Accounting Horizons
Vol. 25, No. 3
2011
pp. 559–576
American Accounting Association
DOI: 10.2308/acch-50009
Structured Finance and Mark-to-Model
Accounting: A Few Simple Illustrations
Anthony Meder, Steven T. Schwartz, Eric E. Spires, and Richard A. Young
SYNOPSIS: We review the development of structured financial products, discuss their
accounting treatment, and illustrate their valuation using simple numerical examples.
The crucial element we incorporate is the possibility that the underlying assets in
structured financial products have correlated returns. The benefit of structured finance is
it uses diversification to protect the senior tranches’ cash flows. However, when the
underlying assets have correlated returns diversification is not as effective. Normally,
structured financial products would be marked ‘‘to market,’’ obviating the need for
analytical valuation techniques. Current accounting standards, however, have significant
provisions for valuing structured financial products based on analytically derived
expectations of future cash flows, especially when markets are illiquid. Therefore, it is
important for both preparers and users of accounting information to understand how
underlying economic fundamentals, such as the correlation in returns, affect
expectations of future cash flows.
Keywords: fair value accounting; structured finance; systematic risk.
INTRODUCTION
D
uring the recent financial crisis, banks, hedge funds, credit-rating agencies, and regulatory
agencies have all come under greater scrutiny. We may add to this list accountants and
accounting standards. The two most prominent questions that have arisen with respect to
accounting are whether mark-to-market accounting helps spread financial contagion (Pozen 2009;
SEC 2008; Westbury 2008) and whether the more opaque mark-to-model method is reliable (Kolev
2008; Song et al. 2010). Marking ‘‘to market’’ refers to valuing financial assets at the price found for
identical assets traded in active markets, whereas marking ‘‘to model’’ refers to valuing financial
assets based on analytically derived expectations of future cash flows. Both approaches come under
the broader rubric of fair value accounting and stand in stark contrast to the more traditional
Anthony Meder is a Graduate Research Associate at The Ohio State University, Steven T. Schwartz is an
Associate Professor at Binghamton University, and Eric E. Spires is an Associate Professor and Richard A.
Young is a Professor, both at The Ohio State University.
The authors thank Anil Arya, John Fellingham, Brian Mittendorf, Robert Russ, Doug Schroeder, Shyam Sunder and
participants at the 2010 Accounting Hall of Fame Conference.
Submitted: June 2010
Accepted: January 2011
Published Online: September 2011
Corresponding author: Steven T. Schwartz
Email: sschwart@binghamton.edu
559
560
Meder, Schwartz, Spires, and Young
historical cost approach. Standard-setters have gradually been moving away from historical cost,
especially for financial assets (SEC 2008).
There have been several theories that attempt to explain how mark-to-market accounting can
spread financial crises (Allen and Carletti 2008; Cifuentes et al. 2005; Heaton et al. 2009; Plantin et
al. 2008). Although the models vary in their assumptions, most feature as the transmission
mechanism some sort of regulatory capital requirement, such as minimum capital ratios. Laux and
Leuz (2010, 93), in generalizing the models, note, ‘‘The write-downs . . . deplete bank capital and
set off a downward spiral, as banks are forced to sell assets at ‘fire sale’ prices, which in turn can
lead to contagion as prices from asset fire sales of one bank become relevant for other banks.’’
Authors such as Ball (2008) and Ryan (2008b) remain skeptical, and claim that mark-to-market
speeds the necessary process of adjustment and bolsters investor confidence. The issue continues to
be debated in scholarly journals and the popular press.1
Despite the unsettled nature of the issue, intense political pressure from Congress, which itself
has been receiving pressure from the financial services industry, caused the Financial Accounting
Standards Board (FASB) to issue Staff Position FSP 157-3, which expands the potential use of
mark-to-model accounting for assets whose markets have become illiquid (Karabell 2008). Despite
this move, many in the financial community did not believe the FASB went far enough (Pozen
2009). Under continued political pressure, the FASB further expanded the circumstances under
which mark-to-model accounting may be used, issuing FSP 157-4 in April of 2009. Pozen (2009,
89) notes, ‘‘Given the FASB’s two recent pronouncements . . . there is no question that banks will
increasingly value illiquid securities by marking them to model.’’ The FASB’s move to expand
mark-to-model accounting has its critics, many of whom have labeled the new accounting standards
as ‘‘mark-to-myth.’’
The goal of this paper is not to pass judgment on the merits of mark-to-model accounting.
Instead, our goal is to use simple numerical examples to illustrate how models can be employed to
value certain types of financial assets. We show that for the financial assets in question, valuation is
critically dependent on how one assesses the potential for systematic risk, that is, risk that cannot be
mitigated through diversification.2
The assets we focus on are structured financial products and related credit derivatives. We use
as an illustrative example a collateralized mortgage obligation (CMO), although the principles we
present apply to many structured financial products. Our choice is not arbitrary. The use of
structured financial products has exploded in recent years.3 For example, a recent International
Monetary Fund (IMF) publication estimates that global structured financial products grew from
around $500 billion in 2000 to a peak of around $3,000 billion in 2006 (IMF 2008). The largest
share of these issuances was mortgage-backed securities, of which a CMO is one type.
1
2
3
Laux and Leuz (2009) provide a review of the current debate. They argue that even if it is true that mark-tomarket accounting helps spread financial crises, changes should be made not to accounting standards, but to
prudential regulations, such as capital ratio requirements, that interact with the accounting standards.
Confusion may ensue between the terms systematic risk and systemic risk. Systematic risk, which is well defined
in the finance literature, refers to that portion of risk that cannot be eliminated through diversification. Systemic
risk is less well defined, but generally refers to the risk of an event with system-wide consequences. Therefore,
one might say that the collapse of a major financial institution (such as Lehman Brothers) is a systemic risk,
whereas correlation in mortgage defaults is a systematic risk. Even here we have simplified somewhat, because
the risk that may not be diversified by holding different mortgages may be diversified by holding non-mortgage
assets, such as treasury bonds or gold.
Bowden and Lorimer (2009) note the sharp growth in structured financial products and their derivatives by
providing a partial list of newly introduced products that includes credit spread options, swaptions, variance
swaps, credit-linked notes, collateralized debt obligation squareds, collateralized loan obligations, synthetic
CDOs, basket products, tranched index trades, equity default swaps, credit contingent swaps, constant maturity
swaps and constant proportion portfolio insurance.
Accounting Horizons
September 2011
Structured Finance and Mark-to-Model Accounting: A Few Simple Illustrations
561
The rest of this paper is organized as follows. The second section provides background
information on structured finance and relevant accounting standards. The third section introduces
the model and extensions. The fourth section concludes the paper.
BACKGROUND
In this section we discuss some of the important concepts related to structured finance and fair
value accounting. Terms in this arena often are used inconsistently and loosely defined; for
example, securitization and structured finance are sometimes used interchangeably. We will try to
be clear in our use of these terms.
Securitization
Securitization refers to the general process of forming a pool of assets, usually some type of
loan or other debt-like instrument that is not sold in a very liquid market (e.g., student loans or auto
loans), removing them from the balance sheet of the originator, and issuing securities with claims
against the cash flows of the assets. The first and still most important type of asset to be securitized
is mortgages, upon which we shall focus.
Prior to the Great Depression, mortgages were difficult for most people to obtain. The reason
was that commercial banks had mainly short-term liabilities (deposits) and did not want to grant
long-term loans against those liabilities. They tended to invest in tradable bonds or the call money
market.4 The largest public source of mortgages at the time was organizations known as building
and loans, as seen in the movie It’s a Wonderful Life. Their supply of capital was limited, because
they drew mainly on small deposits from the local community (Mason 2004). Further, the terms of
such mortgages, such as a 40 percent down payment, were difficult for most people to meet
(Peterson 2007). As a result, homeownership rates were much lower than they are today: 45 percent
in 1920 versus 66.2 percent in 2000, as reported by the U.S. Census Bureau. These numbers
understate how much more difficult it was to obtain a mortgage during that time period, as a greater
percentage of families lived in rural farm areas (with cheaper land) in the 1920s.
During the banking crisis of 1930–1933, the incomplete but functioning mortgage market
collapsed. In order to address the collapse, several new federal agencies were founded, the most
important of which, with respect to securitization, was the Federal National Mortgage Association,
or ‘‘Fannie Mae.’’ Its mission was to buy mortgages, provided the mortgages had already received a
government guarantee from other federal agencies (such as the Veterans Administration). In time
Fannie Mae became an important, though not overwhelming, participant in the mortgage market
(Mason 2004).
In 1968, Fannie Mae became a privately owned corporation and in 1971 began purchasing
mortgages not guaranteed by federal agencies (Bartke 1971). It also helped create a new type of
financial instrument: securities directly backed by mortgages. Along with newcomer Federal Home
Loan Mortgage Corporation, or ‘‘Freddie Mac,’’ and government-owned Government National
Mortgage Association, or ‘‘Ginnie Mae,’’ Fannie Mae issued securities that directly ‘‘passed
through’’ interest and principal to investors (Peterson 2007). These securities are similar to secured
debt; however, pass-through securities pay investors based on the cash flow received, rather than
using a fixed schedule. This new financing vehicle is known as a mortgage-backed security (MBS).
Marketability of early MBSs was substantially enhanced by having government-owned or affiliated
institutions guarantee principal payments—the only risks to investors are that interest rates might
change over the life of the security or that loans would be paid off early.
4
Call money gets its name from the fact that the lender can demand that the loan be repaid at any time.
Accounting Horizons
September 2011
562
Meder, Schwartz, Spires, and Young
Despite privatization, Fannie Mae and Freddie Mac were still prohibited from servicing several
segments of the mortgage market, including loans that exceeded a certain size (‘‘jumbo loans’’) or
loans that otherwise did not meet underwriting standards of the government-affiliated agencies
(Sivesend 1979).5 These markets were serviced by private lenders, but suffered from some of the
same problems that afflicted the mortgage market before the Great Depression. Therefore, an unmet
demand remained for ‘‘securitization’’ of nongovernmental agency mortgages. In 1977, the first
securitization was underwritten by Salomon Brothers and was related to a pool of Bank of America
loans.6
The growth of private mortgage securitization has been truly impressive in the last thirty years.
Starting at nothing in 1977, privately securitized mortgages reached $3 trillion in 2007, according to
the Federal Reserve’s Flow of Funds Report.7 Tirole (2009) provides some economic rationale for
why securitization has become so popular, noting three important benefits: (1) it provides enhanced
liquidity, (2) it provides an investment outlet for emerging economies that have surplus savings, and
(3) it allows for better diversification. Of course, securitization also introduces the potential for
significant costs that are generally associated with information asymmetries (Bernanke and Lown
1991).8
Structured Finance
Of particular note with early MBSs is that all shares were of the same class, so that the credit
risk of any share was identical to the weighted-average credit risk of the underlying mortgages.
However, many potential investors, such as university endowments and pension funds, are
prohibited from investing in less than investment-grade debt (Cantor et al. 2007). 9 Therefore, in
order to increase liquidity for lower quality mortgages successfully, MBSs had to be transformed
into higher-rated securities. This process—referred to as alchemy by some—is a key feature in what
is known as structured finance. Dodd (2007) reports that through proper structuring, 80 percent of
even subprime mortgages can be packaged into investment-grade debt.
Although the underlying legal structure is quite complex, the overarching principle of
structured finance is very simple. Instead of issuing one class of securities, the MBS is broken up
into ‘‘tranches.’’ Each tranche is structured so as to have a targeted risk level, i.e., a particular rating
from a credit-rating agency. To illustrate, a financial intermediary could create a tranche that
incurred losses only if 70 percent of the mortgages defaulted. This tranche would probably be
considered quite safe and hence rated AAA by the rating agencies, making it palatable to almost
5
6
7
8
9
Eventually, Fannie Mae and Freddie Mac entered the subprime mortgage market, becoming involved to a greater
extent after 2000 (Leonnig 2008). However, after 2005 Fannie Mae and Freddie Mac were pushed out of the
subprime market by more aggressive underwriting from private mortgage issuers (Goldstein and Hall 2008). It
was the 2006 and 2007 vintage securitizations that were at the heart of the crisis, as early loans tended to be
refinanced (Zuckerman 2009).
The term ‘‘securitization,’’ in fact, was first coined for this issuance by Salomon Brothers (Ranieri 1996).
The $3 trillion figure includes structured and pass-through securitizations.
For example, Keys et al. (2010) find evidence that loan originators were lax in their screening of borrowers when
the mortgages were more likely to be securitized. There is also evidence that several investment banks bet
against securitized products they sold and profited from their bets (Morgenson and Story 2009). This behavior is
reminiscent of Albert Wiggin, president of Chase National Bank in the 1920s, who sold short his company’s
stock right before the stock market crash, netting a significant sum of money. Even though Wiggin’s actions
were later revealed to Congress, they were not punished because (at that time) they were not illegal.
Investment-grade debt is debt rated BBB or higher by a national rating agency. There are many regulations and
constraints on banks, insurance companies, pension funds and trust funds that make investment-grade debt more
attractive than lower-rated debt (Cantor and Packer 1995). For example, as part of the Basel II bank accords, the
amount of regulatory capital that must be held by banks became a function of the credit rating of the debt
securities they hold (Bowden and Lorimer 2009).
Accounting Horizons
September 2011
Structured Finance and Mark-to-Model Accounting: A Few Simple Illustrations
563
any investor. The tranched MBS is generally referred to as a collateralized mortgage obligation
(CMO).
By carving up MBSs in this way, the capital available for investment is greatly enhanced. In
addition to overcoming restrictions on investments by pension funds, trust funds, and others,
tranching mitigates the adverse selection problem in the mortgage market to a greater extent than
simply securitizing.10 Mitchell (2004) provides a summary of the economic benefits of structured
finance.
Ashcraft and Schuerman (2008, Table 17) provide data on the capital structure of a typical
CMO. In their example there are 17 tranches. The top five tranches are all rated AAA and represent
79.35 percent of the CMO.11 The next nine tranches are rated from AAþ to BBB, with the latter
representing the lowest rating for investment grade securities. These nine tranches are usually
referred to as ‘‘mezzanine tranches.’’ Below BBB are two tranches rated BBþ and BB. Last, there is
an unrated tranche, referred to as over-collateralization or equity tranche. In order to make the sale
of the tranches more palatable to investors, the CMO has to hold in reserve more collateral than the
face value of securities issued. If needed, the over-collateralization will go to pay the principal on
the more senior tranches. However, if after a specified period of time the CMO is performing well,
some of the over-collateralization is released to the equity holders.
The idea behind structured finance is to create as much AAA rated debt as possible; therefore,
financial intermediaries developed methods to ‘‘recycle’’ the mezzanine tranches into new CMOs. A
CMO consisting of other CMO tranches is commonly referred to as a CDO-squared (CDO2).12 The
process of structuring and re-structuring these debt products causes a polarization of debt quality,
wherein a pool of mid-level assets is manufactured into either AAA or speculative grade debt.
Credit Default Swaps
A credit default swap (CDS) is a type of credit derivative. As with all derivatives, a CDS
derives its value from another security, in this case a reference debt security. Although a CDS is
somewhat complex in its mechanics, it can be thought of as a type of insurance. The issuer of a
CDS agrees to pay its holder a specified amount conditional on the default of a reference debt
security. Typically this amount is the par value of the debt, less any value the debt has in default. In
exchange, the holder pays the issuer quarterly premiums, very much like insurance premiums.
Therefore, the value of a CDS to the holder increases as the probability of default on the reference
debt security increases. The value to the issuer is the net of discounted premiums and discounted
expected indemnity payments.
Although there had been a few primitive predecessors, the first real CDS was created by J. P.
Morgan in the mid-1990s. Its goal was to offload credit risks it otherwise would have held on its
balance sheet, specifically a large loan to Exxon in the wake of the Exxon Valdez oil spill. J. P.
Morgan structured a deal wherein it agreed to pay the European Bank for Reconstruction a fixed
periodic premium while the bank assumed the credit risk on the Exxon loan (Tett 2009). Since that
time, the issuance of CDSs has exploded, helped particularly by the lack of regulation ensured with
10
11
12
The basic idea is that when there are informed and uninformed investors, the issuer wants to split the security
into a risk-free bond and a risky asset (Gorton and Pennacchi 1990). The risk-free bond is sold to the uninformed
investors, and the risky asset is sold to the informed investors. Structured finance is a method of approximating
this outcome.
We use the Standard & Poor’s rating system. The higher the letter the lower the default risk is. Also, repeated
letters indicate a lower default risk. So, AAA has lower default risk than BBB and BBB has lower default risk
than BB.
CDO is an acronym for Collateralized Debt Obligation, and as the name implies, it holds a wider variety of debt
than just mortgages. Our example might be more accurately described as a CMO-squared, but that term is rarely
used.
Accounting Horizons
September 2011
564
Meder, Schwartz, Spires, and Young
the passage of the Commodities Futures Modernization Act of 2000 and the standardization of
contract terms by the International Swaps and Derivatives Association (ISDA) in 2003.13 Although
estimates vary, by 2006 the notional value of CDSs outstanding was thought to be somewhere
around $45 trillion (Mengle 2007). That is, the amount of debt ‘‘insured’’ was $45 trillion. It is
important to note that the issuance of CDSs is not limited by the amount of underlying debt
outstanding. For example, in one particular bankruptcy in 2005 there were $28 billion in CDSs on
$2.2 billion of underlying debt (Mengle 2007). Although originally developed for bank loans to
corporations, CDSs have evolved to cover sovereign debt, indices of debt instruments, and
structured finance (Mengle 2007).14
Fair Value Accounting
Fair value accounting, in theory, ‘‘seeks to capture and report the present value of future cash
flows associated with an asset or a liability’’ (Campbell et al. 2008, 32). In practice, the FASB
interprets fair value to be the amount an unrelated party would pay for an asset (or demand to
assume a liability) in an orderly transaction on the measurement date (FASB 2007).15 The FASB’s
definition is silent regarding any difficulties related to market frictions, including those resulting
from imperfect and incomplete markets (Bignon et al. 2009; Christensen and Frimor 2007; Demski
et al. 2008, 2009; Fellingham 2010, Chap. 5).16 We do not address these market frictions directly,
although they are of first-order importance in determining ‘‘fair values’’ for many classes of assets.
We focus on three of the many accounting standards that relate to fair value accounting.17 The
first, SFAS 115, issued in 1993, classified financial assets into three types (FASB 1993). Held-tomaturity securities are debt instruments, for which the holder has both the intention and ability to
hold the instruments until the principal becomes due. They are accounted for using amortized
historical cost. Other than amortization, their value is adjusted only for ‘‘other than temporary’’
declines in value, in which case the losses are charged against either earnings or other
comprehensive income (OCI), depending on the reason for the loss.18 Trading securities, such as
those used by Goldman Sachs for proprietary trading purposes, are recorded at fair value, and
changes in fair value are passed directly through earnings. Available-for-sale securities are those
securities that are neither held-to-maturity nor trading. These assets are recorded at fair value, and
changes in fair value are initially passed through OCI; gains or losses do not impact earnings until
the assets are sold, other than as explained above.
The second accounting standard we consider, SFAS 133 (FASB 1998), addresses financial
derivatives. Prior to the issuance of SFAS 133, derivative instruments were subject to a variety of
13
14
15
16
17
18
The documentary film The Warning, broadcast in the U.S. in 2009 on public television, is helpful in
understanding the history of credit derivative regulation.
CDSs have even been used to create a structured financial product known as a synthetic collateralized debt
obligation (synthetic CDO). In a synthetic CDO, CDSs are issued against assets held by others, such as
mortgages or tranches of CMOs that hold mortgages. The CDO collects quarterly premiums in exchange for
issuing the CDSs and must pay the holders of the CDSs in case the ‘‘insured’’ CMO tranche defaults. In a typical
structure, only a small percent of the notional value of the debt insured by the CDSs is covered by funds received
from outside investors. The unfunded portion, which represents the most senior positions in the structure, is held
on the books of the issuer as ‘‘super-senior’’ risk (Tett 2009, 61–66). The collapse of AIG was largely caused by
their writing CDSs on tens of billions of dollars of super-senior risk (Lewis 2009).
See SEC (2008) for a brief history of fair value accounting in the United States.
Also, Christensen and Demski (2003) and Demski (2004) discuss the difficulty in applying fair value accounting.
They argue that to value assets correctly one must take into account the reason for the assets’ acquisition and use
by the firm, which is of course endogenous.
Ryan (2008a) provides an extended discussion of fair value accounting and structured financial products.
As stated in staff position FSP 115-2 (FASB 2009b), credit losses (borrower’s inability to pay) are charged to
earnings; all other losses are charged to OCI. This rule also applies to available-for-sale debt securities.
Accounting Horizons
September 2011
Structured Finance and Mark-to-Model Accounting: A Few Simple Illustrations
565
rules that were incomplete and allowed similar derivatives to be treated differently. The statement
focuses on derivatives used for hedging, such as airlines buying jet fuel futures. In addition, SFAS
133 states that derivatives not held for hedging are to be reported at fair value, with changes in
value run through earnings. This accounting standard would seemingly apply to the issuer of CDSs.
However, there were very few CDSs when SFAS 133 was issued, and CDSs did not seem to have
been contemplated in the original statement. In succeeding years, the FASB observed the explosion
of CDS use and in 2008 clarified the application of SFAS 133 with FSP 133-1 (FASB 2008b). This
statement of position requires the liability of CDS issuers to be stated at fair value, as well as the
disclosure of significant issues regarding the CDS, such as notional value.
The third and most significant accounting standard related to our discussion is SFAS 157
(FASB 2007). SFAS 157 does not change the requirements for fair value accounting per se. Rather,
it provides overarching guidance on how it is to be applied (Christian and O’Reilly 2009). SFAS
157 introduces a classification scheme for assets conditioned on the type of information available
for their valuation. Level 1 assets are valued using current market prices for identical assets in wellfunctioning markets. Level 2 assets are valued using current market prices in illiquid markets or
current market prices for similar but not identical assets, as well as analytical techniques. Level 3
assets are valued using nonmarket inputs, such as models estimating future cash flows associated
with the assets. Ryan (2008a) characterizes the various levels as follows: Level 1 is pure mark-tomarket accounting, Level 3 is pure mark-to-model accounting, and Level 2 is a hybrid.
Given the problems of illiquidity that have recently gripped many markets, the FASB has
issued continuing guidance on when it would be appropriate to set aside current market prices
because they are not indicative of fair value (FASB 2008a, 2009a). What is most interesting is the
FASB did not add any guidance on acceptable nonmarket methods of evaluation, but simply
clarified and expanded the situations where market inputs can be set aside (Christian and O’Reilly
2009). The existing high degree of discretion is likely to increase the use of models in fair value
accounting.19
The use of models to value structured financial products should not be thought of as being
confined to extreme, once-in-a-generation events. Structured financial products tend to be one-of-akind securities that are thinly traded, as a rule (Zuckerman 2009). Therefore, even under normal
conditions some modeling would be necessary. Laux and Leuz (2010) observe that banks were
using cash flow models to value structured assets in 2007, well before the worst of the financial
crisis hit. Further, Zuckerman (2009) describes how difficult it was in 2007 and 2008 for holders of
certain structured financial products and related derivatives to get meaningful price quotes from
broker dealers, forcing the use of some modeling to value the securities or derivatives.
MODEL
Our model borrows the basic structure suggested by Coval et al. (2009). However, we simplify
the analysis and allow the systematic risk to be more transparent. Therefore, unlike Coval et al.
(2009), we have no need to utilize complex computer simulations. Instead, we supply more
straightforward calculations that can be easily replicated.
No Systematic Risk
To begin the analysis, suppose a CMO is backed by three mortgages. We normalize the
principal balance to one per mortgage. We assume there is no interest paid on the mortgages and no
19
See Negus and Boyles (2009) for practical guidance on valuing illiquid securities under recent FASB
pronouncements.
Accounting Horizons
September 2011
Meder, Schwartz, Spires, and Young
566
recovery in default. Therefore, at some point in the future each mortgage pays either 1 or 0. We also
assume the discount rate is 0. These assumptions greatly simplify the analysis, without any
significant loss of insight. Finally, we assume that the probability of default is 0.1 for each
mortgage. We initially assume the probability of default is independent across mortgages. That is,
there is no systematic risk.
Our CMO comprises three tranches. Tranche A pays 1 to the security holder if fewer than three
mortgages default. Tranche B pays 1 if fewer than two mortgages default. Tranche C pays 1 if no
mortgages default. Panel A of Table 1 provides valuation figures for each tranche. The panel shows
that structured finance techniques have converted an MBS with an average probability of default of
0.1 into three tranches, with two of them having less than a 0.03 probability of default and one
TABLE 1
Valuation of Tranches for a Three-Mortgage CMO
Panel A: No Systematic Risk; Probability of Default = 0.1
Tranche A
Tranche B
Tranche C
a
b
c
Probability Does
Not Pay Out
Probability Pays Out =
Mark-to-Model Value
Issuer Liability
of 1,000 CDSs
0.001a
0.028b
0.271c
0.999
0.972
0.729
1
28
271
(0.1)3.
(0.1)3 þ (3)(0.1)2(0.9).
(0.1)3 þ (3)(0.1)2(0.9) þ (3)(0.1)(0.9)2.
Panel B: Probability of Bad Economy = 0.5; Conditional Probability of Default = 0.2
Tranche A
Tranche B
Tranche C
d
e
f
Probability Does
Not Pay Out
Probability Pays Out =
Mark-To-Model Value
Issuer Liability
of 1,000 CDSs
0.004d
0.052e
0.244f
0.996
0.948
0.756
4
52
244
(0.5)(0.2)3.
(0.5)(0.2)3 þ (0.5)(3)(0.2)2(0.8).
(0.5)(0.2)3 þ (0.5)(3)(0.2)2(0.8) þ (0.5)(3)(0.2)(0.8)2.
Panel C: Probability of Bad Economy = 0.2; Conditional Probability of Default = 0.5
Tranche A
Tranche B
Tranche C
Probability Does
Not Pay Out
Probability Pays Out =
Mark-to-Model Value
Issuer Liability
of 1,000 CDSs
0.025g
0.100h
0.175i
0.975
0.900
0.825
25
100
175
g
(0.2)(0.5)3.
(0.2)(0.5)3 þ (0.2)(3)(0.5)2 (0.5).
i
(0.2)(0.5)3 þ (0.2)(3)(0.5)2 (0.5) þ (0.2)(3)(0.5)(0.5)2.
In the probability calculations a–i, the first term is the probability of all three mortgages defaulting, the second term (if
applicable) is the probability of two mortgages defaulting, and the third term (if applicable) is the probability of one
mortgage defaulting.
h
Accounting Horizons
September 2011
Structured Finance and Mark-to-Model Accounting: A Few Simple Illustrations
567
having a probability of default equal to 0.271.20 This structuring would likely qualify both Tranche
A and B for selection by many pensions and endowments for which they would not otherwise have
qualified.21 For Tranche A, the risk is only slightly more than a U.S. Treasury Bill.
Even though the structured finance techniques have created securities with different risk levels
than the underlying securities, certain features of the underlying securities remain intact. For
example, the average default risk (probability of default) of the three tranches is (0.001 þ 0.028 þ
0.271)/3 = 0.1, which is the same as the overall risk of the underlying CMO. In addition, the payout
promised by each tranche is supported by the underlying CMO. That is, if no mortgages default, the
total received from the mortgages in the CMO is 3, which is exactly the amount needed to satisfy
the requirement that each of the three tranches pays 1. If one mortgage defaults, the total received
from the mortgages in the CMO is 2, which would allow Tranche A and Tranche B each to pay 1.
The analysis is similar for two and three defaults.
Systematic Risk
The reason securitization works so well toward reducing risk in our first example is we
assumed there was no systematic risk. Thus, the protection of the senior tranche, Tranche A, was
achieved through diversification. Given the assumptions that the probability of any particular
mortgage defaulting is 0.1 and statistical independence between mortgages, it is very unlikely that
all three mortgages would default.
Prior to the financial crisis, it was thought that diversifying across borrowers, regions, and loan
amounts would protect the holders of mortgage securities from any localized disruptions in property
values, such as that resulting from the closure of a factory. The potential for systematic risk in the
form of a nationwide decrease in home prices was downplayed. Zuckerman (2009) reports financial
analysts in 2005 as saying, ‘‘Home prices have never gone negative’’ and ‘‘They won’t even go
flat.’’ Joseph Cassano, who headed the AIG financial products unit that sold the CDSs, ultimately
leading to the firm’s demise, said as late as August 2007 in a conference call, ‘‘It is hard for us, and
without being flippant, to even see a scenario within any kind of realm of reason that would see us
losing $1 in any of those transactions’’ (Cohan 2010).
Ryan (2008a) provides an explanation for how systematic risk can be present in the
mortgage market, at least for the case of subprime mortgages. He points out subprime borrowers
generally have a hard time paying their mortgage from current income. Therefore, their loans are
paid off mainly through refinancing. This can be easily accomplished in a period of rising house
prices. However, when prices level off or fall, refinancing is not a viable option. Ryan (2008a,
1608) states, ‘‘[I]nvestment performance of these [subprime] positions exhibits a binary quality
that depends on subprime mortgagors’ ability to obtain cash-out refinancing.’’ That is, they tend
to all perform well, or all perform poorly.22 Tett (2009) describes another type of systematic risk,
in the form of a negative feedback loop. In the event that houses in an area become vacant or
neglected, the values of neighboring houses decline, causing greater defaults on these
neighboring houses and more neglect. Finally, Guiso et al. (2009) describe a type of social
contagion. They find that most people continue to pay on their mortgage even if they owe more
20
21
22
By valuing the tranches in this way, we are essentially using the present value technique described in SFAS 157
(FASB 2007, 37–43). However, SFAS 157 requires that the discount rate incorporate the risk associated with
model error. For simplicity we assume the risk adjusted discount rate is zero.
Using the long-term default data provided by Coval et al. (2009), the lowest level of investment-grade debt has a
default rate of 7.26 percent, whereas the highest level of junk debt has a default rate of 10.18 percent.
See Zuckerman (2009, 100–102) for further support of Ryan’s (2008a) assertions regarding the effect of home
appreciation on the performance of subprime loans.
Accounting Horizons
September 2011
568
Meder, Schwartz, Spires, and Young
than their property is worth. However, if they begin to hear about others defaulting, they
themselves are more likely to default.
We now introduce systematic risk into the model. To capture the phenomenon described
previously, suppose mortgages will default only if the economy is bad. In notation, the probability
the economy is good or bad is Pr(G) and Pr(B), respectively. If the economy is bad, there is a
probability p that any individual mortgage will default, with default being independent events
(conditional on the economy being bad). That is, Pr(defaultjG) = 0 and Pr(defaultjB) = p. We
begin by assuming Pr(G) = Pr(B) = 0.5 and p = 0.2. These assumptions are consistent with the
original assumption that the unconditional probability of any particular mortgage defaulting is 0.1,
calculated as: Pr(G) Pr(defaultjG) þ Pr(B) Pr(defaultjB) = 0.5 (0) þ 0.5 (0.2) = 0.1. Because the
common state (B) has a positive probability of default for each mortgage, it creates (an
unconditional) correlation in defaults across mortgages.23 We show the mark-to-model valuations
of the three tranches in Panel B of Table 1.
Introducing correlation among mortgage defaults while keeping the unconditional probability
of default constant does not change the overall quality of the CMO. It does, however, change the
value of the individual tranches, and in such a way that harms high-level tranches while
benefitting low-level tranches. Specifically, in our example, the probability of default in Tranche
A increases from 0.001 to 0.004, while the probability of default in Tranche C declines from
0.271 to 0.244. Tranche A suffers due to the greater likelihood of all three mortgages defaulting
while Tranche C benefits due to the greater likelihood of none of the mortgages defaulting.
Moving from Panel B to Panel C of Table 1, where Pr(B) = 0.2 and Pr(defaultjB) = p = 0.5, the
incidence of default is even more concentrated in the bad state. Tranche A now has a probability
of default of 0.025, a 25-fold increase over the no systematic risk case. The 2.5 percent default
rate is still low but it is hardly in the category of a U.S. Treasury security. Tranche C now has a
0.175 probability of default, which is close to the unconditional probability of default on a single
mortgage of 0.1.
Although the introduction of correlation has not changed the overall quality of the CMO, that
does not mean that erroneously omitting correlation from valuation models is not costly. The lowlevel tranches that do better with correlation are generally held by hedge funds and others who
presumably are structured to bear greater risk, while the high-level tranches that do worse with
correlation tend to be held by more risk-averse entities. Therefore a misspecified model can cause
overall social welfare to decrease, quite dramatically perhaps, even if overall expected cash flow is
unaffected. A related friction is that not all tranche holders are likely to use the same valuation
assumptions. It may be that those tranche holders who benefit most from correlation assume more
of it than those tranche holders who benefit least.
In summary, a comparison of the three panels of Table 1 illustrates that in marking structured
financial assets to a model, an important characteristic to incorporate is the potential for correlated
asset returns. Although individuals generally may have a good appreciation of the unconditional
likelihood of a particular event, they tend to underestimate the potential for systematic risk. Taleb
(2001) points out such errors can have a profound impact on investor behavior. Further, Salmon
(2009) notes even seasoned financial professionals may underestimate the potential for catastrophic
failure brought on by correlated asset returns.
23
Other modeling attributes that must be considered besides simple correlation are pre-payment risks, depth of the
tranche (shallower tranches will be wiped out more quickly), recovery rates (including costs to repossess and
rehabilitate properties), correlation between default clustering and recovery rates (in times of recession there
tends to be a clustering of defaults coupled with poor recovery rates [Altman et al. 2005]) and agency conflicts
(Ashcraft and Schuermann [2008] cite six examples).
Accounting Horizons
September 2011
Structured Finance and Mark-to-Model Accounting: A Few Simple Illustrations
569
Fair Value of CDSs
We now consider the effect of systematic risk on CDSs. As mentioned, a CDS can be thought
of as a type of insurance. CDSs must be recorded at ‘‘fair value’’ (FASB 1998, 2008b). The
protection issuers of CDSs have when issuing a CDS on a high-level tranche is many mortgages
must default for the tranche to fail. For the CDS in our illustration, suppose an issuer pays 1 in the
event a tranche does not pay out and 0 otherwise. Further, as a simplification, assume quarterly
premiums are zero. Hence, the fair value liability of a single CDS is equal to the probability of
default multiplied by the par value of the debt, which in our case is 1. The right-most column in
Table 1 reports the liability for 1,000 CDSs for the various tranches.
If one does not consider systematic risk, the liability resulting from the CDS on a high-level
tranche is relatively low. However, moving from no systematic risk to even a small amount of
systematic risk can drastically increase the liability. In Table 2, we concentrate on CDSs and
provide data on four levels of correlation between mortgage defaults. Comparing Column 1, where
there is no systematic risk, to Column 2 with some systematic risk Pr(B) = 0.5 and p = 0.2), there
is a sharp difference in the liability of CDS issuers on the most senior tranche relative to the
difference in value for the actual holders of the tranche. While the value of Tranche A declines less
than 1 percent, from 0.999 to 0.996, the expected liability of CDS issuers increases by 400 percent.
Comparing Column 1 to Column 3 where there is greater systematic risk (Pr(B) = 0.2 and p = 0.5),
the expected liability of CDS issuers increases 2,500 percent. One of the truisms of derivative
products is they tend to magnify both the upside and downside risks relative to the securities from
TABLE 2
The Effect of Systematic Risk on the Expected Liability of 1,000 CDSs
Pr(B) = 0.1
Pr(B) = 0.2
Pr(B) = 0.5
Pr(B) = 1.0
p = 1.0
p = 0.5
p = 0.2
p = 0.1
Pr(default) = 0.1 Pr(default) = 0.1 Pr(default) = 0.1 Pr(default) = 0.1
Scenario
0
0.111
0.444
1
Correlation Coefficient
between Any Two
Mortgages
Liability from Tranche A
Liability from Tranche B
Liability from Tranche C
1
28
271
4
52
244
25
100
175
100
100
100
Liabilities from A, B, and C
300
300
300
300
Pr(B) = probability of bad economy.
p = probability of default, conditional on bad economy.
Correlation coefficients measure the degree of association between two mortgages defaulting.
Correlation coefficient of two random variables:
X; Y =
Covariance ðX; YÞ
:
rðXÞrðYÞ
Example of the calculation of the Correlation Coefficient assuming Pr(B) = 0.2 and p = 0.5:
Covariance (X,Y) = E (XY) E (X)E(Y).
Assume that the event of default = 1, the event of non-default = 0, and that X and Y are random variables of the
outcome of any two mortgages. Then:
E (XY) = 0.05 and E (X) = E (Y) = 0.1 so that E (X) E (Y) = 0.01; therefore Covariance (X,Y) = 0.04.
r(X) = r(Y) = 0.3 so that r(X)r(Y) = 0.09.
Correlation Coefficient = 0.04/0.09 = 0.444.
Accounting Horizons
September 2011
570
Meder, Schwartz, Spires, and Young
which they derive their values. Table 2 clearly shows the potential for large losses accruing from
CDSs.24
CDO-Squareds
We close our analysis with an illustration of a CDO-squared. As mentioned, a CDO-squared
typically comprises the middle tranches of other structured financial products, such as CMOs. The
idea is to take tranches that are not rated highly enough to gain the desired liquidity and to
repackage them so that most will now be saleable to a broad range of investors. Using the structure
we developed above, suppose there are many three-tranche CMOs identical to the one described
earlier. Further, suppose a new CMO is formed using the middle tranches (Tranche B) of eight of
our original three-tranche CMOs. The new CMO, denoted CMO 0 , has eight tranches, labeled A 0
through H 0 . The CMO tranches will pay either 1 or 0, and, as before, there is no interest. Tranche A 0
pays 1 if seven or fewer original tranches fail; Tranche B 0 pays 1 if six or fewer original tranches
fail, and so on.
In Table 3, Panel A, we present the expected default rates given a default rate on the original
middle tranche of 0.028, derived from a 0.1 probability of default on mortgages with no systematic
risk (see Panel A of Table 1). We focus on Tranche F 0 . Assuming there is no systematic risk, the
security has a very low probability of default (0.001). Using the ten-year default statistics found in
Coval et al. (2009), the security would earn an AAA debt rating.
In Panel B of Table 3, we consider the same CMO 0 , assuming instead that systematic risk is
present: Pr(B) = 0.5 and p = 0.2. Therefore, the original middle tranche has a default probability of
0.052 instead of 0.028 (see Panel B of Table 1). Table 3 shows the probability of default on Tranche
F 0 has risen from 0.001 to 0.021, with the value of the tranche decreasing from 0.999 to 0.979. At
first, this decline in value may not seem substantial; however, many of the financial institutions
holding structured financial products are highly leveraged. An SEC rule change in 2004, putatively
to help financial institutions remain competitive with their European counterparts, effectively
eliminated leverage restrictions on the broker-dealers (Labaton 2008).25 Bear Stearns, for example,
before its collapse, had issued $13.4 trillion of notional value derivatives against $80 billion in
capital (Evans-Pritchard 2008). Even crisis survivor Goldman Sachs was reported to have a
leverage ratio of 25 to 1 prior to 2008, which means that an asset decline of 4 percent would wipe
out their entire capital (Kassenaar and Harper 2008). Given this extreme degree of leverage, even
small errors in fair value accounting can have large consequences on how investors and regulators
would view a firm.
Perhaps more important, the contingent liability of 1,000 CDSs has increased from 1 to 21. As
described in Tett (2009), Patterson (2010), and Zuckerman (2009), by 2006 much of the issuance of
structured financial products was of the synthetic variety (that is, CDOs that were created from
CDSs). Further, with AIG pulling out of the market for insuring super-senior risk, these CDSs were
issued by the organizers such as Lehman Brothers and Merrill Lynch (Tett 2009). For highly
leveraged institutions, the rise in expected liabilities and collateral calls from CDS issuance can
have catastrophic consequences.
24
25
CDS issuers need not wait for actual default to feel the pain of rising CDS liability. As part of the CDS contract,
issuers must put up collateral if the value of the underlying security falls or if their credit rating declines. The
demise of AIG was as much due to calls for additional collateral as actual defaults (Sorkin 2009).
Leverage limits are much stricter for depositary institutions, such as commercial banks and thrifts. National
agencies, such as the Federal Reserve and the FDIC, and international agencies, such as the Basel Committee on
Banking Supervision, take a more proactive role in banking regulations. Many have blamed the financial crisis
on the so-called shadow-banking system a system of near banks that perform many of the functions of banks
without the regulation or capital of depositary institutions (Farhi and Cintra 2009).
Accounting Horizons
September 2011
Structured Finance and Mark-to-Model Accounting: A Few Simple Illustrations
571
TABLE 3
CDO-Squared Formed from Eight Middle Tranches (Tranche B) of the Three-Tranche CMO
Panel A: CDO-Squared—No Systematic Risk; Risk-Default Probability = 0.028
A
Tranche
Tranche
Tranche
Tranche
Tranche
Tranche
Tranche
Tranche
Maximum Number
that Can Fail for
Tranche to Pay Out (m)
B
Unconditional
Probability Tranche
Does Not Pay Out
C
Unconditional Value
of Tranche =
Mark-to-Model Value
D
Issuer Liability
of 1,000 CDSs
E
7
6
5
4
3
2
1
0
3.77802E-13
1.05299E-10
1.28532E-08
8.97922E-07
3.93036E-05
0.001105884
0.019618665
0.203235236
.0.999999999
.0.999999999
0.999999987
0.999999102
0.999960696
0.998894116
0.980381335
0.796764764
,1
,1
,1
,1
,1
1
20
203
A0
B0
C0
D0
E0
F0
G0
H0
Based on Table 1, Panel A, where unconditional probability tranche B pays out = 0.972.
m = maximum number of Tranche Bs that can fail for tranche to pay out.
Column C:
X8
i = mþ1
8!
0:028i ð1 0:028Þ8i :
i!ð8 iÞ!
Column D: 1 Column C.
Column E: 1,000 (1 Column D).
Panel B: CDO-Squared—Systematic Risk; Risk-Default Probability = 0.052
A
Tranche
Tranche
Tranche
Tranche
Tranche
Tranche
Tranche
Tranche
Maximum Number
that Can Fail for
Tranche to
Pay Out (m)
B
Probability
Tranche
Does Not Pay
In Bad Economy
C
Unconditional
Probability
Tranche Does
Not Pay Out
D
Unconditional
Value of Tranche =
Mark-to-Model
Value
E
Issuer
Liability
of 1,000
CDSs
F
7
6
5
4
3
2
1
0
1.36857E-08
9.56946E-07
2.93999E-05
0.000519493
0.005797421
0.042174522
0.198875883
0.58460231
6.84285E-09
4.78473E-07
1.46999E-05
0.000259747
0.00289871
0.021087261
0.099437942
0.292301155
0.99999999
0.99999952
0.9999853
0.99974025
0.99710129
0.97891274
0.90056206
0.70769885
,1
,1
,1
,1
3
21
99
292
A0
B0
C0
D0
E0
F0
G0
H0
Based on Table 1, Panel B, where unconditional probability tranche B pays out = 0.948.
Pr(B) = probability of bad economy = 0.5.
p = conditional probability of default given bad economy = 0.2.
m = maximum number of tranche Bs that can fail for tranche to pay out.
Pr(tranche B does not pay j p = 0.2 and Economy = Bad) = 0.23 þ (3)(0.2)2 (0.8) = 0.104.
Column C:
X8
i = mþ1
8!
0:104i ð1 0:104Þ8i :
i!ð8 iÞ!
Column D: Pr(B) Column C þ Pr(G) 0 = 0.5 Column C.
Column E: 1 Column D.
Column F: 1,000 (1 Column E).
Accounting Horizons
September 2011
572
Meder, Schwartz, Spires, and Young
Models in Practice
As noted above, modeling in practice is a rather complex task. In order to give an overview of
modeling techniques in practice, we focus on credit-rating agencies. We do so because it is their
purpose to evaluate debt securities without the benefit of price discovery. Also, we know much
more about their models than those used by investment banks or hedge funds. Finally, there are
strong indications that their ratings were heavily relied upon by unsophisticated purchasers of
structured financial products (Adelino 2009; Ammer and Clinton 2004).
The rating agencies use what is called a ‘‘through-the-cycle’’ default likelihood estimation
approach (Löffler 2004). This means they make no attempt to estimate the timing of recessions per
se. Instead, they evaluate the debt instrument across the entire business cycle. In formulating their
rating, the agency considers the likelihood of a stressful event during the term of the debt and the
consequences of plausible outcomes associated with the stressful event. As an example, there might
be a 30 percent chance of a severe recession during the term of the debt in question, with a severe
recession resulting in a 6 percent mortgage default rate. In this sense, the credit-rating agencies’
approach is similar to ours. However, in reality there is a continuum of states—and this is just the
beginning of the analysis. Rating agencies must look at the specific type of mortgages held and their
vulnerability to economic downturns, the potential for low recovery rates, and the effect of
geographic diversification. We have previously mentioned that subprime mortgages tend to
experience very high default rates when home prices stop increasing. Therefore, for an instrument
primarily composed of subprime mortgages, the rating agencies must assess the likelihood that
home prices will remain flat (or decline) over the term of the instrument being rated, whether the
price stagnation will be localized or national, and whether the instrument is geographically
diversified or concentrated. Historical data will be the main source of analysis, but care must be
taken not to assign zero probability to events that have yet to be observed.
All of the credit-rating agencies use some form of the Gaussian copula to model correlation
(Coval et al. 2009). Amato and Gyntelberg (2005, 80–81) provide a simplified illustration and
explanation of Gaussian copula methods. These methods, especially the ‘‘single factor method’’ that
became the dominant approach in the early part of the last decade, have drawn serious criticism
(Donnelly and Embrechts 2010; Salmon 2009). For example, one criticism is the method does not
take into consideration the negative correlation between default clusters and recovery rates. In
response to both critics and obvious flaws, modeling of structured financial products has continued
to evolve (Brigo et al. 2010). However, unlike the physical sciences, the phenomenon of interest
responds to the model. If markets get too comfortable with a model, they can behave in ways that
invalidate the underlying model assumptions. One could argue this should have been the lesson
learned in the collapse of Long-Term Capital Management in 1998 (Clark 2000).
Although the rating agencies are mainly interested in the probability of default, the probability
of default is a key input, along with term, coupon rate, and risk-free rate, in determining the value of
the instrument. One note of caution, though, credit ratings have only a single parameter mapping
into default probabilities (or more generally expected losses). However, as investors consider the
use of ratings in valuing structured financial products, they should be aware that the variance (and
higher order moments) of the returns of CMOs tends to be larger than a claim on an untranched
portfolio of mortgages.26
26
Tranches tend to have all-or-nothing outcomes. Once enough losses have occurred to breach the lower boundary
of the tranche, it will quickly become worthless. In contrast, a claim on an untranched portfolio of mortgages
with positive recovery rates cannot have a zero outcome and, therefore, untranched claims tend to have a less
variable payout (CGFS 2005). See CGFS (2008) for proposals to add a second parameter to credit ratings of
structured financial products. We thank an anonymous reviewer for pointing out the importance of higher order
moments of the outcome distribution.
Accounting Horizons
September 2011
Structured Finance and Mark-to-Model Accounting: A Few Simple Illustrations
573
Of course, we are interested not only in how values are modeled, but whether modeled values
have information content. The evidence so far is consistent with prior research on other accounting
methods with considerable management discretion; the accounting valuations have information
content but are subject to manipulation. Goh et al. (2009), Kolev (2008), and Song et al. (2010) all
find that mark-to-model valuations have information content, but less information content than
mark-to-market valuations. Also, the information content of mark-to-model is increasing in
measures of good corporate governance. However, there is also evidence of mark-to-model being
used as an earnings management tool, both to increase earnings if necessary (Huizinga and Laeven
2009) and to take an ‘‘earnings bath’’ so as to boost future reported income (Fiechter and Meyer
2010). Finally, there is evidence that the percentage of assets disclosed as Level 3 is informative,
with higher percentages meaning lower liquidity (Lev and Zhou 2009).
CONCLUSION
Ryan (2008a, 1606) deems the recent financial crisis as ‘‘the signal researchable-teachable
moment of [his] two-decade-plus career as an accounting academic focused on financial reporting
by financial institutions for financial instruments and transactions.’’ We have taken advantage of
some of the issues exposed by the crisis to illustrate the critical nature of systematic risk in
modeling the value of structured financial products. It is obvious now that correlated returns on
certain types of assets underlying structured financial products can affect their values far beyond
historical norms. Yet the Bank for International Settlements (CGFS 2008) reports that credit-rating
agencies and their customers were surprised by the degree of correlation and its effect on the value
of some structured financial products. Similarly, Salmon (2009) reports that many in the financial
community knew of correlation modeling inadequacies and yet continued to act as if the problems
did not exist.
Our goal is to provide a brief overview of structured financial products and to explore the
sensitivity of their valuation to systematic risk. One reason this is important is the FASB has greatly
expanded the circumstances under which modeling is an appropriate valuation technique. Another
reason is that many structured financial products are inherently unique and hence difficult to value
without some modeling to augment market observables. Finally, as the accounting standards for fair
value continue to come under attack and are likely to face further review, nonmarket data may have
an even greater role in valuing financial assets.
REFERENCES
Adelino, M. 2009. Do Investors Rely Only on Ratings? The Case of Mortgage Backed Securities. Working
paper, Massachusetts Institute of Technology.
Allen, F., and E. Carletti. 2008. Mark-to-market accounting and liquidity pricing. Journal of Accounting
and Economics 45 (August): 358–378.
Altman, E., B. Brady, A. Resti, and A. Sironi. 2005. The link between default and recovery rates: Theory,
empirical evidence, and implications. Journal of Business 78: 2203–2227.
Amato, J., and J. Gyntelberg. 2005. CDS index tranches and the pricing of credit risk correlations. BIS
Quarterly Review (March): 73–87.
Ammer, J., and N. Clinton. 2004. Good news is no news? The impact of credit ratings changes on the prices
of asset backed securities. In International Finance Discussion Papers, No. 809. Washington, D.C.:
Board of Governors of the Federal Reserve System.
Ashcraft, A., and T. Schuerman. 2008. Understanding the Securitization of Subprime Mortgage Credit.
Staff Report No. 318. New York, NY: Federal Reserve Bank of New York.
Ball, R. 2008. Don’t blame the messenger . . . or ignore the message. Available at: http://faculty.
chicagobooth.edu/brian.barry/igm/ShootingtheMessenger2008-10-12.pdf
Accounting Horizons
September 2011
574
Meder, Schwartz, Spires, and Young
Bartke, R. 1971. Fannie Mae and the secondary mortgage market. Northwestern Law Review 66 (1): 1–78.
Bernanke, B., and C. Lown. 1991. The credit crunch. Brookings Papers on Economic Activity: 205–247.
Bignon, V., Y. Biondi, and X. Ragot. 2009. An Economic Analysis of Fair Value: Accounting as a Vector of
Crisis. Cournot Centre for Economic Studies (August). Available at: http://papers.ssrn.com/sol3/
papers.cfm?abstract_id=1474228#
Bowden, R., and D. Lorimer. 2009. Credit Securitization and System Stability. Working paper, Ulm
University.
Brigo, D., A. Pallavicini, and R. Torresetti. 2010. Credit Models and the Crisis: A Journey into CDOs,
Copulas, Correlations and Dynamic Models. Hoboken, NJ: Wiley Finance.
Campbell, R., L. Owens-Jackson, and D. Robinson. 2008. Fair value accounting from theory to practice.
Strategic Finance 90 (1): 31–37.
Cantor, R., O. Gwilym, and S. Thomas. 2007. The use of credit ratings in investment management in the
U.S. and Europe. Journal of Fixed Income 17 (2): 13–26.
Cantor, R., and F. Packer. 1995. The credit rating industry. The Journal of Fixed Income 5 (3): 10–34.
Christensen, J., and J. Demski. 2003. Accounting Theory: An Information Perspective. Boston, MA:
McGraw-Hill/Irwin.
Christensen, J, and H. Frimor. 2007. Fair value, accounting aggregation and multiple sources of
information. In Essays in Honor of Joel S. Demski, edited by R. Antle, F. Gjesdal, and P. Liang, New
York, NY: Springer.
Christian, C., and D. O’Reilly. 2009. FASB issues guidance on estimating fair value. Bank Accounting and
Finance (August-September): 21–28.
Cifuentes, R., G. Ferrucci, and H. Shin. 2005. Liquidity risk and contagion. Journal of the European
Economic Association 3: 556–566.
Clark, M., producer. 2000. The trillion dollar bet. Available at: http://www.pbs.org/wgbh/nova/
stockmarket/
Cohan, W. 2010. The fall of AIG: The untold story. Institutional Investor. Available at: http://www.
iimagazine.com/InstitutionalInvestor/Articles/2460649/The_Fall_of_AIG_The_Untold_Story.html
Committee on the Global Financial System (CGFS). 2005. The Role of Ratings in Structured Finance:
Issues and Implications. Basel, Switzerland: Bank for International Settlements.
Committee on the Global Financial System (CGFS). 2008. Ratings and Structured Finance: What Went
Wrong and What Can Be Done to Address Shortcomings? Basel, Switzerland: Bank for International
Settlements.
Coval, J., J. Jurek, and E. Stafford. 2009. The economics of structured finance. Journal of Economic
Perspectives 23 (1): 3–25.
Demski, J. 2004. Endogenous expectations. The Accounting Review 79 (2): 519–539.
Demski, J., H. Lin, and D. Sappington. 2008. Asset revaluation regulation with multiple information
sources. The Accounting Review 83 (4): 869–891.
Demski, J., H. Lin, and D. Sappington. 2009. Asset revaluation regulations. Contemporary Accounting
Research 26 (3): 843–865.
Dodd, R. 2007. Subprime: Tentacles of a crisis. Finance and Development 44 (4): 15–19.
Donnelly, C., and P. Embrechts. 2010. The devil is in the tails: Actuarial mathematics and the subprime
mortgage crisis. ASTIN Bulletin 40 (1): 1–33.
Evans-Pritchard, A. 2008. Fed’s rescue halted a derivatives Chernobyl. Telegraph.uk.co (March 23).
Available at: http://www.telegraph.co.uk/finance/newsbysector/banksandfinance/2786816/Fedsrescue-halted-a-derivatives-Chernobyl.html
Farhi, M., and M. Cintra. 2009. The financial crisis and the global shadow banking system. Revue de la
régulation 5 (1st Semester). Available at: http://regulation.revues.org/index7473.html
Fiechter, P., and C. Meyer. 2010. Big Bath Accounting Using Fair Value Measurement Discretion during
the Financial Crisis. Working paper, University of Zurich.
Fellingham, J. 2010. Accounting Structure. Unpublished Monograph. Available at: http://fisher.osu.edu/
;fellingham_1/523/index.html
Accounting Horizons
September 2011
Structured Finance and Mark-to-Model Accounting: A Few Simple Illustrations
575
Financial Accounting Standards Board (FASB). 1993. Accounting for Certain Investments in Debt and
Equity Securities. Statement of Financial Accounting Standards No. 115. Financial Accounting
Series. Norwalk, CT: FASB.
Financial Accounting Standards Board (FASB). 1998. Accounting for Derivative Instruments and Hedging
Activities. Statement of Financial Accounting Standards No. 133. Financial Accounting Series.
Norwalk, CT: FASB.
Financial Accounting Standards Board (FASB). 2007. Fair Value Measurements. Statement of Financial
Accounting Standards No. 157. Financial Accounting Series. Norwalk, CT: FASB.
Financial Accounting Standards Board (FASB). 2008a. In Determining the Fair Value of a Financial Asset
When the Market for That Asset Is Not Active. FASB Staff Position No. 157-3. Financial Accounting
Series. Norwalk, CT: FASB.
Financial Accounting Standards Board (FASB). 2008b. In Disclosures about Credit Derivatives and
Certain Guarantees. An Amendment of FASB Statement No. 133 and FASB Interpretation No. 45;
and Clarification of the Effective Date of FASB Statement No. 161. FASB Staff Position No. 133-1
and FIN 45-4. Financial Accounting Series. Norwalk, CT: FASB.
Financial Accounting Standards Board (FASB). 2009a. In Determining Fair Value When the Volume and
Level of Activity for the Asset or Liability Have Significantly Decreased and Identifying Transactions
That Are Not Orderly. FASB Staff Position No. 157-4. Financial Accounting Series. Norwalk, CT:
FASB.
Financial Accounting Standards Board (FASB). 2009b. In Recognition and Presentation of Other-thanTemporary Impairments. FASB Staff Position No. 115-2. Financial Accounting Series. Norwalk, CT:
FASB.
Goh, B., J. Ng, and K. Yong. 2009. Market Pricing of Banks’ Fair Value Assets Reported under SFAS 157
during the 2008 Economic Crisis. Working paper, Singapore Management University.
Goldstein, D., and K. Hall. 2008. Private sector loans, not Fannie and Freddie, triggered crisis. Mcclatchy
Newspapers (October 12). Available at: http://www.mcclatchydc.com/2008/10/12/53802/privatesector-loans-not-fannie.html
Gorton, G., and G. Pennacchi. 1990. Financial intermediaries and liquidity creation. Journal of Finance 45
(March): 49–71.
Guiso, L., P. Sapienza, and L. Zingales. 2009. Moral and social constraints to strategic default on
mortgages. Working Paper No. 15145. Washington, D.C.: NBER.
Heaton, J., D. Lucas, and R. McDonald. 2009. Is mark-to-market accounting destabilizing? Analysis and
implications for policy. Journal of Monetary Economics 57 (1): 64–75.
Huizinga, H., and L. Laeven. 2009. Accounting Discretion of Banks during a Financial Crisis. Working
paper, CEPR.
International Monetary Fund (IMF). 2008. Containing Systemic Risks and Restoring Financial Soundness.
Global Financial Stability Report. World Economic and Financial Surveys (April). Washington,
D.C.: IMF Publications.
Karabell, Z. 2008. Bad accounting rules help sink AIG. Wall Street Journal (September 18): A25.
Kassenaar, L., and C. Harper. 2008. Goldman Sachs paydays suffer on lost leverage with Fed scrutiny.
Bloomberg.com Available at: http://www.bloomberg.com/apps/news?pid=20601109&sid=aA_
lSAQLywYs
Keys, B., T. Mukherjee, A. Seru, and V. Vig. 2010. Did securitization lead to lax screening? Evidence from
subprime loans 2001–2006. Quarterly Journal of Economics 125 (1): 307–362.
Kolev, K. 2008. Do Investors Perceive Marking-to-Model as Marking-to-Myth? Early Evidence from FAS
157 Disclosure. Working paper, Yale University.
Labaton, S. 2008. Agency’s ‘04 rule lets banks pile up new debt. The New York Times Times (October 2):
A1.
Laux, C., and C. Leuz. 2009. The crisis of fair-value accounting: Making sense of the recent debate.
Accounting, Organizations and Society 34: 826–834.
Laux, C., and C. Leuz. 2010. Did fair-value accounting contribute to the financial crisis? Journal of
Economic Perspectives 24 (1): 93–118.
Accounting Horizons
September 2011
576
Meder, Schwartz, Spires, and Young
Leonnig, C. 2008. How HUD mortgage policy fed the crisis. The Washington Post (June 10).
Lev, B., and N. Zhou. 2009. Unintended Consequence: Fair Value Accounting Informs on Liquidity Risk.
Working paper, New York University.
Lewis, M. 2009. The man who crashed the world. Vanity Fair (August).
Löffler, G. 2004. An anatomy of rating through the cycle. Journal of Business and Finance 28 (3): 695–720.
Mason, D. 2004. From Buildings and Loans to Bail Outs: A History of the American Savings and Loan
Industry, 1831–1995. Cambridge, UK: Cambridge University Press.
Mengle, D. 2007. Credit derivatives: An overview. Federal Reserve Bank of Atlanta Economic Review
(Fourth Quarter): 1–24.
Mitchell, J. 2004. Financial Intermediation Theory and the Sources of Value in Structured Finance
Markets. Working paper, National Bank of Belgium.
Morgenson, G., and L. Story. 2009. Banks bundled bad debt, bet against it and won. The New York Times
(December 23): A1.
Negus, J., and C. Boyles. 2009. Structured product valuation methodologies in illiquid and volatile markets.
Bank Accounting and Finance 23 (1): 25–31.
Patterson, S. 2010. The Quants: How a New Breed of Math Whizzes Conquered Wall Street and Almost
Destroyed It. New York, NY: Crown Business.
Peterson, C. 2007. Predatory structured finance. Cardozo Law Review 28 (5): 2185–2284.
Plantin, G., H. Sapra, and H. Shin. 2008. Marking-to-market: Panacea or Pandora’s box? Journal of
Accounting Research 46: 435–460.
Pozen, R. 2009. Is it fair to blame fair value accounting for the financial crisis? Harvard Business Review
(November): 85–92.
Ranieri, L. 1996. The origins of securitization, sources of its growth and it future potential. In A Primer on
Securitization, edited by L. Kendall and M. Fishman. Cambridge, MA: MIT Press.
Ryan, S. 2008a. Accounting in and for the subprime crisis. The Accounting Review 83: 1605–1638.
Ryan, S. 2008b. Fair value accounting: Understanding the issues raised by the credit crunch. Washington,
D.C.: Council of Institutional Investors.
Salmon, F. 2009. Recipe for disaster: The formula that killed Wall Street. Wired Magazine (February 23).
Securities and Exchange Commission (SEC). 2008. Report and Recommendations Pursuant to Section 133
of the Emergency Economic Stabilization Act of 2008: Study on Mark-to-Market Accounting.
Washington, D.C.: Government Printing Office.
Sivesend, C. 1979. Mortgage-backed securities: The revolution in real estate finance. Federal Reserve Bank
of New York Quarterly Review (Autumn): 1–10.
Song, C., W. Thomas, and H. Yi. 2010. Value relevance of FAS 157 fair value hierarchy information and
the impact of corporate governance mechanisms. The Accounting Review 85 (4): 1375–1410.
Sorkin, A. R. 2009. Too Big to Fail: The Inside Story of How Wall Street and Washington Fought to Save
the Financial System—and Themselves. New York, NY: Viking.
Taleb, N. 2001. Fooled by Randomness: The Hidden Role of Chance in the Markets and in Life. New York,
NY: Texere.
Tett, G. 2009. Fool’s Gold. New York, NY: Free Press.
Tirole, J. 2009. Illiquidity and All Its Friends. Working paper, Toulouse School of Economics.
Westbury, B. 2008. How to start healing now. Wall Street Journal (October 1): A25.
Zuckerman, G. 2009. The Greatest Trade Ever. New York, NY: Broadway Books.
Accounting Horizons
September 2011