enhancing, interest-only
strips.
a securitization after all payments to investors, servicing fees,
and charge-offs. As such, excess spread—a measure of how
profitable the securitization is—provides assurance to
investors in the deal that they will be paid as promised. Excess
spread accounts are the first line of defense against losses to
investors, as the accounts must be exhausted before even the
most subordinated investors incur losses.
The second class of enhancements, subordinated securities
and other residual interest, is a standard-form credit
enhancement. By holding a subordinated or junior claim, the
bank that securitized the assets is in the position of being a first-
loss bearer, thereby providing protection to more senior
claimants. In that sense, subordination serves basically as a
buffer or collateral. However, in the asymmetric information
context, holding a subordinate claim gives the bank the stake
that can motivate it to screen the loans carefully before it
securitizes them and to continue monitoring the loans after it
securitizes them. The bank’s willingness to keep some risk may
serve as a signal that it has screened loans adequately and plans
to monitor diligently.
The third class of enhancements, standby letters of credit,
obligates the bank to provide funding to a securitization
structure to ensure that investors receive timely payment on
the issued securities (for example, by smoothing timing
differences in the receipt of interest and principal payments) or
to ensure that investors receive payment in the event of market
disruptions. The facility is counted as an enhancement if and
only if advances through the facility are subordinate to other
claims on the cash flow from the securitized assets.
8
Although not technically classified as an enhancement, a
fourth item on Schedule HC-S that we consider is unused
commitments to provide liquidity. Unused commitments
represent the undrawn balance on previous commitments.
We include this variable simply as a control; we do not venture
a hypothesis about how it will enter any of our regressions.
It is important to note that the HC-S data we study,
particularly subordination, are measures of risk retention by
BHCs and not necessarily a total credit enhancement for a
securitization deal. For example, a deal could have 20 percent
subordination (say, a $1 billion mortgage pool divided into an
$800 million senior bond and a $200 million junior bond)
without the BHC holding (retaining) any of the subordinated
piece. In that case, the enhancement would not show up in our
data. Our basic question, however, remains: Is risk retention
important because it is a buffer against observable risk or
because it is a signal of unobservable quality? Indeed, Title 9
of the Dodd-Frank Act requires federal regulators to set
8
Note that banks also provide enhancements in the form of representation
and warranties that obligate the issuer to take back the loan if it defaults early
in its life.
38 The Role of Bank Credit Enhancements in Securitization
Source: Federal Reserve System, Form FR Y-9C, Schedule HC-S.
Chart 1
Total Credit Enhancements
by Bank Holding Companies
Billions of U.S. dollars Percent
Enhancements as share
of outstanding securitizations
Scale
Total enhancements
Scale
0
20
40
60
80
100
120
140
160
180
09080706050403022001
0
1
2
3
4
5
6
7
8
Source: Federal Reserve System, Form FR Y-9C, Schedule HC-S.
Chart 2
Credit Card Enhancements
by Bank Holding Companies
Billions of U.S. dollars Percent
Enhancements as
share of outstanding
credit card securitizations
Scale
Total enhancements
Scale
0
20
40
60
80
100
120
140
160
09080706050403022001
0
1
2
3
4
5
6
7
8
mandatory retention standards for sponsors of asset-backed
securities, suggesting that some policymakers believe that
enhancements in the form of retentions can ameliorate the
incentive and information problems endemic to securitization.
Because the enhancement data in Schedule HC-S have not,
to our knowledge, been studied publicly before, we briefly
examine the data in graphic form to get a sense of the size,
trends, and volatility of enhancements by BHCs. The data run
from 2001:Q2, when BHCs were first required to disclose
securitization activity, to 2009:Q4, when BHCs were required,
per Financial Accounting Standards Board ruling 167,
9
to bring
securitized assets back on their balance sheets (and thus ceased
to report most enhancements).
Chart 1 plots total enhancements in billions of dollars and
as a percentage of outstanding securitizations. Measured per
securitized asset, enhancements were more or less stable at
between 2 and 3 percent until 2009:Q1, although there is a
slight upward trend in the series to that point. In dollar terms,
total enhancements trended upward from about $25 billion
in 2001:Q2 to about $70 billion in 2009:Q1. In the following
quarter, total enhancements more than doubled, to
$164 billion, and enhancements per securitized asset rose
to about 6 percent.
Chart 2 shows that the abrupt increase in total enhance-
ments in 2009 came about almost entirely because of a rise in
enhancements on securitized credit card loans. The increase
in credit card enhancements, in turn, came about because of
increased enhancements at two BHCs: Bank of America and
9
See http://www.fasb.org/cs/ContentServer?c=FASBContent_C&pagename
=FASB/FASBContent_C/NewsPage&cid=1176156240834.
JPMorgan Chase (JPMC). The increase at Bank of America
followed purchases of new securitization trusts after it acquired
Merrill Lynch in 2009. More interestingly, perhaps, the
increase in enhancements at JPMC in 2009 occurred primarily
because several classes of notes issued by Chase Issuance Trust,
one of its master trusts of securitized credit card assets, were
placed on credit watch and one class of notes was down-
graded.
10
That case illustrates how enhancements are used
to maintain a given rating level, whether by providing
a buffer against collateral losses, a signal of faith in the quality
of the assets, or both.
For completeness, Chart 3 plots the enhancements, both
by level and per securitized asset, for non–credit card
enhancements. The only feature of note is the downward trend
in non–credit card enhancements per securitized non–credit
card asset. That finding implies that the upward trend in overall
enhancements per securitized asset evident in Chart 1 results
from the upward trend in credit card enhancements per
securitized asset evident in Chart 2.
Chart 4 breaks out total enhancements into enhancements
of the BHCs’ own securitized assets (“self-enhancements”) and
enhancements provided to third parties (“third-party
enhancements”). Apart from the beginning and the end of the
sample period, self-enhancements were roughly stable at
between $30 billion and $40 billion. By contrast, third-party
enhancements began trending upward in about 2004:Q4 to
reach a peak of about $25 billion in 2008:Q2. Third-party
10
See “Fitch: Chase Increases Credit Enhancement in Credit Card Issuance
Trust (CHAIT),” http://www.reuters.com/article/2009/05/12/
idUS260368+12-May-2009+BW20090512.
FRBNY Economic Policy Review / July 2012 39
Source: Federal Reserve System, Form FR Y-9C, Schedule HC-S.
Chart 3
Non–Credit Card Enhancements
Billions of U.S. dollars Percent
Enhancements as share
of outstanding non–credit
card enhancements
Scale
Total enhancements
Scale
0
5
10
15
20
25
30
35
40
45
50
0908070605040302
2001
0
5
10
15
20
25
Source: Federal Reserve System, Form FR Y-9C, Schedule HC-S.
Chart 4
Self-Enhancements and Third-Party Enhancements
Billions of U.S. dollars
Third-party
enhancements
Self-enhancements
0
20
40
60
80
100
120
140
160
180
09080706050403022001
enhancements dropped noticeably during the financial crisis,
presumably because BHCs’ own solvency and liquidity came
into question.
While Charts 1-4 tell us something about trends in
enhancements within the banking industry, we were also
interested in how enhancements by bank holding companies
compared with those by financial institutions in the shadow
banking system, namely, insurance companies. Insurance
companies provide enhancements to structured finance
products through guaranties and credit default swaps (CDS).
As there is no central source of data on enhancements provided
by insurance companies, we turned to their 10-K forms for
data. Starting with the nineteen publicly traded insurance
companies, we determined that only six or seven (depending
on the year) provided guaranties for asset-backed securities.
These included firms such as Ambac, MBIA, and Radian.
11
While the companies usually provided a reasonable breakdown
of guarantee coverage—such as residential and consumer loans
and the like—the classifications were not uniform across
companies. Thus, for each company we summed guaranties
across categories and then summed across companies to obtain
the aggregate level of guaranties by publicly traded insurance
companies in a given year.
11
The sample excludes American International Group, Inc. AIG was a
prominent seller of CDS protections on collateralized debt obligations (CDOs)
through one of its subsidiaries, AIG Financial Products. AIG experienced a
severe liquidity crisis due to its rating downgrade in late 2008, and the
subsequent bailout resulted in a substantial decline in outstanding net notional
amount of AIG’s CDS portfolio written on CDO products. Including AIG in
our analysis would therefore cause a more significant downward trend in
insurance companies’ presence in the financial guarantee market for the
sample period. We exclude AIG to make a conservative comparison of the
aggregate volumes of protection provided by banks and insurance firms in the
securitization market.
Chart 5 plots the ratio of guaranties by insurance companies
to total enhancements provided by bank holding companies.
The level of guaranties provided by insurance companies
clearly swamps the level of enhancements provided by bank
holding companies; at its peak in 2004:Q4, the ratio was more
than ten to one. However, apart from some notable
fluctuations, including a drop in 2009 because of the increase
in credit card enhancements at JPMC and Bank of America, the
ratio has been fairly trendless, indicating that banks have
maintained (or perhaps increased) their share of the credit
enhancement business.
As noted in the introduction, we found that insurers would
often cite (in their 10-Ks) competition from banks for
enhancement business. Here is another example:
Financial guarantee insurance also competes with other
forms of credit enhancement, including senior-subordinated
structures, credit derivatives, letters of credit and guarantees
(for example, mortgage guarantees where pools of mortgages
secure debt service payments) provided by banks and other
financial institutions, some of which are governmental
agencies. Letters of credit are most often issued for periods
of less than 10 years, although there is no legal restriction
on the issuance of letters of credit having longer terms. Thus,
financial institutions and banks issuing letters of credit
compete directly with our Insurers to guarantee short-term
notes and bonds with a maturity of less than 10 years. To the
extent that banks providing credit enhancement may begin
to issue letters of credit with commitments longer than
10 years, the competitive position of financial guarantee
insurers could be adversely affected (MBIA Inc. 2008,
form 10-K, p. 24).
40 The Role of Bank Credit Enhancements in Securitization
Sources: Federal Reserve System, Form FR Y-9C, Schedule HC-S;
insurance companies’ 10-K forms.
Chart 5
Guaranties to Asset-Backed Securities Provided
by Insurance Companies/Credit Enhancements
Provided by Bank Holding Companies
Ratio
2
4
6
8
10
12
09080706050403022001
3. Panel Regression Results
In this section, we investigate the relationship between the
performance of securitized assets and the extent of credit
enhancements. According to the buffer hypothesis, where
enhancements are a buffer against observable risks, one would
expect a negative relationship between enhancements and
performance. Under the signaling hypothesis, where
enhancements are a signal of unobserved quality, we would
expect a positive relationship between enhancements and
performance.
To investigate that question, we estimate the following
fixed-effect regression models:
(1) Severe Delinquency Rate
it
Total Enhancements
it
Controls .
For each loan category (mortgages, credit card loans, and the
like), the dependent variable is the sum of securitized assets
ninety or more days past due and loans charged off, divided by
total securitized assets outstanding at BHC i in quarter t. The
main independent variable, TotalEnhancements, is the sum of
the three types of credit enhancements discussed earlier scaled
by total outstanding securitizations for each BHC in each
quarter.
12
The controls are unused commitments divided by
total loans in each category, the log of on balance sheet assets,
leverage (total common equity divided by total balance sheet
i
t
++=
+
it
+
assets), ROA (quarterly net income divided by total balance
sheet assets), and risk-weighted assets divided by total balance
sheet assets (a measure of risk). All the variables in this and
subsequent regressions are defined in the appendix. The BHC
and time-fixed (quarter-year) effects control for constant
differences in performance across BHCs and time. We report
Huber-White robust standard errors for all quarter-BHC
observations with nonmissing, nonzero outstanding
securitization. The standard errors are clustered by BHCs. The
equation is estimated from 2001:Q2 to 2007:Q2 , that is, up to
but not including the financial crisis. A BHC is included in the
regression if it had nonzero securitization for a given loan type.
12
Besides the aggregate enhancement, Schedule HC-S reports disaggregated
numbers cross several categories, including retained interest-only strips,
standby letters of credit, subordinated securities, and other enhancements,
as discussed earlier. We focus on the aggregate amount, as discussions with
professionals in this business sector suggest that the overall amount of
enhancements is the most relevant term in the deal-making process.
Table 1
Summary Statistics
Variable Observations Mean
Standard
Deviation
Severe delinquency ratio
a
Residential real estate 3,394 0.006 0.025
Home equity 536 0.012 0.024
Credit card 703 0.012 0.018
Auto 686 0.005 0.011
Other consumer 444 0.027 0.032
Commercial and industrial 717 0.003 0.008
All other 968 0.002 0.008
Total 4,589 0.005 0.017
Total enhancements (ratio)
b
Residential real estate 3,394 0.037 0.150
Home equity 536 0.062 0.108
Credit card 703 0.024 0.071
Auto 686 0.060 0.104
Other consumer 444 0.063 0.095
Commercial and industrial 717 0.037 0.124
All other 968 0.062 0.170
Total 4,589 0.041 0.150
Source: Federal Reserve System, Form FR Y-9C, Schedule HC-S.
a
Severe delinquency ratio = securitized loans ninety days past due plus
charge-offs divided by total loans in that category.
b
Total enhancements = sum of credit-enhancing, interest-only strips
and excess spread accounts, subordinated securities, and other residual
interest; standby letters of credit; and other enhancements divided by
total loans in that category.
FRBNY Economic Policy Review / July 2012 41
Summary statistics are reported in Table 1, and the
regression results are in Table 2. In the regressions, the point
estimates on total enhancements are positive in every loan
category but the residual “all other” and are significantly
different from zero in four of the eight categories: residential
real estate, home equity, auto, and total. Thus, we find no
evidence for the signaling hypothesis and some evidence for
the hypothesis that enhancements serve as a buffer against
observable risk. It is possible that enhancements serve as both
a buffer and a signal but the buffering role dominates.
Although we do not claim that the relationship between
delinquency and enhancements is causal, it is still interesting to
gauge the magnitude of the relationship between the two. To
do so, we calculate how much delinquency rates rise relative to
the average when total enhancements increase by one standard
deviation. Specifically, we calculate the product of the point
estimate for each loan category and the standard deviation
of total enhancements for that category; we then scale that
product by the mean delinquency rate for that category. The
result yields the estimated percentage change in delinquency
(relative to the mean delinquency rate) per standard deviation
change in total enhancements. The results imply a fairly stable
relationship between total enhancements and delinquency
rates in cases where the relationship was statistically significant:
residential real estate (0.43), home equity (0.81), auto (0.56),
and total (0.45).
Table 2
Panel Regression Results
Dependent Variable: Severely Delinquent Loans / Total Securitized Loans
Pre-Crisis (2001:Q2 to 2007:Q2)
Residential
Real Estate
Home
Equity Credit Card Auto
Other
Consumer
Commercial
and Industrial All Other Total
Total enhancements 0.017 0.09 0.044 0.027 0.037 0.003 -0.007 0.015
[2.54]** [4.49]*** [1.07] [2.38]** [1.34] [0.29] [0.86] [2.40]**
Unused commitments -0.083 -0.015 3.714 -0.009 -0.034 -0.004 -0.002 -0.001
[1.85]* [1.22] [1.92]* [1.05] [1.44] 1.05] [0.85] [0.17]
Leverage -0.047 0.015 -0.125 0.001 0.218 0.026 -0.094 -0.04
[1.61] [0.07] [3.35]*** [0.11] [1.18] [0.55] [3.64]*** [1.08]
Return on assets 0.226 -0.11 -0.52 0.011 0.009 0.029 -0.131 -0.042
[1.09] [0.20] [8.58]*** [1.26] [0.03] [0.17] [1.39] [0.49]
Risk-weighted assets/total assets -0.017 0.006 0.01 0.028 -0.003 -0.013 0.002 0.008
[1.75]* [0.20] [0.39] [3.63]*** [0.05] [0.83] [0.24] [0.91]
Log asset size 0.002 0.013 0.009 -0.002 0.022 0.004 -0.002 0.004
[0.92] [1.36] [1.36] [0.88] [1.15] [1.49] [0.71] [1.36]
Observations 3,358 532 703 685 444 706 960 4,543
Number of entities 166 27 34 32 22 35 48 225
R
2
0.04 0.18 0.36 0.2 0.17 0.05 0.06 0.06
Source: Authors’ calculations.
Notes: Robust t-statistics appear in brackets. Time dummies are not reported. Variables are defined in the appendix.
***Statistically significant at the 1 percent level.
***Statistically significant at the 5 percent level.
***Statistically significant at the 10 percent level.
42 The Role of Bank Credit Enhancements in Securitization
4. Event Studies: What Do Stock
Price Reactions and Implied
Volatility Tell Us about
the Role of Enhancements?
We next investigate the role of credit enhancements in
securitization by looking at market reactions to the new
disclosure requirement adopted in 2001:Q2 on BHCs’
securitizations. Beginning in that quarter, BHCs started
including in the quarterly “Reports of Condition and Income”
a new schedule that detailed their securitization activities.
The new schedule requires BHCs to disclose comprehensive
information on the volume and performance
13
of seven
categories of securitized assets (the same categories we study
in the panel analysis above). Significantly, BHCs are required
to report the maximum amount of credit exposure they face
through the credit enhancements described above. This new
information first became public after BHCs’ reports for
2001:Q2 were disclosed in August and September 2001. This
event provides a unique opportunity for assessing how banks’
securitization and the associated credit exposure through
enhancements affect shareholders.
We focus on the valuation and risk implications of the newly
disclosed securitization activities. First, we conduct a standard
event study on a sample of 267 BHCs. A one-factor market
model is estimated for each firm using monthly return data
from July 1996 to June 2001, with the S&P 500 index being the
factor. Monthly abnormal returns are calculated for August
and September 2001 and then summed to reach a two-month
cumulative abnormal return (CAR) for each bank.
To see how the newly disclosed securitization activities and
credit enhancements affect valuation, we relate the CARs to
several securitization-related variables through the following
regression:
(2) CAR
i
Securitization
i
Total_Enhancements
i
Total_Enhancements
i
Delinquency
i
Delinquency
i
Unused Commitments
i
Stock Volatility
i
The dependent variable CAR
i
is the two-month cumulative
abnormal return for bank i. All independent variables are
constructed using data from the Federal Reserve Y-9C reports,
which bank holding companies filed as of 2001:Q2 under
the revised reporting rules. Securitization
i
represents the
outstanding principal balance of assets sold and securitized by
bank i, with servicing retained or with recourse or other seller-
provided credit enhancements, and is normalized by the bank’s
total outstanding loans on the balance sheet. This measure
13
The performance metrics include past-due amounts, charge-offs, and
recoveries on assets sold and securitized.
1
+=
2
3
4
5
6
i
reflects the extent to which bank i has moved its loans off the
balance sheet through securitization. Total_Enhancements
i
and Unused Commitments
i
are defined in Section 3 (and the
appendix). While the scale of securitization activities is
captured by Securitization
i
, Total_Enhancements
i
reflects the
extent to which bank i could still be “on the hook” should the
securitized assets perform poorly. We measure performance by
Delinquency
i
, defined as the sum of past-due loan amounts and
year-to-date net charge-offs divided by the total outstanding
securitized assets. Last, to control for a BHC’s risk, we include
the stock volatility estimated using the daily returns in the
252 trading days prior to the disclosure period.
Equation 2 also includes an interaction between
Total_Enhancements
i
and Delinquency
i
. Per our earlier
discussion, we postulated two hypotheses on the role of
enhancements. Under the signaling hypothesis, keeping risk
through enhancements signals bank i’s private knowledge of
good loan quality, implying a positive relationship between
high enhancements and CAR. Under the buffer hypothesis,
banks securitizing riskier collateral need more enhancements
to meet rating agencies’ criteria. In this case, high enhancements
are associated with observably riskier deals, implying a negative
valuation impact. If loan performance is a reasonable proxy
for the observable riskiness of the securitized assets, we expect
the signaling effect to dominate among relatively better-
performing (lower-delinquency) deals, where observable risk
is less a concern, resulting in an overall positive relationship
between Total Enhancements and CAR. When deals are
performing poorly (high delinquency), however, concerns
over “observable risk” would heighten and the buffer role
of enhancements would dominate, leading to a negative
relationship between Total Enhancements and CAR. As a result,
we expect a positive coefficient for Total Enhancements ()
and a negative coefficient for the interaction Total
Enhancements Delinquency ().
Table 3 presents the least-squares regression coefficient
estimates with Huber-White robust standard errors. Each
model estimated includes one of two versions of the
Delinquency
i
variable. Models 1 and 2 use a delinquency
measure based on all past-due loans, while models 3 and 4
use one that includes severe delinquencies only. The cross-
section variation of the CARs appears to be significantly
associated with the securitization-related variables. The
impact of Securitization is significantly positive in all
specifications, suggesting that more favorable market
reactions are associated with larger-scale securitizations as
first disclosed by banks in 2001. This finding is consistent
with the notion that securitization transactions were
generally viewed as positive-NPV (that is, profitable)
projects in 2001 and that the market reacted more favorably
2
3
FRBNY Economic Policy Review / July 2012 43
when banks reported that a higher portion of their assets
was being securitized.
In columns 1 and 3 of Table 3, Total_Enhancements,
Delinquency, Unused Commitments, and Stock Volatility are all
statistically insignificant. Securitization is the only significant
variable in those models.
Models 2 and 4 suggest that the insignificance of Total
Enhancements in models 1 and 3 is likely due to the omitted
interaction between enhancements and loan performance. In
both models 2 and 4, Total Enhancements alone is significantly
positive and has a strongly negative interaction effect with loan
performance, Total Enhancements Delinquency. A simple
numerical exercise further illustrates the importance of the
interplay between enhancement and loan performance in
determining which of the two hypotheses dominates. Using
model 4 as an example, we can compare the relationship
between the market reaction (CAR) and enhancements at
different levels of severe loan delinquencies off the balance
sheet. For example, with no severe delinquency (Delinquency =
0 percent), the overall effect of Total Enhancements is
0.168 + (-15.567) 0 = 0.168, a positive wealth effect of
enhancements consistent with the signaling hypothesis.
As the severe delinquency ratio rises, however, the effect of
Total Enhancements weakens monotonically but remains
positive until severe delinquency reaches 0.168/15.567 =
1.08 percent.
14
Once the delinquency rate exceeds
1.08 percent, the net effect of Total Enhancements on CAR
becomes increasingly negative as delinquency further rises.
For example, when severe delinquency is 1.18 percent,
15
the net effect of Total Enhancements on CAR becomes
0.168 + (-15.567) 1.18 percent = -1.6 percent. This negative
relationship is consistent with the notion that investors
become increasingly concerned when a bank with poorly
performing securitized assets discloses a high level of credit
enhancements, just as the buffer hypothesis would predict.
We next focus on the risk implications of banks’
securitization activities. Specifically, we examine changes in
option-implied volatilities around the event period. For
fifty-one banks in our sample, we obtained data from the
OptionMetrics Ivy database, which features implied volatilities
calculated using the Cox, Ross, and Rubinstein (1979)
binomial model adjusted for dividends. Because some banks
have numerous exchange-traded options, we impose a number
of widely used sample restrictions.
16
We calculate weighted-
average implied volatilities at the firm level, using each option’s
vega as the weight (Latané and Rendleman 1976). We then run
the following regression:
(3) [log (implied_vol
i
)]
Securitization
i
+ Total_Enhancements
i
+ Total_Enhancements
i
Delinquency
i
Delinquency
i
Unused Commitments
i
Stock Volatility
i
14
This number corresponds to the 90th percentile of the severe delinquency
ratio in our sample.
15
This number corresponds to the 92nd percentile of the severe delinquency
ratio in our sample.
16
Specifically, several studies (see, for example, Patell and Wolfson [1981])
report that implied volatility estimates behave erratically during the last two
to four weeks before expiration and also that options with a very long time to
expiration are less sensitive to volatility changes). We therefore study only
those options with expiration dates between 28 and 100 days away from the
event day, with the latter criterion due to Deng and Julio (2005). Last, we
require each option to have nonzero trading volume in the event window.
1
+=
2
3
4
5
6
i
Table 3
Regression Analysis of Cumulative Abnormal
Equity Returns
Dependent Variable: Cumulative Abnormal Equity Returns,
August 2001-September 2001
(1) (2) (3) (4)
Constant -0.002 -0.0034 -0.0016 -0.0027
[0.34] [0.57] [0.28] [0.0059]
Securitization 0.04 0.032 0.042 0.036
[3.48]*** [2.62]*** [3.88]*** [2.99]***
Total enhancements 0.047 0.295 0.06 0.168
[0.58] [4.40]*** [0.82] [3.45]***
Delinquencies (all) -0.106 0.499
[0.46] [1.59]
Delinquencies (all)
total enhancements
-10.933
[3.06]***
Delinquencies
(severe)
-0.337
[0.78]
0.759
[1.50]
Delinquencies
(severe) total
enhancements
-15.567
[3.32]***
Unused commitments 0.014 -0.122 0.005 -0.041
[0.20] [2.31]** [0.08] [0.83]
Stock volatility -1.498 -1.33 -1.496 -1.374
[1.53] [1.36] [1.53] [1.41]
Observations 267 267 267 267
R
2
(percent) 5757
Source: Authors’ calculations.
Notes: Robust t-statistics appear in brackets. Variables are defined
in the appendix.
***Statistically significant at the 1 percent level.
***Statistically significant at the 5 percent level.
***Statistically significant at the 10 percent level.
44 The Role of Bank Credit Enhancements in Securitization
The dependent variable [log(implied_vol
i
)] measures the
change in log(implied_vol
i
) from the beginning of August 2001
to the end of September 2001. All the independent variables
remain the same as in equation 2.
17
Overall, the significantly positive coefficient estimates for
Securitization suggest that higher securitization activities are
associated with higher risk as perceived in the forward-looking
option market (Table 4). This result, coupled with the positive
valuation effect of securitization just noted, suggests that
securitization was generally viewed as increasing both
shareholder value and risk. Unused commitments were also
17
We cannot control for market movement in the current regression setup.
As an alternative, we define excess implied volatility as the difference between
each option’s implied volatility and market volatility and use it to calculate the
dependent variable. The results are quantitatively similar to those in Table 4.
associated with higher risk, despite the lack of valuation
effect (see Table 2). Total Enhancements are always positive
and significant, which is sensible given that enhancements
represent exposure to the securitizing bank. Unlike the analysis
of valuation impact, we do not observe any significant
interaction effect between Total Enhancements and
Delinquency in the risk effect of credit enhancements. Overall,
the evidence suggests that both securitization activities and the
associated credit enhancements are perceived to add risk to the
securitizing bank, even though underlying assets have been
moved off the balance sheet.
5.Conclusion
This article focuses on credit enhancements provided by banks
in the U.S. securitization market. Contrary to the impression
that banks have been surpassed by other financial institutions
in the shadow banking system, we show that banks have held
their own relative to monoline insurance companies in the
business of providing credit enhancements.
Having shown that banks are still important in providing
enhancements, we also investigate the role of bank enhance-
ments in the securitization process. Enhancements obviously
serve as a buffer against observable risk, but we are interested
in the hypothesis, commonly advanced by academics, that
enhancements also serve as a signal of unobservable quality.
By keeping “skin in the game,” banks offering enhancements
may signal to investors or raters that the assets being securitized
are of high quality.
Our event study of banks’ first-time disclosure in 2001 of
their securitization activities finds evidence that the buffer
effect and the signal hypothesis could both be at play, with the
dominant effect depending on the riskiness of the securitized
assets. Specifically, we find that stock prices reacted favorably
to high enhancement provisioning among banks with better-
performing (lower-delinquency) securitizations, consistent
with the signaling hypothesis. Among banks with poorly
performing securitizations (high delinquency), however, stock
prices reacted negatively to higher levels of enhancements,
suggesting that the buffer role of enhancements dominates
under observably risky securitizations.
Evidence from cross-sectional regressions favors the buffer
hypothesis of enhancements. There we find a positive
relationship between delinquency rates on banks’ securitized
assets and credit enhancements, contrary to what the signaling
hypothesis suggests. Of course, it could be that enhancements
do serve a signaling role, but that role is dwarfed by the
buffering role.
Table 4
Regression Analysis of Changes in Implied Volatility
Dependent Variable: [log( Implied Volatility )]
(1) (2) (3) (4)
Constant 0.414 0.404 0.418 0.41
[5.12]*** [4.95]*** [5.37]*** [5.25]***
Securitization 0.045 0.043 0.049 0.047
[2.21]** [1.80]* [2.39]** [2.12]**
Total enhancements 0.279 0.367 0.289 0.316
[2.64]** [2.06]** [2.62]** [2.64]**
Delinquencies (all) -0.137 0.15
[0.27] [0.15]
Delinquencies (all)
total enhancements
-4.744
[0.44]
Delinquencies (severe) -0.538 -0.125
[0.71] [0.08]
Delinquencies (severe)
total enhancements
-5.53
[0.38]
Unused commitments 0.187 0.145 0.177 0.172
[2.32]** [1.53] [2.10]** [1.92]*
One-year lagging
daily stock return
standard deviation
-10.945
[3.46]***
-10.626
[3.39]***
-10.952
[3.55]***
-10.726
[3.48]***
Observations 52 52 52 52
R
2
(percent) 29303030
Source: Authors’ calculations.
Notes: Robust t-statistics appear in brackets. Variables are defined
in the appendix.
***Statistically significant at the 1 percent level.
***Statistically significant at the 5 percent level.
***Statistically significant at the 10 percent level.
FRBNY Economic Policy Review / July 2012 45
Delinquencies (All): Securitized loans thirty or more days past
due plus charge-offs divided by total securitized loans in the
category.
Delinquencies (All) (Total Enhancements): Delinquencies
(all) times total credit enhancements.
Delinquencies (Severe): Securitized loans ninety days past due
plus charge-offs divided by total securitized loans in the
category.
Delinquencies (Severe) (Total Enhancements): Delinquencies
(severe) times total credit enhancements.
Leverage: Total common equity divided by total balance sheet
assets.
Log Asset Size: Natural log of total balance sheet assets.
Risk-Weighted Assets/Total Assets: Total risk-weighted assets
divided by total balance sheet assets.
ROA: Quarterly net income divided by total balance sheet
assets.
Securitization: Total securitized loans divided by total balance
sheet loans.
Severely Delinquent Loans/Total Securitized Loans: Securitized
loans ninety days past due plus charge-offs divided by total
securitized loans in the category.
Stock Volatility: One-year lagging daily stock return standard
deviation.
Total (Credit) Enhancements: Sum of interest-only strips,
subordinated securities, and other residual interest; standby
letters of credit; and other enhancements divided by total loans
in the category.
Unused Commitments: Unused commitments to provide
liquidity divided by total loans in the category.
Appendix: Variable Definitions
References
46 The Role of Bank Credit Enhancements in Securitization
Albertazzi, U., G. Eramo, L. Gambacorta, and C. Salleo. 2011.
“Securitization Is Not that Evil after All.” BIS Working Paper
no. 341, March.
Ashcraft, A. B., and T. Schuermann. 2008. “Understanding the
Securitization of Subprime Mortgage Credit.” Federal Reserve
Bank of New York Staff Reports, no. 318, March.
Ashcraft, A. B., J. Vickery, and P. Goldsmith-Pinkham. 2010. “MBS
Ratings and the Mortgage Credit Boom.” Federal Reserve Bank
of New York Staff Reports, no. 449, May.
Berger, A. N., and G. F. Udell. 1990. “Collateral, Loan Quality, and
Bank Risk.” Journal of Monetary Economics 25, no. 1
(January): 21-42.
Bernanke, B. S. 2012. “Some Reflections on the Crisis and the Policy
Response.” Remarks delivered at the Russell Sage Foundation
and Century Foundation Conference on “Rethinking Finance,”
New York City, April 13.
Besanko, D., and A. Thakor. 1987. “Collateral and Rationing: Sorting
Equilibria in Monopolistic and Competitive Credit Markets.”
International Economic Review 28, no. 3 (October): 671-89.
Cetorelli, N., and S. Peristiani. 2012. “The Role of Banks in Asset
Securitization.” Federal Reserve Bank of New York Economic
Policy Review 18, no. 2 (July): 47-63.
Chan, Y., and G. Kanatas. 1985. “Asymmetric Valuations and the Role
of Collateral in Loan Agreements.” Journal of Money, Credit,
and Banking 17, no. 1 (February): 84-95.
Cox, J. C., S. A. Ross, and M. Rubinstein. 1979. “Option Pricing:
A Simplified Approach.” Journal of Financial Economics 7,
no. 3 (September): 229-63.
Deng, Q., and B. Julio. 2005. “The Informational Content of Implied
Volatility around Stock Splits.” University of Illinois at
Urbana-Champaign working paper, September.
Downing, C., D. Jaffee, and N. Wallace. 2009. “Is the Market for
Mortgage-Backed Securities a Market for Lemons?” Review
of Financial Studies 22, no. 7 (July): 2457-94.
Erel, I., T. D. Nadauld, and R. M. Stulz. 2011. “Why Did U.S. Banks
Invest in Highly Rated Securitization Tranches?” Fisher College
of Business Working Paper no. 2011-03-016, July 25.
James, C. M. 2010. “Mortgage-Backed Securities: How Important Is
‘Skin in the Game’?” Federal Reserve Bank of San Francisco
Economic Letter, no. 2010-37, December 13.
Latané, H. A., and R. J. Rendleman, Jr. 1976. “Standard Deviations
of Stock Price Ratios Implied in Option Prices.” Journal
of Finance 31, no. 2 (May): 369-81.
Levitin, A. J. 2011. “Skin in the Game: Risk Retention Lessons from
Credit Card Securitization.” Georgetown Law and Economics
Research Paper no. 11-18, August 9.
Patell, J. M., and M. A. Wolfson. 1981. “The Ex Ante and Ex Post Price
Effects of Quarterly Earnings Announcements Reflected in Option
and Stock Prices.” Journal of Accounting Research 19, no. 2
(autumn): 434-58.
Pozsar, Z., T. Adrian, A. Ashcraft, and H. Boesky. 2010. “Shadow
Banking.” Federal Reserve Bank of New York Staff Reports,
no. 458, July.
Standard and Poor’s. 2008. “The Basics of Credit Enhancement
in Securitizations.” June 24.
The views expressed are those of the authors and do not necessarily reflect the position of the Federal Reserve Bank of New York
or the Federal Reserve System. The Federal Reserve Bank of New York provides no warranty, express or implied, as to the
accuracy, timeliness, completeness, merchantability, or fitness for any particular purpose of any information contained in
documents produced and provided by the Federal Reserve Bank of New York in any form or manner whatsoever.
