Description
Taking as its starting point Alan Greenspan’s ‘shocked disbelief’ in the failure of institutional
self interest to prevent the credit crisis, this paper sets out to explore two related
questions. How was self interest constructed in financial markets? And how might we
account for its failure? Conceptually the paper draws upon Callon’s (1998) analysis of
‘agent–networks’, the importance this gives to the agency of non-humans, and his complementary
notions of ‘framing’/‘disentanglement’ and ‘overflowing’ as these allow and subvert
the calculation of self interest. Empirically, the paper then presents a sketch of
these processes in the rise and then fall of the market for collateralised debt obligations
(CDOs) that was central to the credit crisis. The final substantive section of the paper
reflects on the role and ‘hyperreal’ interaction of accounting and models as ‘mediators’
in these processes.
Accounting for self interest in the credit crisis
John Roberts
*
, Megan Jones
University of Sydney, Discipline of Accounting, Faculty of Economics and Business Building H69, Cnr. Codrington and Rose Streets, Darlington, NSW 2006, Australia
a r t i c l e i n f o a b s t r a c t
Taking as its starting point Alan Greenspan’s ‘shocked disbelief’ in the failure of institu-
tional self interest to prevent the credit crisis, this paper sets out to explore two related
questions. How was self interest constructed in ?nancial markets? And how might we
account for its failure? Conceptually the paper draws upon Callon’s (1998) analysis of
‘agent–networks’, the importance this gives to the agency of non-humans, and his comple-
mentary notions of ‘framing’/‘disentanglement’ and ‘over?owing’ as these allow and sub-
vert the calculation of self interest. Empirically, the paper then presents a sketch of
these processes in the rise and then fall of the market for collateralised debt obligations
(CDOs) that was central to the credit crisis. The ?nal substantive section of the paper
re?ects on the role and ‘hyperreal’ interaction of accounting and models as ‘mediators’
in these processes.
Ó 2009 Elsevier Ltd. All rights reserved.
Introduction
‘I made a mistake in presuming that the self interest of
organizations, speci?cally banks and others, were such
that they were best capable of protecting their own share-
holders and their equity in the ?rms. So the problem here is
something which looked to be a very solid edi?ce, and
indeed, a critical pillar to market competition and free
markets, did break down. And I think that, as I said did
shock me. I still do not understand fully why it happened
and, obviously, to the extent that I ?gure out where it hap-
pened and why, I will change my views. I found a ?aw in
the model that I perceived is the critical functioning struc-
ture that de?nes how the world works, so to speak’. Alan
Greenspan, October 2008.
Alan Greenspan’s ‘shocked disbelief’ is an appropriately
cautious starting point for any attempt to make sense of
the credit crisis. His belief in the effectiveness of self inter-
est as a pillar of market competition, having served him
well for the last 40 years, had broken down in the face of
evidence from the current credit crisis. As he observed la-
ter in questioning by the House Oversight Committee, we
all need an ideology or conceptual framework in order to
act, and it is painful, disorienting and disabling when it
breaks down.
Our own interest in the crisis arose from a related start-
ing point. At the beginning of 2008 it was clear that some-
thing unexpected and unanticipated was happening in
?nancial markets and that knowledge had failed market
participants in important ways. As Latour (2005) observes,
the assembly or collapse of networks of relations repre-
sents an ideal moment for research, for then associations
that are otherwise invisible and simply taken for granted
come into clear view. We wanted to try to understand
what was happening, and set out to explore the rise and
fall of the market for collateralised debt obligations (CDOs)
that seemed to be a central player in the crisis. We began
our research with interviews with market participants,
but as the crisis developed these became increasingly dif?-
cult to organise. However, these dif?culties were more
than compensated for by the host of analyses that began
to emerge based on the research of bodies such as the
SEC, the Counterparty Risk Management Policy Group
(CRMPG), the IMF and Financial Stability Forum – groups
that had much better access and research resources than
us.
The credit crisis will no doubt absorb the energies
of researchers for many years to come. In what follows,
0361-3682/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2009.03.004
* Corresponding author. Tel.: +61 2 9351 6638.
E-mail address: [email protected] (J. Roberts).
Accounting, Organizations and Society 34 (2009) 856–867
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
drawing upon the above reports and other analyses of the
crisis, we want to offer a sketch of some of the key ele-
ments that contributed to the CDO market’s rise and fall
in the hope of casting some light on the catastrophic failure
of understanding that the crisis seems to embody. In par-
ticular, the paper explores the role played by models and
accounting as ubiquitous ‘mediators’ of almost all market
relationships, arguing that they fed both the illusion of
rationality (greed) that fuelled market growth, and com-
pounded fear and panic as the market fell. As a framework
for this more process-oriented view of the crisis, we begin
with a discussion of some elements of Callon’s seminal
contribution to social studies of ?nance in ‘The Laws of
the Markets’ (1998) (see also Callon & Muniesa, 2005;
MacKenzie, Muniesa, & Siu, 2007). When viewed through
the lens of the current credit crisis this analysis seems par-
ticularly prescient and points to the misconceptions and
lacunae that possibly produced Greenspan’s and others’
‘shocked disbelief’.
The construction of self interest
Callon begins his analysis of markets by noting North’s
observation that a peculiar omission in economics is its
failure to discuss markets. Callon follows Guesnerie’s de?-
nition of a market as a ‘coordination device’, in which
agents pursue their own interests through economic calcu-
lations. The generally divergent interests of agents lead
them to engage in market transactions that resolve con?ict
by de?ning a price. For Callon such economic conceptions
of market coordination ‘have the common feature of pro-
viding autonomous – over autonomous – and isolated –
over isolated – agents with the social relations which, by
opening them up to their environment, enable them to
co-ordinate their actions with those of other agents’
(1998, p. 7). His alternative is to reverse the economic
assumption of atomism by taking agents’ dependence on
their environment as his starting point.
This reversal is very important for it is a decisive re-con-
ceptualisation of a founding assumption of economics and
?nance – that of a self-seeking and opportunistic (calculat-
ing) individual as the basic unit of analysis. In place of this
traditional dualism of ‘agent’ and market relations or ‘net-
works’, Callon insists that the unit of analysis must be the
‘agent–network’; ‘agent and network are, in a sense, two
sides of the same coin’. A network from this perspective
is not something that merely connects already existing
entities, but rather itself produces the calculating agents;
‘the agents, their dimensions, and what they are and do,
all depend upon the morphology of the relations in which
they are involved’ (1998, p. 9). The image of the market
here is not that of an encompassing context for agency that
contains its own invisible hand magically transforming the
pursuit of self interest into a public good, but rather of the
market only as myriad transactions being repeatedly en-
acted; ‘once the transaction has been concluded the agents
are quits; they extract themselves from anonymity only
momentarily, slipping back into it immediately after-
wards’. Here we want to follow two elements of Callon’s
subsequent analysis of markets – his analysis of processes
of ‘framing’/‘disentanglement’ and ‘over?owing’ – as these
each build upon his insistence of the indivisibility of
‘agent–networks’.
From an outsider’s perspective one of the most remark-
able features of the world of ?nance is the sheer energy
that individuals and institutions put into the business of
making money. Rather than see such self interest as natu-
ral and reliably predictable, Callon insists that we have to
work very hard to produce the self, and the self interest,
that Greenspan took as a given. Critically it depends upon
‘calculation’, which Callon argues should be understood
neither as a cognitive characteristic nor a cultural phenom-
ena, but rather as dependent upon the complex temporal
‘framing’ of the relationships in which a person is engaged.
He draws the concept of ‘framing’ from Goffman’s micro-
analyses of the ‘staging’ of interaction, and argues that in
relation to markets it should be understood as a process
of ‘disentanglement’. As he puts it:
Framing is an operation used to de?ne agents (an indi-
vidual person or group of persons) who are clearly dis-
tinct and dissociated fromone another. It also allows for
the de?nition of objects, goods and merchandise which
are perfectly identi?able, and can be separated not only
from other goods, but also from the actors involved, for
example in their conception, production, circulation
and use. It is owing to this framing that the market
can exist and that distinct agents and distinct goods
can be brought into play (1998, p. 17).
In relation to the CDO market in following Callon we
can think of the energetic pursuit of self interest then not
as natural, but rather as the product of myriad processes
and devices that allowed the self, and self interestedness,
to be calculated, pursued and realised. Only this ‘framing’
of market relations allows the self and self interest to be
‘disentangled’ within the relationships through which it
is realised. The work of framing/disentanglement was mas-
sive. The development of the CDO in its multiple forms,
and the new opportunities and relationships that these
products opened up and made possible were themselves
an essential part of this framing/disentanglement. So too
were legal and employment contracts, bonus schemes,
institutional identities and roles, market indices and
countless forms of market data, etc. (Callon, Millo, & Mu-
niesa, 2007). But in analysing such active processes of
framing, Callon gives a key role to the agency of what he
calls ‘calculating tools’ and in particular to accounting; ‘full
signi?cance has to be restored to that humble, disclaimed
and misunderstood practice; accounting and the tools it
elaborates’ (1998, p. 23).
Latour’s (2005) later distinction between what he calls
‘intermediaries’ and ‘mediators’ is useful in clarifying the
importance Callon gives to ‘calculating tools’. Whilst ‘inter-
mediaries’ act merely as dependable relays in a network,
‘mediators transform, translate, distort, and modify the
elements they are supposed to carry’ (2005, p. 39). The self
image, or at least public image, of accounting is that of an
intermediary – it does no more than render ‘what is’ ‘trans-
parent’. But for Callon (1998) and Miller (1998; Miller &
O’Leary, 2007) whose work he is following, accounting
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 857
should be seen as a mediator. As Callon puts it, calculating
tools ‘do not merely record a reality independent of them-
selves; they contribute powerfully to shaping, simply by
measuring it, the reality they measure’ (1998, p. 23). Part
of this agency involves the ways that, as accounting tools
are constantly recon?gured, they ‘authorize decisions that
are more and more calculated or (to use the commonly ac-
cepted word) more and more rational’. In this respect the
CDO market was from its inception absolutely dependent
on the use of mathematical modelling techniques (Mac-
Kenzie, 2006; Morgan & Morrison, 1999), and its evolution
involved a dizzying ‘bricolage’ of ever more complex prod-
uct forms designed to entice further pro?table growth
(Engelen, Ertuk, Froud, Leaver, & Williams, 2008). Along-
side this product innovation, accounting standards and
practices were themselves forced to evolve, but this itself
provided fuel for the development of ‘new calculative
strategies’. Accounting is, of course, also vital as the ?nal
measure of the achievements of self interest:
Money comes in last in a process of quanti?cation and
production of ?gures, measurement and correlations
of all kinds. It is the ?nal piece, the keystone in a metro-
logical system that is already in place and of which it
merely guarantees the unity and coherence (Callon,
1998, p. 22).
Accounting only captures the pro?ts (losses) momen-
tarily on their way to others. In investment banks this typ-
ically involved the equal sharing of pro?ts between
employee annual bonuses and shareholders (Augar, 2006).
One moment in Callon’s analysis of markets as coordi-
nating devices suggests then the necessity, if one is to
understand the current crisis, of exploring the intense la-
bour of framing/disentanglement that took place in ?nan-
cial markets in the construction of individual and
institutional self interest, and the ubiquitous role of non-
humans agents in this – notably the models and account-
ing that mediated almost all relationships. However, the
full signi?cance of his analysis of markets as ‘agent–net-
works’ only emerges with his exploration of the corollary
of framing in the parallel process of what he calls ‘over-
?owing’. Callon argues that ‘without this framing the states
of the world cannot be described and listed, and conse-
quently, the effects of the different conceivable actions
cannot be anticipated’ (1998, p. 17). There is an echo here
of the current crisis and the paralysis of action that arises
when calculation becomes unreliable or even impossible.
Callon argues that for economics framing is regarded as
the norm, and that over?ows which create ‘externalities’
are from this perspective exceptional and ‘accidental leaks’
to be corrected. However, by insisting that agents are
inseparable from, and indeed constituted only in and
through the networks of relations in which they are
embedded, Callon reverses this assumption. From a con-
structivist perspective, he argues, over?ows should be re-
garded as the norm and framing as itself both ‘expensive
and always imperfect’ (1998, p. 252). This is because fram-
ing is always incomplete and must be so if value is to be
created; ‘it is because an actor’s output gets necessarily be-
yond her entire control so as to generate pro?ts, that the
actor is unable to avoid externalities’ (1998, p. 23). Impor-
tantly, Callon suggests that over?owing is multidirectional
but nevertheless travels back along the very paths (net-
works) that are initially created in the pursuit of individual
and institutional self interest. However, such over?owing
is initially invisible and must itself be subjected to mea-
surement in order to be recognised and then ideally re-
framed.
Taken together Callon’s notions of framing/disentangle-
ment and over?owing offer a very powerful lens through
which to re-think our assumptions about the nature of self
interest in markets. Typically, as evidenced in Greenspan’s
comments, interests are taken to refer only to the self; as if
interests were somehow internal to the self (or institution)
and that it is the pursuit/defence of these that de?nes the
nature of economic rationality. For Callon, however, this is
to misunderstand and ignore the tension that is implicit in
self interest. The etymology of interests as ‘inter-esse’ re-
veals this tension for interests always lie between selves
rather than within the individual (institution). There is in this
sense no self, or possibility of a self, divorced from the rela-
tionships that are the very ground of its own agency. From
this perspective ?nancial markets are a ?nely balanced
high-wire act in which network interdependencies are
endlessly elaborated and intensi?ed through the ever more
calculated pursuit of individual and institutional gain. As
Callon puts it:
How can one perform framing when one has to be
attentive to all this over?owing? How is it possible to
become homo clausus when survival requires one to
be homo apertus? This question is at the heart of the
stock market and the speculative behaviour that it
spawns. Nowhere is the tension between framing and
over?owing so intense and so dif?cult to control
(1998, p. 25).
So in this sense crises and volatility, the jumps from
boom to bust, are to be expected and result from no more
than a small shift in the precarious balance between the la-
bour of framing and consequent over?owing that is ever
present in markets.
In the remainder of the paper we take these metaphors
of framing/disentanglement and over?ows and use them
as the basis for a sketch of some of the key elements in-
volved in the construction of self interest in the rise and
fall of the market for CDOs. In the next section, we explore
the growth of the CDO market; the evolution of the prod-
ucts, the core ‘framing/disentanglement’ processes of secu-
ritisation and tranching that were at the heart of these
innovations, and the work done by models and accounting
in facilitating this growth. A key element in the story is the
greatly extended network of interests that was created
through these innovative forms of securitisation. While
markets were apparently working well, these seemed only
to offer participants ever expanding opportunities to
‘trade’ risk widely and pro?tably. However, we will then
seek to describe how, with the trigger of subprime mort-
gage defaults and a housing market downturn, these net-
works became channels for losses and panic. In
particular, we want to trace the peculiar ‘over?owing’ of
858 J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867
risk that has taken place from a seemingly safely framed
and pro?table ‘credit risk’, into ‘market risk’ in which as-
sets prices fell in unexpected ways, and ‘counterparty risk’
where fear of the potential exposure of others to losses
quite literally brought credit markets to a standstill, and
thereby created the current liquidity crisis (CRMPG,
2008). The ?nal substantive section of the paper draws to-
gether our re?ections on the role of accounting and its
interaction with models in the crisis.
Framing/disentanglement in the CDO markets
The market for CDOs emerged in the early 1980s but
only became a major asset class in the late 1990s, after
which there was an explosive growth in the volume and
value of transactions, particularly from 2003 to 2007 (SIF-
MA, 2007). Between 1995 and 2007 the total value of out-
standing asset-backed securities grew from $500bn to
$2500bn. Within this growth home equity loans grew in
importance, accounting for some $600bn in 2007 (Felsen-
heimer & Gisdakis, 2008). In 2001 the value of new sub-
prime mortgages originated was $160bn, a ?gure that
increased to $600bn in 2006 (Blackburn, 2008).
Securitisation
What is common to this family of products are pro-
cesses of ‘securitisation’ and ‘tranching’. Essentially securi-
tisation involves packaging up illiquid debt into securities
that can then be sold on to others, thereby creating both
liquidity and the dissipation of the risk attached to the
debt. Securitisation has a long history but in relation to
mortgages it involves a shift from an ‘originate to hold’
model of traditional mortgage lending to an ‘originate to
distribute’ model. The shift is important for, when com-
pared to the traditional model in which borrower and len-
der are bound together in a dyadic long term relationship,
the new model involves an extended chain of new institu-
tional relationships and inter-dependencies. With mort-
gage-backed securities the borrower and mortgage
broker/lending bank still begin the process but individual
mortgages are then bundled together and sold on to inves-
tors. The key rationale for securitisation is drawn from
portfolio theory and, in particular, the notion that ‘diversi-
?ed’ pools of assets – say different grades of mortgage,
from different regions or countries, or indeed different
combinations of assets – create less ‘idiosyncratic risk’
(Markowitz, 1959). The ‘disentanglement’ achieved by sec-
uritisation is therefore of the individual borrower from
individual lender.
Tranching
Whilst in a securitisation of assets cash ?ows are simply
passed on to investors, with CDOs asset-backed securities
are then subjected to a further process of what is called
‘tranching’. Tranching is a particularly creative form of dis-
entanglement that allows the creation of different invest-
ment products with very different risk characteristics
from an underlying portfolio of assets. It also involves a
seemingly magic process through which low quality debt
can, in part, be transformed into higher quality, investment
grade assets that, by virtue of this ‘credit enhancement’,
can then be sold to/taken on by a much larger population
of investors. The simple example given of such a process
suggests a threefold tranching of the assets into an ‘equity’
or ‘?rst loss’ tranche which references say 0–3% of the port-
folio, a ‘mezzanine’ tranche referencing say a further 12% of
the portfolio, and a ‘senior’ tranche which references the
remaining 85% of the portfolio of assets. Tranching effects
a skewed distribution of the risks and returns associated
with the underlying assets. What is termed a ‘waterfall
principle’ (an ominous anticipation of future over?ows)
dictates that cash ?ows from the pool of assets go ?rst to
the senior tranche, then mezzanine, then equity. The allo-
cation of risks and consequently higher returns, however,
follows a ‘reverse waterfall’ principle in which any losses
arising, say from defaults on subprime mortgages, are ?rst
absorbed by the equity tranche until it is completely
eroded, then the mezzanine and only then effect the senior
tranche. As a result of tranching the most senior tranches
can be rated as investment grade securities offering higher
returns than similarly rated corporate bonds, by virtue of a
structure in which others absorb the risks of initial defaults
in their entirety for a yet higher return. Often the equity
piece was retained by the originator as a mark of con?-
dence, or sold on to hedge funds which had an appetite
for such risk and return. However, the relative dif?culty
of passing on these riskier tranches was also the inspira-
tion for more complex structured ?nance CDOs that essen-
tially re-securitised and re-tranched these lower tranches
thereby creating yet more senior level tranches.
The effectiveness of such tranching procedures depends
critically upon the nature of the assets involved and, in
particular, the ‘correlation’ of risks between different as-
sets in the portfolio. Whilst securitisation of assets ideally
results in a diversi?ed portfolio that reduces ‘idiosyncratic
risk’, there remains the ‘systemic’ risk created by the pos-
sibility that macro events – for example, a downturn in
the housing market – might cause the default risk in the
portfolio to be highly correlated such that one default is
tied to multiple defaults. As a result the success of tran-
ching depends critically upon the calculation of a whole
variety of variables, including such ‘correlation’ risk. This
was one of the critical spaces in which the expertise of
‘quants’ staff and the models they employed became vital
to the operation of credit markets. The other spaces where
models became all important were in their parallel use by
credit rating agencies and in ‘risk management’ processes
in the banks, investment banks and hedge funds, by means
of which institutional risks were appraised and monitored
(Millo & MacKenzie, 2008; Power, 2007).
For the modelling of a mortgage backed CDO, for exam-
ple, data inputs would be divided into pre-payment vari-
ables, such as interest rates, asset prices, macroeconomic
and housing data, and default variables, such as the loan
to value ratios, estimates of default probabilities and likely
recovery rates (Fender & Kiff, 2004). Once assembled, such
data would then be subjected to both random and selected
scenario simulations to gauge the possible correlation ef-
fects the loans might display under different conditions.
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 859
On this basis the capital structure of the CDO could then be
calculated – the size and ‘thickness’ of the different tran-
ches – and cash ?ows discounted and allocated to the dif-
ferent tranches. This, in turn, allowed the originator to
assess the likelihood of excess spread in the loan pool each
month over the term of the investment, and to ensure that
payments of interest and principal covered proposed pay-
ments to investors. The resultant structure could then be
‘stress tested’ in a variety of ways against current market
and historical data to verify the modelling process. Since
the attraction and marketability of such products de-
pended upon senior tranches receiving an investment
grade (AAA) rating then, in practice, the ?nal structure of
the CDO was the product of a process of negotiation be-
tween the originator and the credit rating agency. The
involvement of the credit rating agencies was critical since
their rating served as an authoritative, apparently indepen-
dent, mark of assurance in relation to what is an invariably
complex and opaque set of calculations of the risk/return
pro?le of the different tranches (Coffee, 2008; SEC, 2008a).
The simplest version of a CDO was called a ‘cash ?ow’ or
‘balance sheet’ CDO; so called because it was managed off
balance sheet through a bankruptcy remote Special Pur-
pose Vehicle (SPV). In this simple version, the assets are
bought and packaged together, and sold to the SPV, which
after tranching and rating, then passes on the risk but also
the interest and principal to the ultimate investors. Press
attention has focused on CDOs backed by subprime mort-
gages but in reality a whole variety of different kinds of
loans and assets, including mortgages (Prime, Alt A and
Sub-prime) as well as student loans, credit card and car
loans, as well as combinations of these and other assets,
were used as the basis for CDO products (Felsenheimer &
Gisdakis, 2008).
Enrolling interests to the CDO
If we stay for the moment with CDOs backed, or par-
tially backed, by subprime mortgages then we can think
about the growth of the market through Callon’s (1986)
language of ‘translation’; the attractiveness of the CDO to
all the interests that it enrolled. For the sub-prime bor-
rower it offered access to the ‘American Dream’ of home
ownership despite their no income, no job status. For lend-
ing banks securitisation allowed them to pass on the risks
they had formerly held to maturity, providing them with
relief from regulatory capital requirements and thus allow-
ing further lending. For investment banks the CDO opened
up a new and highly pro?table stream of work; as deal
makers there were origination fees to be earned as well
as margin on the tranched assets as fuel for this year’s bo-
nus pool. By 2006 30% of all investment bank earnings
were from structured ?nance activities (Pleven & Craig,
2008). Similarly, for hedge funds charging high fees and
offering high returns, the mezzanine and equity tranches
were grasped as an ideal product. This then fed back into
pro?table prime brokerage business for the investment
banks in which they provided clearing services and ?nance
for the leveraged strategies of the hedge funds. For the
credit rating agencies the CDO opened up a whole new cat-
egory of work in addition to their traditional low margin
bond rating work (Coffee, 2008). Likewise for the monoline
insurers who insured the senior tranches as part of the
credit enhancement process. At the end of the chain were
a whole range of individual and institutional investors
who, in a time when yields were low, were able to invest
through the CDO in products that were AAA rated but of-
fered a substantial premium to similarly rated corporate
bonds. Finally, we should mention market regulators.
Greenspan and others pronounced themselves pleased
with these new structured ?nance products which, they
argued, had contributed both liquidity and stability to
?nancial markets by virtue of the dissipation of risk that
they allowed (Greenspan, 2005; Joint Forum Credit Risk
Transfer, 2004).
Synthetic CDOs
Perhaps precisely because of the attraction and success
of the cash ?ow CDO, in the early 2000s a new variety of
product emerged in the form of so called ‘synthetic’ CDOs.
With this version of the product there is no passing on of
the principal, interest and ownership of the assets to the
ultimate investor (this was not always legally possible),
rather these are retained by the originator and credit de-
fault swaps that mimic the risk pro?le of the assets are en-
tered into with the SPV which in turn sells the risk on to
investors. Credit default swaps (CDS) are typically ex-
plained using the analogy of an insurance policy, with
the originator buying and investors effectively selling pro-
tection on the risk of potential losses on the underlying as-
sets. Unlike insurance companies, however, the market is
unregulated. Synthetic CDOs made use of CDSs in two
ways. In funded synthetics the CDS sits between the origi-
nator and SPV but investors pay in the full amount of the
tranche, which is then used as collateral or invested in high
quality securities. There is a further distinction here be-
tween managed and unmanaged funds – essentially
whether the asset pool remains static or is actively man-
aged through sales and purchases of assets throughout
the life of the product. By contrast, in unfunded synthetic
CDOs there is no initial investment; investors receive a
periodic spread payment on the CDSs related to their tran-
che but are required to cover the cost of any defaults if they
arise.
The key element of both funded and unfunded synthetic
CDOs is that the risk is disentangled from the underlying
assets and allowed to travel. As Jones and Peat explain:
‘synthetic CDOs do not involve any actual sale of assets
by the originator – only the underlying credit risk of the
reference assets is transferred to the counterparty. The
originator remains the legal and bene?cial owner of those
assets’ (2008, p. 216). One important feature of the CDS
market is that it is an ‘over-the-counter’ (OTC) market
and therefore the distribution of risk that contracts allow
remains opaque; a factor that was the basis of more recent
counterparty risk. The CDS had itself apparently proved
immensely popular as a credit instrument, both because
of its risk disentanglement and dissipation capabilities as
well as its unfunded nature. This fuelled an explosive
growth in the market with some $45trn of contracts out-
standing as of 2007 (Morris, 2008).
860 J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867
The disentanglement of risk from the underlying assets
also changes the nature of the risk from ‘default’ risk to
‘market’ risk. As Felsenheimer and Gisdakis explain:
The great achievement of credit markets is that they
have enabled investors to buy and sell credit risky
assets, thus transforming an event risk (i.e. the risk of
a default event) into a price risk (i.e. the risk that the
value of an asset declines over time). The price
expresses the probability that such an event occurs
and the anticipated loss that accompanies the event.
This transformation process was accompanied by the
development of derivative structures in order to
improve the tradability of the underlying risks (2008,
p. 154).
The emergence of synthetic CDOs was also driven by a
very different set of motives. Whilst balance sheet CDOs
were used to of?oad certain assets for the purposes of risk
reduction or re?nancing, the development of the synthetic
CDO was motivated primarily by the trading and pro?t
(arbitrage) opportunities offered by these instruments
(Das, 2006). Demand for the CDO was then driven as much
by investors as by the supply of assets whose risk de-
manded securitisation. This then fuelled the further
growth of the market as well as the demand for assets that
could be processed in this way; subprime mortgages were
particularly attractive because of the premium charged to
the borrower over the life of the loan. As the Joint Forum
on Credit Risk Transfer summarised it: ‘the credit risk
transfer market is characterised by signi?cant product
innovation, an increasing number of market participants,
growth in overall transaction volumes, and perceived con-
tinued pro?t opportunities for ?nancial intermediaries’
(2004, p. 1).
The development of the synthetic CDO led to, and was
facilitated by, the creation of new American and European
‘indices’ for CDSs, CDX and iTraxx, including the ABX.HE,
which was also known as the sub-prime index. These then
allowed investors to trade in the average credit spread of
an underlying portfolio and led to the development of
CDO
2
and CDO
3
– CDOs of CDOs of CDOs. Other yet more
innovative and exotic versions of the CDO included Lever-
aged Super Senior Tranches, Constant Proportion Debt
Obligations, and Structured Investment Vehicles (SIVs),
which involved the combination of high quality assets
and high levels of leverage to offer extraordinary but
apparently low risk returns. SIVs were important because
they used short term asset-backed commercial paper to
fund longer term debts, thus creating direct links between
underlying mortgage-backed securities and short term
money markets. These off balance sheet vehicles were typ-
ically structured to include various triggers designed to
limit leveraged losses, as well as liquidity facilities with
their sponsoring banks.
Accounting
Up until the middle of 2007 accounting’s role in the
growth of the CDO markets could be seen as that of a
complex but innocent ‘intermediary’ that merely recogni-
sed the pro?tability of these new innovations. In response
to the rapid growth of the structured credit universe both
the FASB and IASB had been required to develop new
standards in relation to the valuation of ?nancial instru-
ments. One critical set of issues concerned the accounting
treatment of off balance sheet entities (OBSEs) and here,
post-Enron, such structures were only allowed if no single
institution held the majority of the risks and rewards or
control rights. For the measurement of on balance sheet
assets a distinction was made between whether assets
are available for sale (AFS) or to be held to maturity
(HTM). Both the IASB and FASB required the measure-
ment of AFS assets at ‘fair value’, which is de?ned by
the FASB as ‘the price that would be received to sell an
asset or paid to transfer a liability in an orderly transac-
tion between market participants at the measurement
date’ (FAS 157, para 5), and by the IASB as ‘the amount
for which an asset could be exchanged between knowl-
edgeable, willing parties in an arms length transaction’
(IAS 16, para 6). Both accounting standards then offer a
threefold hierarchy of bases for the determination of fair
value. Least problematic is the use of quoted prices for
identical assets and liabilities. If these are not available,
the second level of the hierarchy stipulates the use of ‘ob-
servable inputs’, which might include quoted prices for
similar assets, or inputs derived from an observable index.
In the absence of such liquid and active markets, or com-
parable prices, then valuation has to be done on a ‘mark-
to-model’ basis, rather than a ‘mark-to-market’ basis. This
serves to create an immediate tie between accounting
valuations and model assumptions that are clearly depen-
dent upon judgement (see, Ryan, 2008; Whittington,
2008). SEC data suggests that 15% were measured at level
1, 76% at level 2 and 9% at level 3 with relatively small
differences between sectors (SEC, 2008b).
Leverage
In 2005/2006 the spreads, particularly on indices for
synthetic CDOs, became increasingly narrow – a possible
signal that risk was being ‘under-priced’ – but by then such
was the con?dence of the market in the CDO that the re-
sponse was simply to add leverage to the deals (Adrian &
Shin, 2008a). Leverage involves the simplest of calculations
– multiplication. As Felsenheimer and Gisdakis explain:
The investment rationale was very simple; if the risk
premium is low, then risk has to be low. And if the risk
premiumprovides only a quarter of the return, then just
invest four times as much. This strategy was supported
by ?nancial markets, since in a low yield, low spread
environment the cost of capital is also low (2008, p.
156).
The levels of leverage were considerable; in banks it
was around 12 to 1, for investment banks around 30–1,
and for some of the more exotic CDO products up to 60–
1. Importantly the tranching process itself creates ‘embed-
ded’ leverage in the lower tranches so that the CDO
2
and
CDO
3
could include leverage upon embedded leverage.
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 861
Whilst markets were stable such leverage was an obvious
way to multiply pro?ts but it was also laying the seeds of
the subsequent liquidity crisis (Adrian & Shin, 2008b).
In the above we have attempted a brief sketch of some
of the key processes of ‘framing’/‘disentanglement’
through which the CDO product and market was consti-
tuted. The move from an ‘originate to hold’ to an ‘originate
to distribute’ model greatly increased the degree of entan-
glement of ?nancial market participants, creating new and
extended networks of relationships that tied the sub-prime
borrower and lender through to investment banks, insur-
ers, rating agencies, hedge funds, money markets and insti-
tutional and individual investors. The disentanglement of
default risk was achieved, in part, through securitisation
and then tranching; processes that combined then re-di-
vided the pools of assets into different risk/return seg-
ments. Risk was then further disentangled from the
underlying assets with the development of synthetic prod-
ucts using credit default swaps. Common to almost every
link in these now extended networks was the use of mod-
els to construct, rate, risk manage and price the products.
Accounting was indeed the ‘keystone’ to these processes
in its ability to calculate, recognise and thereby expropri-
ate the pro?ts that were being earned at every stage of
the process; pro?ts that were multiplied through the
extensive use of leverage. Up until 2007 these ‘framing’
processes apparently worked well and, as such, could per-
haps be relied upon and taken for granted by participants.
For outsiders to ?nancial markets, beyond the ready access
to cheap credit, these processes were completely invisible.
Over?owing in CDO markets
As is now well known, rising interest rates in late 2006/
early 2007 and the beginnings of a decline in the housing
market as a whole, set the environment for defaults on
mortgages in the USA. Sub-prime borrowers were particu-
larly vulnerable to these changes and by the third-quarter
of 2007 an alarming 42% of subprime adjustable rate and
12% of subprime ?xed rate mortgages had started the pro-
cess of foreclosure (Mortgage Bankers Association, 2008).
2006/2007 vintages of subprime mortgages, in particular,
had higher rates of default than previous years and created
a bleak prospect for 2008 given that some $250bn worth of
these were due to reset in this period (IMF, 2008). These
actual and anticipated levels of default far exceeded the
assumptions of default risk that had been modelled into
CDOs based on subprime mortgages. Further, and critically,
in a declining housing market modelled assumptions of
recovery rates post-default and the degree of correlation
between defaults had also been drastically underestimated
(Ashraft & Schuermann, 2008; Chomsisengphet & Penning-
ton-Cross, 2006; Ryan, 2008; SEC, 2008a). This was the
shock that passed through into ?nancial markets and, sig-
ni?cantly, it was the non-human agents, the models and
accounting, that acted as the messengers. In this context
we can only touch on some of the major shocks of the per-
iod in order to give a ?avour of the nature of the over?ows.
In mid-June 2007, two of Bear Sterns’ mortgage related
hedge funds collapsed. They had been used to house sub-
prime RMBS and CDOs, many of which were structured
using the worst performing 2006 vintage mortgages. Dur-
ing the ?rst four months of 2007 one fund lost 23% of its
value. This was compounded by its borrowings, which at
$6bn were ten times larger that its $600m capital (Econo-
mist, 2007). As a result, the funds were unable to meet calls
from their counterparties for additional margin or
collateral.
In July the rating agencies, who were now in possession
of new default data, started to re-rate CDOs and there fol-
lowed a huge number of downgrades, often by multiple
notches. Moody’s for example, downgraded 252 AAA rated
CDOs sold in 2006/2007, or 22% of the 1155 deals issued in
that period with a combined value of $12bn; the average
magnitude of each downgrade was eight notches. The
meaning for market participants was ambiguous – was
the default prospect so much worse or had the original rat-
ings been faulty, and by implication were the modelling
processes upon which they had relied themselves funda-
mentally ?awed? The volume and severity of these down-
grades pushed market prices even lower, especially for
junior tranche, subprime CDOs. Downgrades on the senior
tranches raised fears that ‘rating sensitive’ investors might
be forced to sell assets.
In August, the world of SIVs began to unravel when
Sachsen LB announced that German savings banks had
provided a Eur 17bn credit facility given the dif?culties it
was facing in re?nancing short term commercial paper
upon which it depended, raising concerns about asset sales
in this $350bn segment of the market. Sachsen LB was
immediately bought to avoid default because of its invest-
ments in US subprime assets. Later in August, London
based SIV Cheyne Finance reported losses and the possible
sale of assets because of similar problems related to short
term money market re?nancing. In October JP Morgan,
Bank of America and Citigroup announced plans for an
$80bn bailout for their SIVs. In November HSBC announced
a bailout of two SIVs of $45bn, whilst in December Citi-
group announced a bailout of $49bn for its six SIVs.
More shocks came when the large investment banks
started reporting large write downs in September 2007.
Between then and early 2008 Merrill Lynch reported
write-downs of $24.5bn, Citigroup of $22bn, UBS $18bn,
HSBC $10bn and Morgan Stanley $9.5bn. The scale of these
losses fed huge falls in their stock market capitalisation,
creating an urgent need either for de-leveraging and credit
rationing or for new capital injections. As seriously, given
the dependence of the synthetic CDO on credit default
swaps, this fed into the spreads in their debt instruments
(Felsenheimer & Gisdakis, 2008, p. 123). These concerns
in relation to CDSs were heightened when in December
and January the ratings agencies began to downgrade their
ratings on the monoline insurers. Their exposure to struc-
tured ?nance was only 30% of their business but hugely
leveraged and losses had begun to eat very rapidly into
their capital. More seriously the downgrades put at risk
the entire $2.4trn of structured ?nance, municipal bonds
and securities that they had guaranteed. The near collapse
of AIG, and the backlash from the collapse of Lehman
Brothers re?ects the dangers of over?ow to the vast CDS
market.
862 J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867
In the light of our above discussion of framing these
over?ows can be understood since, as Callon asserts, ini-
tially they arose from the ways that subprime defaults
fed rapidly through the networks that CDOs had created.
Most obviously, increased defaults often wiped out the
equity and mezzanine tranches and began to eat into what
had been rated as apparently risk remote senior tranches.
Having believed that CDOs allowed them to pass on risks
to others, banks and hedge funds found themselves incur-
ring large and immediate losses that, in turn, created the
need for re-capitalisation. But as described earlier, the syn-
thetic versions of CDOs had also created new ‘market’ and
‘counterparty’ risks that had not been assessed as part of
the rating process. Even without defaults, falling prices
on indices created unrealised mark-to-market losses,
whilst margin calls for increased collateral, or trigger
mechanisms in the highly leveraged CDOs, forced ?re sales,
which only further depressed prices in a potentially self
feeding vicious circle. The reliance of SIVs on short term
commercial paper that could not now be re?nanced cre-
ated an immediate connection to money markets, forcing
up spreads and feeding the growing liquidity crisis. Liquid-
ity facilities offered by the sponsors of such vehicles, as
well as considerations of ‘reputational risk’, brought these
until then invisible liabilities back onto balance sheets fur-
ther squeezing capital, etc. To each of these channels for
over?owing must be added the multiplier effects of lever-
age. Now, however, it was losses rather than pro?ts that
were being multiplied.
A loss of faith in calculating tools
Although such post-hoc tracing of the channels along
which markets over?owed is possibly helpful, it is as
important to observe that, as these over?ows appeared in
reported accounting losses, they were both on a shocking
scale, and for most people completely mysterious. Like
blind men touching different parts of an elephant, or like
a pixelated TV screen, there was no sight of what was gen-
erating these effects. Instead, accounting losses signalled
only that previously reliable frames were no longer pro-
ducing pro?ts but not why. Entity accounting provides part
of the explanation for this since systemic effects were only
visible in the ways these touched an individual institution.
However, it was also the case that some effects were com-
ing from entities that had not even been reported on – the
off balance sheet SPVs and SIVs that were now suddenly
appearing on the balance sheets of their sponsoring banks –
as well as the effects on valuations of the still invisible
but vast set of counterparty risks held in the OTC CDS mar-
ket. That losses could appear unexpectedly in these ways
told market participants that what they had previously ta-
ken as a complete frame had been no such thing. Even
within entity accounting, say in the investment banks,
the reported aggregated data initially did not allow inves-
tors to see suf?cient detail of exposures in relation to par-
ticular activities and asset classes. The variable timing and
‘jumps’ in the values of write downs from quarter to quar-
ter were also shocking, and there was particular investor
concern about level 3, mark-to-model valuations. Investors
feared this could be being used to minimise reported
losses, whilst for entities it was often simply a necessity gi-
ven the collapse of liquid markets (see IMF, 2008, chap. 2).
When all was going well models could be taken as evi-
dence and embodiments of the rationality so prized by ?-
nance, and accounting as no more than a neutral
observer of the fruits of this rationality; both models and
accounting were treated as reliable ‘intermediaries’. With
the crisis, however, both models and accounting came to
be seen as unreliable just as they were most needed; they
were now viewed as possibly distorting ‘mediators’. This
collapse of belief in what had been seen as reliable and
dependable ways of knowing upon which all still de-
pended, was one of the routes through which panic en-
tered the scene and this, in turn, served only to heighten
a different kind of self interest. Freud’s de?nition of panic
seems relevant here: ‘a panic arises if a group becomes dis-
integrated. . .each individual is solicitous only on his own
account, and without any consideration for the rest’ (cited
in Weick, 1993).
Such panic can involve an intensi?cation of calculation
intended to defend institutional and individual self interest
from market turmoil. Unintentionally, it feeds a paranoia
that then tries to run ahead of current facts; not just sub-
prime but the entire market for securitisation has effec-
tively closed; spreads on CDO and CDS indices have
blown out in anticipation of future losses and, perhaps
most consequentially, money markets have frozen as insti-
tutions seek to preserve their own liquidity in the now cer-
tain knowledge that other institutions with whom they
formerly traded face counterparty risks of which they too
are ignorant. Whilst seemingly rational from an individual,
institutional point of view, such extreme self defence is, of
course, self ful?lling (Ghoshal & Moran, 1996; Roberts,
2001). As with the anticipation of future pro?ts, the ra-
tional attempt to protect the institution from future losses
can unintentionally serve to bring these into reality. At the
time of writing, despite governments’ attempts to breathe
life back into inter-bank lending by injecting billions of
dollars into markets, homo economicus still remains in a
state of ‘shocked disbelief’ and is largely closed for
business.
Accounting for self interest
In the above we have followed Callon’s insistence that
self interest should be seen, not as a given and de?ning fea-
ture of the ‘individual’ as it is in economics and ?nance, but
rather as something that is actively constructed. We have
then sought to trace, if only in sketch form, the intense la-
bour of framing/disentanglement through which CDO
products and markets were created and elaborated, only
then to over?ow to produce the current crisis. Within this
account we have also followed Callon’s insistence on the
importance of ‘calculating tools’ as agents or ‘mediators’
within these processes, and, in particular, have sought to
trace some of the effects of mathematical models and
accounting in producing the greed and then fear that have
characterised the rise and fall of the CDO market. Latour’s
distinction between neutral ‘intermediaries’ and distorting
‘mediators’ possibly seems obscure, but is re?ected in the
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 863
intense, post-crisis debates as to the possibly distorting
and misleading effects of both models and fair value.
In relation to models we have already mentioned some
of the erroneous assumptions that have subsequently
come to light. The SEC subsequently summarised these
concisely:
The performance history of the type of subprime mort-
gages that dominated many of the RMBS portfo-
lios. . .had been very short. Further, the performance
history that did exist occurred under very benign eco-
nomic conditions. These conditions included: consistent
high economic growth, interest rates at historic lows,
very low volatility in interest rates and a period where
housing prices increased consistently year over year’
(SEC, 2008a, p. 35).
With the wisdom of hindsight it is obvious that a model
can only be as good as the assumptions that are built into
it, and in this case, assumptions about default rates, and
the way these might feed into a wider housing downturn
that radically changes default correlations, were simply
wrong.
Whilst these wrong assumptions were at the heart of
the distortions that had unwittingly been built into the
models, other investigations suggested more pervasive
problems in the reliance on models. The CRMPG, for exam-
ple, noted a range of problems in risk management pro-
cesses which they suggest ‘struggled to keep up with the
complexities of product design and development’ in a
way that made their hedging strategies ineffective. They
also noted that the reliance of large integrated ?nancial
intermediaries on Value at Risk metrics masked the effects
of ‘correlations between exposures’ both within institu-
tions and in their counterparties. It was also argued that
models simply cannot predict low probability ‘extreme
events’ (Taleb, 2007).
The CRMPG concluded that ‘risk management profes-
sionals and senior management must recognise the limita-
tions of mathematical models’ suggesting that
‘incremental analytical detail must not be allowed to over-
whelm users of the data’ (2008, p. 83). There was arguably
no need to rely on sophisticated models in order to antici-
pate the possibility of highly correlated defaults on sub-
prime mortgages; a moment of thought would have been
suf?cient. Possibly the reassurance offered by complex
modelling had taken the place of thought. But when the
models proved to have been misleading, this in turn led to
‘a collapse in con?dence in a very broad range of structured
product ratings and a collapse in liquidity for such products’
(CRMPG, 2008, p. 53). The dilemma here, however, is intense
for, as Millo and MacKenzie (2008) have recently observed,
the markets simply could not have functioned without the
models and, in this sense, their organisational usefulness
was as signi?cant as their accuracy.
Post the onset of the crisis similar doubts have been
raised about the effects of fair value accounting. Supporters
argued that blaming fair value was like ‘shooting the mes-
senger’ or ‘going to a doctor for a diagnosis and then blam-
ing him for telling you are sick’ (Morgan, 2008). The SEC
investigation reported that, if anything, investors wanted
more transparency (SEC, 2008b, pp. 139–156). Opponents
suggested that the application of fair value was exacerbat-
ing market instability by applying the valuations arising
from sales in abnormal market conditions across all portfo-
lios. Rather than seek to resolve this either/or debate, the
important point here is to recognise that accounting was
indeed an agent in its own right – creating con?dence
(unwarranted at times) through the ‘transparency’ it cre-
ated as well as triggering decisions that fed the crisis. What
we want to observe here is the possibly ‘hyperreal’ interac-
tion of models and fair value accounting.
Macintosh, Shearer, Thornton, and Welker (2000) fol-
lowing Baudrillard de?ne ‘hyperreality’ as ‘the current con-
dition of postmodernity where simulacra are no longer
associated with any real referent and where signs, images,
and models circulate, detached from any real material ob-
jects’ (2000, p. 14). The capacity of CDO structures to ‘dis-
entangle’ risk from the underlying assets and make it
tradeable arguably opened the door to such hyperreality,
with models and accounting then creating a mutually
self-referencing hall of mirrors that was only shattered
when real defaults and correlations started to occur. As
we have seen, the price of a CDO tranche was initially mod-
el derived through the calculation of future cash ?ows on
RMBS, based on multiple assumptions including antici-
pated default and recovery rates and correlation. Subse-
quent valuation was then either model derived for
illiquid products, or marked-to-market against a compara-
ble product or indices. The apparently immediate focus of
fair value on the current ‘exit value’ thereby possibly
masked the way that the price re?ected only the modelled
assumptions of future pro?tability. In this way the success
of the CDO seems, at least with the wisdom of hindsight,
to have been based on no more than the modelled anticipa-
tion of pro?tability – and of course that is indeed precisely
how risk came to have a price. But then real pro?ts and
front loaded bonuses were taken on this basis, creating
incentives for further growth. For example, recent reports
suggest that between 2002 and 2008, the $76bn in net
pro?t of the ?ve largest US investment banks was dwarfed
by $190bn in bonuses, whilst 2008 data suggests that com-
bined net losses of $25.3bn nevertheless yielded $26bn in
bonuses (Lucchetti & Karnitschnig, 2009).
Far from signalling a return to reality, the dynamics of
the credit crisis retain a similarly hyperreal quality, as con-
cerns for market and liquidity risk drove indexed values
below those implied by underlying cash-?ows. As the
IMF fearfully anticipated in October 2007:
A small loss in value can force funds to sell large
amounts of assets as liquidations to meet margin calls
and, simultaneously, their redemptions, increase. Such
‘?re sales’ could lead to a vicious circle of forced sales,
as the widening of spreads forces hedge funds and oth-
ers who mark portfolios to market to post losses, possi-
bly sparking investor withdrawals and further forced
sales’ (2007, p. 20).
What must ?rst be observed in this anticipation is the
repeated use of the word ‘forced’. There is nothing natural
at work here but rather, in the terms of our preceding dis-
864 J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867
cussion, it is the framing of CDO deals – collateral require-
ments, liquidity facilities, regulations, accounting stan-
dards, etc. – that are the ‘forces’ at work, as well as panic
driven calculation. But equally important is the IMF’s rec-
ognition that such a defence of self interest might itself
then produce a vicious self feeding circle. Six months on,
the IMF starkly concluded that: ‘although structured ?-
nance can be bene?cial by allowing risks to be diversi?ed,
some complex and multi-layered products added little eco-
nomic value to the ?nancial system’ (IMF, 2008, p. 54).
Conclusions – the leverage of ‘Moral Hazard’
In seeking to account for self interest in the current
credit crisis this paper has asked two different but related
questions. The ?rst concerns accounting’s role in producing
the self, and the calculation of ‘self’ interest, and here we
must observe that accounting, and associated modelling
processes and practices, have been indispensable allies
for both individuals and institutions. Accounting in its
capacity to recognise (disentangle) pro?ts within a stream
of transactions has provided the motive, means and rea-
lised gains (and now losses) for all concerned. In a recent
speech Christopher Cox, the chairman of the SEC argued
that ‘accounting standards should not be viewed as a ?scal
policy tool to stimulate or moderate economic growth, but
rather as a means of producing neutral and objective mea-
surement of the ?nancial performance of public compa-
nies’ (2008). Such a view re?ects accounting’s public
image as no more than a neutral ‘intermediary’ that simply
passes on the truth to investors and markets. In this paper,
however, we have insisted that accounting and modelling
should be understood as agents in their own right – as
‘mediators’ that in their actual effects fed both the illusions
of rationality (greed) that fuelled market growth, and com-
pounded fear and panic as ‘credit risk’ escaped its pro?t-
able framing and became systemic (FSA, 2009).
This takes us back to the second aspect of accounting for
self interest that we have explored here; Greenspan’s
‘shocked disbelief’ as to how self interest could have failed
so dramatically. The key paradox of the current credit crisis
is that it has been created by the very processes that,
according to ?nance theory and those who marketed, secu-
ritised and tranched CDO products, were supposed to bet-
ter manage risk to the bene?t of all. The paradox is intense
– vast amounts of money have been earned in the process
of creating products and markets that in the end them-
selves ‘manufactured’ risk both for ?nancial markets and
for the ‘real’ economy (Beck, 1992; Giddens, 1999). In this
particular manifestation of the ‘risk society’ the proffered
solution for better risk management – the CDO – itself be-
came the source of risk. The problem of Greenspan’s ‘self
interest’ was not that it was insuf?ciently calculating, but
rather that it was fundamentally confused as to the real
nature of interests which, as Callon insists, and the current
crisis violently illustrates, always lie between rather than
within selves or institutions.
Here we can observe both the absolute limit and
responsibility of accounting, for in framing and possibly
hermetically sealing the calculative mind-set, it effects an
illusory separation between one entity and another, and
between personal agency and market dynamics. Or to re-
turn to Callon’s language, accounting is key in reproducing
a sense of self as ‘isolated – too isolated – and autonomous –
too autonomous’, such that it is largely blind and/or
indifferent to the unintended ‘side effects’ of the calculated
pursuit of self interest. Long before accounting numbers
reached investors for their decision making they were
informing the ever more single minded pursuit of pro?t
by participants, leveraged by pay structures that offered
immediate rewards for the trading of longer term risks.
At an individual level, fair value accounting merely mim-
icked the deal culture that it fed whilst encouraging indi-
viduals to be indifferent to the wider processes of which
they were a part. At an institutional level, accounting’s en-
tity focus encouraged the now obviously erroneous belief
that within ?nancial markets capital could feed endlessly
upon itself in a permanently expanding universe of pro?t-
able deals.
As noted earlier, the striking difference between a tradi-
tional lending relationship and the ‘originate to distribute’
model was the complex web of inter-institutional net-
works and relationships (inter-ests) that the latter created.
Signi?cantly, these inter-ests were rendered largely invisi-
ble by accounting. As losses started to ?ow so too did
blame and, on the back of this, there followed a host of
investigations which sought to go beyond the numbers to
the relational processes that had generated them. Through
these investigations it became evident that not only pro?ts
but also ‘moral hazard’ had been leveraged through the
network of market relations. Whilst in traditional dyadic
lending relationships the calculation of self interest in-
cluded a concern for the other’s ability to pay, with the
new model such concern was apparently judged to be
unnecessary. Evidence has emerged of both predatory bor-
rowing and lending practices (Ornstein, Tallman, & Hola-
han, 2006; Shiller, 2008). Signi?cantly, Mian and Su?
(2008) found that the incidence of default on subprime
mortgages was higher where these were securitised, sug-
gesting that with the originate to distribute model there
was little perceived incentive to monitor or appraise the
credit risk of individual borrowers since the risk was al-
most immediately passed on, earning an upfront commis-
sion. Similar ‘moral hazard’ applied in the banks and
investment banks since they too planed to pass on the risk
to other investors. Hedge funds were also, it appeared,
working on extraordinary fee structures earned in part
through short selling practices that arguably further feed
market volatility (Lo, 2008, chap. 10).
The rating agencies were also discovered, after the
event, to be caught up in all sorts of con?icts of interest
and poor organisational practices; they were short staffed
and relied on historical data of corporate bond defaults to
rate products that had no history. One staff email obtained
by the SEC noted concerns that the ?rm’s model did not
capture half of the deal’s risk, and that ‘it could be struc-
tured by cows and we would rate it’ (SEC, 2008a). Coffee
(2008) further observed that rating agency dependence
on a handful of originators and ‘the issuer pays’ fee struc-
ture fundamentally compromised their assumed indepen-
dence in the rating process. Finally, we should mention
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 865
regulation itself, which in the context of ?nancial markets
seems to take for granted the thought that all regulation
will itself furnish new opportunities for regulatory arbi-
trage. As Power has recently observed, regulatory style
has come to rely increasingly on the ‘self governing capac-
ities of organizations’, and this has set up a tension be-
tween public conformity with the normative ideal
embodied in audited compliance, and practice where par-
ticipants ‘realise they may be participating in the costly
construction of an illusion or fantasy of control’ (2007, p.
199). In the context of the credit crisis, reassurance seems
to have been drawn from the routines of risk management,
which nevertheless failed to identify and manage substan-
tive sources of institutional and inter-institutional risk.
So the tension that becomes visible here is between in-
ter-dependencies (inter-ests) that were systemic and the
calculations of individual and institutional ‘self’ interest
that were seemingly indifferent to these. The real hazard,
however, was not just the lack of care exercised in the bor-
rowing, lending, tranching, rating, leveraging, insuring and
buying of these ‘pass through’ products. It was also that, in
reality, they turned out to be anything but ‘pass through’
since the unanticipated consequences of this carelessness
then revisited agents at every point in the networks. The
error of ‘self’ interest lies in the way in which it imagines
the self as separate and separable from others. Its calcula-
tion conceives of others only in instrumental terms, and so
it remains blind or simply indifferent to the consequences
of its conduct for others. The moment of panic signi?es the
rediscovery of the self as vulnerable and dependent, but
this reality is instantly ?ed from in a renewed and often
self-defeating attempt to defend the self from others.
Whilst there is now an intense effort to de?ne new trans-
parencies that will effectively re-frame the pursuit of self
interest, arguably what are also needed are new forms of
accounting that make interdependencies visible, and new
‘intelligent’ processes of accountability that secure the
relational nature of interests.
Acknowledgements
We would like to thank Tyrone Carlin, Graeme Dean,
David Johnstone, Stewart Jones, Jan Mouritsen and Sue
Newberry for their comments on earlier drafts of this
paper.
References
Adrian, T., & Shin, H. (2008a). Liquidity and leverage. Federal Reserve Bank
of New York staff reports no. 328 (May).
Adrian, T., & Shin, H. (2008b). Liquidity and ?nancial contagion. Financial
stability review. Banque de France (11 February).
Ashraft, A. & Schuermann, T. (2008). Understanding the securitization of
subprime mortgage credit. Federal Reserve Bank of New York staff
reports no. 318 (March).
Augar, P. (2006). The greed merchants: How the investment banks played the
free market game. Penguin: Harmondsworth.
Beck, U. (1992). Risk society: Towards a new modernity. New Delhi: Sage.
Blackburn, R. (2008). The subprime crisis. New Left Review, March/April,
63–106.
Callon, M. (1986). Some elements of a sociology of translation:
Domestication of the scallops and the ?shermen of St. Brieuc Bay.
In J. Law (Ed.), Power, action and belief: A new sociology of knowledge?
(pp 196–223). London: Routledge.
Callon, M. (1998). The laws of the markets. Sociological review monograph.
Oxford: Blackwell.
Callon, M., Millo, Y., & Muniesa, F. (2007). Market devices. Sociological
review monograph. Oxford: Blackwell.
Callon, M., & Muniesa, F. (2005). Economic markets as calculative
collective devices. Organization Studies, 26(8), 1229–1250.
Chomsisengphet, S., & Pennington-Cross, A. (2006). The evolution of the
subprime mortgage market. Federal Reserve Bank of St. Louis review
(January/February).
Coffee, J. (2008). The role and impact of credit rating agencies on the
subprime credit markets. In Testimony before the Senate Banking
Committee (September 26).
Counterparty Risk Management Policy Group III (2008). Containing
systemic risk: The road to reform (August).
Cox, C. (2008). In Remarks before the AICPA national conference on current
SEC and PCAOB developments (December). <http://www.sec.gov/news/
speech/2008/spch120808cc.htm>.
Das, S. (2006). Traders, guns & money. London: Prentice Hall.
Economist (2007). Bearish turns – The subprime meltdown continued (June
23).
Engelen, E., Ertuk, I., Froud, J., Leaver, A., & Williams, K., (2008). Financial
innovation: Frame, conjuncture and bricolage. CRESC working paper
no. 59.
Felsenheimer, J., & Gisdakis, P. (2008). Credit crises: From tainted loans to
global ?nancial meltdown. Weinham: Wiley-Vch.
Fender, I., & Kiff, J. (2004). CDO rating methodology: Some thoughts on
model risk and its implications. Bank for International settlements
working paper no. 163 (November).
Financial Accounting Standards Board (FASB) (2006). SFAS 157. Fair value
measurements (September).
FSA (Financial Services Authority) (2009). The Turner Review: A regulatory
response to the global banking crisis, March. <http://www.fsa.gov.uk-
pubs-other-turner_review.pdf>.
Ghoshal, S., & Moran, P. (1996). Bad for practice. A critique of
transaction cost theory. Academy of Management Review, 21(1),
13–47.
Giddens, A. (1999). Risk and responsibility. The Modern Law Review, 62(1),
1–10.
Greenspan, A. (2005). Risk transfer and ?nancial stability. In Remarks at
the Federal Reserve Bank of Chicago’s 41st annual conference on bank
structure , May 5, Chicago, IL.
Greenspan, A. (2008). Transcript of testimony to House
Oversight Committee. <http://oversight.house.gov/documents/
20081024163819.pdf>.
IMF (2007). Global ?nancial stability report: Financial market turbulence –
Causes, consequences and policies (October). <http://www.imf.org-
External-Pubs-FT-GFSR-2007-02-pdf-text.pdf>.
IMF (2008). Global ?nancial stability report: Containing systemic risks and
restoring ?nancial soundness (April). <http://www.imf.org-external-
pubs-ft-gfsr-2008–1-pdf-text.pdf>.
International Accounting Standards Board (2005). IAS 39. Financial
instruments: Recognition and measurement.
Joint Forum on Credit Risk Transfer (2004). Bank for International
Settlements (March).
Jones, S., & Peat, M. (2008). Credit derivatives: Current practices and
controversies. In S. Jones & D. Henscher (Eds.), Advances in credit risk
modelling and corporate bankruptcy prediction (pp. 207–241). New
York: Cambridge University Press.
Latour, B. (2005). Reassembling the social: An introduction to actor-network-
theory. Oxford: Clarendon.
Lo, A. (2008). Hedge funds: An analytic perspective. Princeton NJ: Princeton
University Press.
Lucchetti, A., & Karnitschnig, M. (2009). Wall street workers brace for
more pay shrinkage. Wall Street Journal (February 2).
Macintosh, N., Shearer, T., Thornton, D., & Welker, M. (2000). Accounting
as simulacrum and hyperreality: Perspectives on income and capital.
Accounting, Organizations and Society, 25(1), 13–50.
MacKenzie, D. (2006). An engine, not a camera: How ?nancial models shape
markets. Cambridge MA: MIT Press.
MacKenzie, D., Muniesa, F., & Siu, L. (2007). Do economists make markets?
On the performativity of economics. Princeton: Princeton University
Press.
Markowitz, H. (1959). Portfolio selection: Ef?cient diversi?cation of
investments. New York: John Wiley and Sons.
Mian, A., & Su?, A., (2008). The consequences of mortgage credit expansion:
Evidence from the 2007 mortgage default crisis (May). <http://ssrn.com/
abstract=1072304>.
Miller, P. (1998). The margins of accounting. European Accounting Review,
7, 605–621.
866 J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867
Miller, P., & O’Leary, T. (2007). Mediating instruments and making
markets: Capital budgeting, science and the economy. Accounting,
Organizations and Society, 32(7/8), 701–734.
Millo, Y., & MacKenzie, D. (2008). The usefulness of inaccurate models:
Towards an understanding of the emergence of ?nancial risk
management. Accounting, Organizations and Society, doi:10.1016/
j.aos.2008.10.002.
Morgan, J.P. (2008). Shooting the messenger. Global Equity Research
(September).
Morgan, M., & Morrison, M. (1999). Models as mediators: Perspectives on
natural and social science. Cambridge: Cambridge University Press.
Morris, C. (2008). The trillion dollar meltdown. New York: Perseus Books.
Mortgage Bankers Association (2008). Delinquencies and foreclosures
increase in latest MBA national deliquency survey (March 3). <http://
www.mortgagebankers.org/NewsandMedia/PressCenter/
60619.htm>.
Ornstein, S., Tallman, D., & Holahan, J. (2006). Predatory lending and the
secondary market. Journal of Structured Finance, Fall, 54–64.
Pleven, L., & Craig, S. (2008). Deal fees under ?re amid mortgage crisis –
Guaranteed rewards of bankers, middlemen are in the spotlight. Wall
Street Journal (January 17).
Power, M. (2007). Organized uncertainty: Designing a world of risk
management. Oxford: Oxford University Press.
Roberts, J. (2001). Trust and control in Anglo-American systems of
corporate governance. The individualising and socialising effects of
processes of accountability. Human Relations, 54(12), 1547–1572.
Ryan, S. (2008). Accounting in and for the subprime crisis. Stern School of
Business, New York University. <http://ssrn.com/abstract=1115323>.
SEC (2008a). Summary report of issues identi?ed in the commission staff’s
examination of select credit rating agencies (July).
SEC (2008b). Report and recommendations pursuant to section 133 of the
emergency economic stabilization act of 2008: Study of mark-to-market
accounting (December).
SIFMA (2007). Securities Industry and Financial Markets Association: Global
CDO Market Issuance Data. <http://archives1.sifma.org/assets/?les/
SIFMA_CDOIssuanceData2007q1.pdf>.
Shiller, R. (2008). The subprime solution: How today’s global ?nancial crisis
happened and what to do about it. Princeton: Princeton University
Press.
Taleb, N. (2007). The black swan: The impact of the highly improbable. New
York: Random House.
Weick, K. (1993). The collapse of sense-making in organisations:
The Mann Gulch disaster. Administrative Science Quarterly, 38,
628–652.
Whittington, G. (2008). Fair value and the IASB/FASB conceptual
framework project: An alternative view. Abacus, 44(2), 139–168.
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 867
doc_875233860.pdf
Taking as its starting point Alan Greenspan’s ‘shocked disbelief’ in the failure of institutional
self interest to prevent the credit crisis, this paper sets out to explore two related
questions. How was self interest constructed in financial markets? And how might we
account for its failure? Conceptually the paper draws upon Callon’s (1998) analysis of
‘agent–networks’, the importance this gives to the agency of non-humans, and his complementary
notions of ‘framing’/‘disentanglement’ and ‘overflowing’ as these allow and subvert
the calculation of self interest. Empirically, the paper then presents a sketch of
these processes in the rise and then fall of the market for collateralised debt obligations
(CDOs) that was central to the credit crisis. The final substantive section of the paper
reflects on the role and ‘hyperreal’ interaction of accounting and models as ‘mediators’
in these processes.
Accounting for self interest in the credit crisis
John Roberts
*
, Megan Jones
University of Sydney, Discipline of Accounting, Faculty of Economics and Business Building H69, Cnr. Codrington and Rose Streets, Darlington, NSW 2006, Australia
a r t i c l e i n f o a b s t r a c t
Taking as its starting point Alan Greenspan’s ‘shocked disbelief’ in the failure of institu-
tional self interest to prevent the credit crisis, this paper sets out to explore two related
questions. How was self interest constructed in ?nancial markets? And how might we
account for its failure? Conceptually the paper draws upon Callon’s (1998) analysis of
‘agent–networks’, the importance this gives to the agency of non-humans, and his comple-
mentary notions of ‘framing’/‘disentanglement’ and ‘over?owing’ as these allow and sub-
vert the calculation of self interest. Empirically, the paper then presents a sketch of
these processes in the rise and then fall of the market for collateralised debt obligations
(CDOs) that was central to the credit crisis. The ?nal substantive section of the paper
re?ects on the role and ‘hyperreal’ interaction of accounting and models as ‘mediators’
in these processes.
Ó 2009 Elsevier Ltd. All rights reserved.
Introduction
‘I made a mistake in presuming that the self interest of
organizations, speci?cally banks and others, were such
that they were best capable of protecting their own share-
holders and their equity in the ?rms. So the problem here is
something which looked to be a very solid edi?ce, and
indeed, a critical pillar to market competition and free
markets, did break down. And I think that, as I said did
shock me. I still do not understand fully why it happened
and, obviously, to the extent that I ?gure out where it hap-
pened and why, I will change my views. I found a ?aw in
the model that I perceived is the critical functioning struc-
ture that de?nes how the world works, so to speak’. Alan
Greenspan, October 2008.
Alan Greenspan’s ‘shocked disbelief’ is an appropriately
cautious starting point for any attempt to make sense of
the credit crisis. His belief in the effectiveness of self inter-
est as a pillar of market competition, having served him
well for the last 40 years, had broken down in the face of
evidence from the current credit crisis. As he observed la-
ter in questioning by the House Oversight Committee, we
all need an ideology or conceptual framework in order to
act, and it is painful, disorienting and disabling when it
breaks down.
Our own interest in the crisis arose from a related start-
ing point. At the beginning of 2008 it was clear that some-
thing unexpected and unanticipated was happening in
?nancial markets and that knowledge had failed market
participants in important ways. As Latour (2005) observes,
the assembly or collapse of networks of relations repre-
sents an ideal moment for research, for then associations
that are otherwise invisible and simply taken for granted
come into clear view. We wanted to try to understand
what was happening, and set out to explore the rise and
fall of the market for collateralised debt obligations (CDOs)
that seemed to be a central player in the crisis. We began
our research with interviews with market participants,
but as the crisis developed these became increasingly dif?-
cult to organise. However, these dif?culties were more
than compensated for by the host of analyses that began
to emerge based on the research of bodies such as the
SEC, the Counterparty Risk Management Policy Group
(CRMPG), the IMF and Financial Stability Forum – groups
that had much better access and research resources than
us.
The credit crisis will no doubt absorb the energies
of researchers for many years to come. In what follows,
0361-3682/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2009.03.004
* Corresponding author. Tel.: +61 2 9351 6638.
E-mail address: [email protected] (J. Roberts).
Accounting, Organizations and Society 34 (2009) 856–867
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
drawing upon the above reports and other analyses of the
crisis, we want to offer a sketch of some of the key ele-
ments that contributed to the CDO market’s rise and fall
in the hope of casting some light on the catastrophic failure
of understanding that the crisis seems to embody. In par-
ticular, the paper explores the role played by models and
accounting as ubiquitous ‘mediators’ of almost all market
relationships, arguing that they fed both the illusion of
rationality (greed) that fuelled market growth, and com-
pounded fear and panic as the market fell. As a framework
for this more process-oriented view of the crisis, we begin
with a discussion of some elements of Callon’s seminal
contribution to social studies of ?nance in ‘The Laws of
the Markets’ (1998) (see also Callon & Muniesa, 2005;
MacKenzie, Muniesa, & Siu, 2007). When viewed through
the lens of the current credit crisis this analysis seems par-
ticularly prescient and points to the misconceptions and
lacunae that possibly produced Greenspan’s and others’
‘shocked disbelief’.
The construction of self interest
Callon begins his analysis of markets by noting North’s
observation that a peculiar omission in economics is its
failure to discuss markets. Callon follows Guesnerie’s de?-
nition of a market as a ‘coordination device’, in which
agents pursue their own interests through economic calcu-
lations. The generally divergent interests of agents lead
them to engage in market transactions that resolve con?ict
by de?ning a price. For Callon such economic conceptions
of market coordination ‘have the common feature of pro-
viding autonomous – over autonomous – and isolated –
over isolated – agents with the social relations which, by
opening them up to their environment, enable them to
co-ordinate their actions with those of other agents’
(1998, p. 7). His alternative is to reverse the economic
assumption of atomism by taking agents’ dependence on
their environment as his starting point.
This reversal is very important for it is a decisive re-con-
ceptualisation of a founding assumption of economics and
?nance – that of a self-seeking and opportunistic (calculat-
ing) individual as the basic unit of analysis. In place of this
traditional dualism of ‘agent’ and market relations or ‘net-
works’, Callon insists that the unit of analysis must be the
‘agent–network’; ‘agent and network are, in a sense, two
sides of the same coin’. A network from this perspective
is not something that merely connects already existing
entities, but rather itself produces the calculating agents;
‘the agents, their dimensions, and what they are and do,
all depend upon the morphology of the relations in which
they are involved’ (1998, p. 9). The image of the market
here is not that of an encompassing context for agency that
contains its own invisible hand magically transforming the
pursuit of self interest into a public good, but rather of the
market only as myriad transactions being repeatedly en-
acted; ‘once the transaction has been concluded the agents
are quits; they extract themselves from anonymity only
momentarily, slipping back into it immediately after-
wards’. Here we want to follow two elements of Callon’s
subsequent analysis of markets – his analysis of processes
of ‘framing’/‘disentanglement’ and ‘over?owing’ – as these
each build upon his insistence of the indivisibility of
‘agent–networks’.
From an outsider’s perspective one of the most remark-
able features of the world of ?nance is the sheer energy
that individuals and institutions put into the business of
making money. Rather than see such self interest as natu-
ral and reliably predictable, Callon insists that we have to
work very hard to produce the self, and the self interest,
that Greenspan took as a given. Critically it depends upon
‘calculation’, which Callon argues should be understood
neither as a cognitive characteristic nor a cultural phenom-
ena, but rather as dependent upon the complex temporal
‘framing’ of the relationships in which a person is engaged.
He draws the concept of ‘framing’ from Goffman’s micro-
analyses of the ‘staging’ of interaction, and argues that in
relation to markets it should be understood as a process
of ‘disentanglement’. As he puts it:
Framing is an operation used to de?ne agents (an indi-
vidual person or group of persons) who are clearly dis-
tinct and dissociated fromone another. It also allows for
the de?nition of objects, goods and merchandise which
are perfectly identi?able, and can be separated not only
from other goods, but also from the actors involved, for
example in their conception, production, circulation
and use. It is owing to this framing that the market
can exist and that distinct agents and distinct goods
can be brought into play (1998, p. 17).
In relation to the CDO market in following Callon we
can think of the energetic pursuit of self interest then not
as natural, but rather as the product of myriad processes
and devices that allowed the self, and self interestedness,
to be calculated, pursued and realised. Only this ‘framing’
of market relations allows the self and self interest to be
‘disentangled’ within the relationships through which it
is realised. The work of framing/disentanglement was mas-
sive. The development of the CDO in its multiple forms,
and the new opportunities and relationships that these
products opened up and made possible were themselves
an essential part of this framing/disentanglement. So too
were legal and employment contracts, bonus schemes,
institutional identities and roles, market indices and
countless forms of market data, etc. (Callon, Millo, & Mu-
niesa, 2007). But in analysing such active processes of
framing, Callon gives a key role to the agency of what he
calls ‘calculating tools’ and in particular to accounting; ‘full
signi?cance has to be restored to that humble, disclaimed
and misunderstood practice; accounting and the tools it
elaborates’ (1998, p. 23).
Latour’s (2005) later distinction between what he calls
‘intermediaries’ and ‘mediators’ is useful in clarifying the
importance Callon gives to ‘calculating tools’. Whilst ‘inter-
mediaries’ act merely as dependable relays in a network,
‘mediators transform, translate, distort, and modify the
elements they are supposed to carry’ (2005, p. 39). The self
image, or at least public image, of accounting is that of an
intermediary – it does no more than render ‘what is’ ‘trans-
parent’. But for Callon (1998) and Miller (1998; Miller &
O’Leary, 2007) whose work he is following, accounting
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 857
should be seen as a mediator. As Callon puts it, calculating
tools ‘do not merely record a reality independent of them-
selves; they contribute powerfully to shaping, simply by
measuring it, the reality they measure’ (1998, p. 23). Part
of this agency involves the ways that, as accounting tools
are constantly recon?gured, they ‘authorize decisions that
are more and more calculated or (to use the commonly ac-
cepted word) more and more rational’. In this respect the
CDO market was from its inception absolutely dependent
on the use of mathematical modelling techniques (Mac-
Kenzie, 2006; Morgan & Morrison, 1999), and its evolution
involved a dizzying ‘bricolage’ of ever more complex prod-
uct forms designed to entice further pro?table growth
(Engelen, Ertuk, Froud, Leaver, & Williams, 2008). Along-
side this product innovation, accounting standards and
practices were themselves forced to evolve, but this itself
provided fuel for the development of ‘new calculative
strategies’. Accounting is, of course, also vital as the ?nal
measure of the achievements of self interest:
Money comes in last in a process of quanti?cation and
production of ?gures, measurement and correlations
of all kinds. It is the ?nal piece, the keystone in a metro-
logical system that is already in place and of which it
merely guarantees the unity and coherence (Callon,
1998, p. 22).
Accounting only captures the pro?ts (losses) momen-
tarily on their way to others. In investment banks this typ-
ically involved the equal sharing of pro?ts between
employee annual bonuses and shareholders (Augar, 2006).
One moment in Callon’s analysis of markets as coordi-
nating devices suggests then the necessity, if one is to
understand the current crisis, of exploring the intense la-
bour of framing/disentanglement that took place in ?nan-
cial markets in the construction of individual and
institutional self interest, and the ubiquitous role of non-
humans agents in this – notably the models and account-
ing that mediated almost all relationships. However, the
full signi?cance of his analysis of markets as ‘agent–net-
works’ only emerges with his exploration of the corollary
of framing in the parallel process of what he calls ‘over-
?owing’. Callon argues that ‘without this framing the states
of the world cannot be described and listed, and conse-
quently, the effects of the different conceivable actions
cannot be anticipated’ (1998, p. 17). There is an echo here
of the current crisis and the paralysis of action that arises
when calculation becomes unreliable or even impossible.
Callon argues that for economics framing is regarded as
the norm, and that over?ows which create ‘externalities’
are from this perspective exceptional and ‘accidental leaks’
to be corrected. However, by insisting that agents are
inseparable from, and indeed constituted only in and
through the networks of relations in which they are
embedded, Callon reverses this assumption. From a con-
structivist perspective, he argues, over?ows should be re-
garded as the norm and framing as itself both ‘expensive
and always imperfect’ (1998, p. 252). This is because fram-
ing is always incomplete and must be so if value is to be
created; ‘it is because an actor’s output gets necessarily be-
yond her entire control so as to generate pro?ts, that the
actor is unable to avoid externalities’ (1998, p. 23). Impor-
tantly, Callon suggests that over?owing is multidirectional
but nevertheless travels back along the very paths (net-
works) that are initially created in the pursuit of individual
and institutional self interest. However, such over?owing
is initially invisible and must itself be subjected to mea-
surement in order to be recognised and then ideally re-
framed.
Taken together Callon’s notions of framing/disentangle-
ment and over?owing offer a very powerful lens through
which to re-think our assumptions about the nature of self
interest in markets. Typically, as evidenced in Greenspan’s
comments, interests are taken to refer only to the self; as if
interests were somehow internal to the self (or institution)
and that it is the pursuit/defence of these that de?nes the
nature of economic rationality. For Callon, however, this is
to misunderstand and ignore the tension that is implicit in
self interest. The etymology of interests as ‘inter-esse’ re-
veals this tension for interests always lie between selves
rather than within the individual (institution). There is in this
sense no self, or possibility of a self, divorced from the rela-
tionships that are the very ground of its own agency. From
this perspective ?nancial markets are a ?nely balanced
high-wire act in which network interdependencies are
endlessly elaborated and intensi?ed through the ever more
calculated pursuit of individual and institutional gain. As
Callon puts it:
How can one perform framing when one has to be
attentive to all this over?owing? How is it possible to
become homo clausus when survival requires one to
be homo apertus? This question is at the heart of the
stock market and the speculative behaviour that it
spawns. Nowhere is the tension between framing and
over?owing so intense and so dif?cult to control
(1998, p. 25).
So in this sense crises and volatility, the jumps from
boom to bust, are to be expected and result from no more
than a small shift in the precarious balance between the la-
bour of framing and consequent over?owing that is ever
present in markets.
In the remainder of the paper we take these metaphors
of framing/disentanglement and over?ows and use them
as the basis for a sketch of some of the key elements in-
volved in the construction of self interest in the rise and
fall of the market for CDOs. In the next section, we explore
the growth of the CDO market; the evolution of the prod-
ucts, the core ‘framing/disentanglement’ processes of secu-
ritisation and tranching that were at the heart of these
innovations, and the work done by models and accounting
in facilitating this growth. A key element in the story is the
greatly extended network of interests that was created
through these innovative forms of securitisation. While
markets were apparently working well, these seemed only
to offer participants ever expanding opportunities to
‘trade’ risk widely and pro?tably. However, we will then
seek to describe how, with the trigger of subprime mort-
gage defaults and a housing market downturn, these net-
works became channels for losses and panic. In
particular, we want to trace the peculiar ‘over?owing’ of
858 J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867
risk that has taken place from a seemingly safely framed
and pro?table ‘credit risk’, into ‘market risk’ in which as-
sets prices fell in unexpected ways, and ‘counterparty risk’
where fear of the potential exposure of others to losses
quite literally brought credit markets to a standstill, and
thereby created the current liquidity crisis (CRMPG,
2008). The ?nal substantive section of the paper draws to-
gether our re?ections on the role of accounting and its
interaction with models in the crisis.
Framing/disentanglement in the CDO markets
The market for CDOs emerged in the early 1980s but
only became a major asset class in the late 1990s, after
which there was an explosive growth in the volume and
value of transactions, particularly from 2003 to 2007 (SIF-
MA, 2007). Between 1995 and 2007 the total value of out-
standing asset-backed securities grew from $500bn to
$2500bn. Within this growth home equity loans grew in
importance, accounting for some $600bn in 2007 (Felsen-
heimer & Gisdakis, 2008). In 2001 the value of new sub-
prime mortgages originated was $160bn, a ?gure that
increased to $600bn in 2006 (Blackburn, 2008).
Securitisation
What is common to this family of products are pro-
cesses of ‘securitisation’ and ‘tranching’. Essentially securi-
tisation involves packaging up illiquid debt into securities
that can then be sold on to others, thereby creating both
liquidity and the dissipation of the risk attached to the
debt. Securitisation has a long history but in relation to
mortgages it involves a shift from an ‘originate to hold’
model of traditional mortgage lending to an ‘originate to
distribute’ model. The shift is important for, when com-
pared to the traditional model in which borrower and len-
der are bound together in a dyadic long term relationship,
the new model involves an extended chain of new institu-
tional relationships and inter-dependencies. With mort-
gage-backed securities the borrower and mortgage
broker/lending bank still begin the process but individual
mortgages are then bundled together and sold on to inves-
tors. The key rationale for securitisation is drawn from
portfolio theory and, in particular, the notion that ‘diversi-
?ed’ pools of assets – say different grades of mortgage,
from different regions or countries, or indeed different
combinations of assets – create less ‘idiosyncratic risk’
(Markowitz, 1959). The ‘disentanglement’ achieved by sec-
uritisation is therefore of the individual borrower from
individual lender.
Tranching
Whilst in a securitisation of assets cash ?ows are simply
passed on to investors, with CDOs asset-backed securities
are then subjected to a further process of what is called
‘tranching’. Tranching is a particularly creative form of dis-
entanglement that allows the creation of different invest-
ment products with very different risk characteristics
from an underlying portfolio of assets. It also involves a
seemingly magic process through which low quality debt
can, in part, be transformed into higher quality, investment
grade assets that, by virtue of this ‘credit enhancement’,
can then be sold to/taken on by a much larger population
of investors. The simple example given of such a process
suggests a threefold tranching of the assets into an ‘equity’
or ‘?rst loss’ tranche which references say 0–3% of the port-
folio, a ‘mezzanine’ tranche referencing say a further 12% of
the portfolio, and a ‘senior’ tranche which references the
remaining 85% of the portfolio of assets. Tranching effects
a skewed distribution of the risks and returns associated
with the underlying assets. What is termed a ‘waterfall
principle’ (an ominous anticipation of future over?ows)
dictates that cash ?ows from the pool of assets go ?rst to
the senior tranche, then mezzanine, then equity. The allo-
cation of risks and consequently higher returns, however,
follows a ‘reverse waterfall’ principle in which any losses
arising, say from defaults on subprime mortgages, are ?rst
absorbed by the equity tranche until it is completely
eroded, then the mezzanine and only then effect the senior
tranche. As a result of tranching the most senior tranches
can be rated as investment grade securities offering higher
returns than similarly rated corporate bonds, by virtue of a
structure in which others absorb the risks of initial defaults
in their entirety for a yet higher return. Often the equity
piece was retained by the originator as a mark of con?-
dence, or sold on to hedge funds which had an appetite
for such risk and return. However, the relative dif?culty
of passing on these riskier tranches was also the inspira-
tion for more complex structured ?nance CDOs that essen-
tially re-securitised and re-tranched these lower tranches
thereby creating yet more senior level tranches.
The effectiveness of such tranching procedures depends
critically upon the nature of the assets involved and, in
particular, the ‘correlation’ of risks between different as-
sets in the portfolio. Whilst securitisation of assets ideally
results in a diversi?ed portfolio that reduces ‘idiosyncratic
risk’, there remains the ‘systemic’ risk created by the pos-
sibility that macro events – for example, a downturn in
the housing market – might cause the default risk in the
portfolio to be highly correlated such that one default is
tied to multiple defaults. As a result the success of tran-
ching depends critically upon the calculation of a whole
variety of variables, including such ‘correlation’ risk. This
was one of the critical spaces in which the expertise of
‘quants’ staff and the models they employed became vital
to the operation of credit markets. The other spaces where
models became all important were in their parallel use by
credit rating agencies and in ‘risk management’ processes
in the banks, investment banks and hedge funds, by means
of which institutional risks were appraised and monitored
(Millo & MacKenzie, 2008; Power, 2007).
For the modelling of a mortgage backed CDO, for exam-
ple, data inputs would be divided into pre-payment vari-
ables, such as interest rates, asset prices, macroeconomic
and housing data, and default variables, such as the loan
to value ratios, estimates of default probabilities and likely
recovery rates (Fender & Kiff, 2004). Once assembled, such
data would then be subjected to both random and selected
scenario simulations to gauge the possible correlation ef-
fects the loans might display under different conditions.
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 859
On this basis the capital structure of the CDO could then be
calculated – the size and ‘thickness’ of the different tran-
ches – and cash ?ows discounted and allocated to the dif-
ferent tranches. This, in turn, allowed the originator to
assess the likelihood of excess spread in the loan pool each
month over the term of the investment, and to ensure that
payments of interest and principal covered proposed pay-
ments to investors. The resultant structure could then be
‘stress tested’ in a variety of ways against current market
and historical data to verify the modelling process. Since
the attraction and marketability of such products de-
pended upon senior tranches receiving an investment
grade (AAA) rating then, in practice, the ?nal structure of
the CDO was the product of a process of negotiation be-
tween the originator and the credit rating agency. The
involvement of the credit rating agencies was critical since
their rating served as an authoritative, apparently indepen-
dent, mark of assurance in relation to what is an invariably
complex and opaque set of calculations of the risk/return
pro?le of the different tranches (Coffee, 2008; SEC, 2008a).
The simplest version of a CDO was called a ‘cash ?ow’ or
‘balance sheet’ CDO; so called because it was managed off
balance sheet through a bankruptcy remote Special Pur-
pose Vehicle (SPV). In this simple version, the assets are
bought and packaged together, and sold to the SPV, which
after tranching and rating, then passes on the risk but also
the interest and principal to the ultimate investors. Press
attention has focused on CDOs backed by subprime mort-
gages but in reality a whole variety of different kinds of
loans and assets, including mortgages (Prime, Alt A and
Sub-prime) as well as student loans, credit card and car
loans, as well as combinations of these and other assets,
were used as the basis for CDO products (Felsenheimer &
Gisdakis, 2008).
Enrolling interests to the CDO
If we stay for the moment with CDOs backed, or par-
tially backed, by subprime mortgages then we can think
about the growth of the market through Callon’s (1986)
language of ‘translation’; the attractiveness of the CDO to
all the interests that it enrolled. For the sub-prime bor-
rower it offered access to the ‘American Dream’ of home
ownership despite their no income, no job status. For lend-
ing banks securitisation allowed them to pass on the risks
they had formerly held to maturity, providing them with
relief from regulatory capital requirements and thus allow-
ing further lending. For investment banks the CDO opened
up a new and highly pro?table stream of work; as deal
makers there were origination fees to be earned as well
as margin on the tranched assets as fuel for this year’s bo-
nus pool. By 2006 30% of all investment bank earnings
were from structured ?nance activities (Pleven & Craig,
2008). Similarly, for hedge funds charging high fees and
offering high returns, the mezzanine and equity tranches
were grasped as an ideal product. This then fed back into
pro?table prime brokerage business for the investment
banks in which they provided clearing services and ?nance
for the leveraged strategies of the hedge funds. For the
credit rating agencies the CDO opened up a whole new cat-
egory of work in addition to their traditional low margin
bond rating work (Coffee, 2008). Likewise for the monoline
insurers who insured the senior tranches as part of the
credit enhancement process. At the end of the chain were
a whole range of individual and institutional investors
who, in a time when yields were low, were able to invest
through the CDO in products that were AAA rated but of-
fered a substantial premium to similarly rated corporate
bonds. Finally, we should mention market regulators.
Greenspan and others pronounced themselves pleased
with these new structured ?nance products which, they
argued, had contributed both liquidity and stability to
?nancial markets by virtue of the dissipation of risk that
they allowed (Greenspan, 2005; Joint Forum Credit Risk
Transfer, 2004).
Synthetic CDOs
Perhaps precisely because of the attraction and success
of the cash ?ow CDO, in the early 2000s a new variety of
product emerged in the form of so called ‘synthetic’ CDOs.
With this version of the product there is no passing on of
the principal, interest and ownership of the assets to the
ultimate investor (this was not always legally possible),
rather these are retained by the originator and credit de-
fault swaps that mimic the risk pro?le of the assets are en-
tered into with the SPV which in turn sells the risk on to
investors. Credit default swaps (CDS) are typically ex-
plained using the analogy of an insurance policy, with
the originator buying and investors effectively selling pro-
tection on the risk of potential losses on the underlying as-
sets. Unlike insurance companies, however, the market is
unregulated. Synthetic CDOs made use of CDSs in two
ways. In funded synthetics the CDS sits between the origi-
nator and SPV but investors pay in the full amount of the
tranche, which is then used as collateral or invested in high
quality securities. There is a further distinction here be-
tween managed and unmanaged funds – essentially
whether the asset pool remains static or is actively man-
aged through sales and purchases of assets throughout
the life of the product. By contrast, in unfunded synthetic
CDOs there is no initial investment; investors receive a
periodic spread payment on the CDSs related to their tran-
che but are required to cover the cost of any defaults if they
arise.
The key element of both funded and unfunded synthetic
CDOs is that the risk is disentangled from the underlying
assets and allowed to travel. As Jones and Peat explain:
‘synthetic CDOs do not involve any actual sale of assets
by the originator – only the underlying credit risk of the
reference assets is transferred to the counterparty. The
originator remains the legal and bene?cial owner of those
assets’ (2008, p. 216). One important feature of the CDS
market is that it is an ‘over-the-counter’ (OTC) market
and therefore the distribution of risk that contracts allow
remains opaque; a factor that was the basis of more recent
counterparty risk. The CDS had itself apparently proved
immensely popular as a credit instrument, both because
of its risk disentanglement and dissipation capabilities as
well as its unfunded nature. This fuelled an explosive
growth in the market with some $45trn of contracts out-
standing as of 2007 (Morris, 2008).
860 J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867
The disentanglement of risk from the underlying assets
also changes the nature of the risk from ‘default’ risk to
‘market’ risk. As Felsenheimer and Gisdakis explain:
The great achievement of credit markets is that they
have enabled investors to buy and sell credit risky
assets, thus transforming an event risk (i.e. the risk of
a default event) into a price risk (i.e. the risk that the
value of an asset declines over time). The price
expresses the probability that such an event occurs
and the anticipated loss that accompanies the event.
This transformation process was accompanied by the
development of derivative structures in order to
improve the tradability of the underlying risks (2008,
p. 154).
The emergence of synthetic CDOs was also driven by a
very different set of motives. Whilst balance sheet CDOs
were used to of?oad certain assets for the purposes of risk
reduction or re?nancing, the development of the synthetic
CDO was motivated primarily by the trading and pro?t
(arbitrage) opportunities offered by these instruments
(Das, 2006). Demand for the CDO was then driven as much
by investors as by the supply of assets whose risk de-
manded securitisation. This then fuelled the further
growth of the market as well as the demand for assets that
could be processed in this way; subprime mortgages were
particularly attractive because of the premium charged to
the borrower over the life of the loan. As the Joint Forum
on Credit Risk Transfer summarised it: ‘the credit risk
transfer market is characterised by signi?cant product
innovation, an increasing number of market participants,
growth in overall transaction volumes, and perceived con-
tinued pro?t opportunities for ?nancial intermediaries’
(2004, p. 1).
The development of the synthetic CDO led to, and was
facilitated by, the creation of new American and European
‘indices’ for CDSs, CDX and iTraxx, including the ABX.HE,
which was also known as the sub-prime index. These then
allowed investors to trade in the average credit spread of
an underlying portfolio and led to the development of
CDO
2
and CDO
3
– CDOs of CDOs of CDOs. Other yet more
innovative and exotic versions of the CDO included Lever-
aged Super Senior Tranches, Constant Proportion Debt
Obligations, and Structured Investment Vehicles (SIVs),
which involved the combination of high quality assets
and high levels of leverage to offer extraordinary but
apparently low risk returns. SIVs were important because
they used short term asset-backed commercial paper to
fund longer term debts, thus creating direct links between
underlying mortgage-backed securities and short term
money markets. These off balance sheet vehicles were typ-
ically structured to include various triggers designed to
limit leveraged losses, as well as liquidity facilities with
their sponsoring banks.
Accounting
Up until the middle of 2007 accounting’s role in the
growth of the CDO markets could be seen as that of a
complex but innocent ‘intermediary’ that merely recogni-
sed the pro?tability of these new innovations. In response
to the rapid growth of the structured credit universe both
the FASB and IASB had been required to develop new
standards in relation to the valuation of ?nancial instru-
ments. One critical set of issues concerned the accounting
treatment of off balance sheet entities (OBSEs) and here,
post-Enron, such structures were only allowed if no single
institution held the majority of the risks and rewards or
control rights. For the measurement of on balance sheet
assets a distinction was made between whether assets
are available for sale (AFS) or to be held to maturity
(HTM). Both the IASB and FASB required the measure-
ment of AFS assets at ‘fair value’, which is de?ned by
the FASB as ‘the price that would be received to sell an
asset or paid to transfer a liability in an orderly transac-
tion between market participants at the measurement
date’ (FAS 157, para 5), and by the IASB as ‘the amount
for which an asset could be exchanged between knowl-
edgeable, willing parties in an arms length transaction’
(IAS 16, para 6). Both accounting standards then offer a
threefold hierarchy of bases for the determination of fair
value. Least problematic is the use of quoted prices for
identical assets and liabilities. If these are not available,
the second level of the hierarchy stipulates the use of ‘ob-
servable inputs’, which might include quoted prices for
similar assets, or inputs derived from an observable index.
In the absence of such liquid and active markets, or com-
parable prices, then valuation has to be done on a ‘mark-
to-model’ basis, rather than a ‘mark-to-market’ basis. This
serves to create an immediate tie between accounting
valuations and model assumptions that are clearly depen-
dent upon judgement (see, Ryan, 2008; Whittington,
2008). SEC data suggests that 15% were measured at level
1, 76% at level 2 and 9% at level 3 with relatively small
differences between sectors (SEC, 2008b).
Leverage
In 2005/2006 the spreads, particularly on indices for
synthetic CDOs, became increasingly narrow – a possible
signal that risk was being ‘under-priced’ – but by then such
was the con?dence of the market in the CDO that the re-
sponse was simply to add leverage to the deals (Adrian &
Shin, 2008a). Leverage involves the simplest of calculations
– multiplication. As Felsenheimer and Gisdakis explain:
The investment rationale was very simple; if the risk
premium is low, then risk has to be low. And if the risk
premiumprovides only a quarter of the return, then just
invest four times as much. This strategy was supported
by ?nancial markets, since in a low yield, low spread
environment the cost of capital is also low (2008, p.
156).
The levels of leverage were considerable; in banks it
was around 12 to 1, for investment banks around 30–1,
and for some of the more exotic CDO products up to 60–
1. Importantly the tranching process itself creates ‘embed-
ded’ leverage in the lower tranches so that the CDO
2
and
CDO
3
could include leverage upon embedded leverage.
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 861
Whilst markets were stable such leverage was an obvious
way to multiply pro?ts but it was also laying the seeds of
the subsequent liquidity crisis (Adrian & Shin, 2008b).
In the above we have attempted a brief sketch of some
of the key processes of ‘framing’/‘disentanglement’
through which the CDO product and market was consti-
tuted. The move from an ‘originate to hold’ to an ‘originate
to distribute’ model greatly increased the degree of entan-
glement of ?nancial market participants, creating new and
extended networks of relationships that tied the sub-prime
borrower and lender through to investment banks, insur-
ers, rating agencies, hedge funds, money markets and insti-
tutional and individual investors. The disentanglement of
default risk was achieved, in part, through securitisation
and then tranching; processes that combined then re-di-
vided the pools of assets into different risk/return seg-
ments. Risk was then further disentangled from the
underlying assets with the development of synthetic prod-
ucts using credit default swaps. Common to almost every
link in these now extended networks was the use of mod-
els to construct, rate, risk manage and price the products.
Accounting was indeed the ‘keystone’ to these processes
in its ability to calculate, recognise and thereby expropri-
ate the pro?ts that were being earned at every stage of
the process; pro?ts that were multiplied through the
extensive use of leverage. Up until 2007 these ‘framing’
processes apparently worked well and, as such, could per-
haps be relied upon and taken for granted by participants.
For outsiders to ?nancial markets, beyond the ready access
to cheap credit, these processes were completely invisible.
Over?owing in CDO markets
As is now well known, rising interest rates in late 2006/
early 2007 and the beginnings of a decline in the housing
market as a whole, set the environment for defaults on
mortgages in the USA. Sub-prime borrowers were particu-
larly vulnerable to these changes and by the third-quarter
of 2007 an alarming 42% of subprime adjustable rate and
12% of subprime ?xed rate mortgages had started the pro-
cess of foreclosure (Mortgage Bankers Association, 2008).
2006/2007 vintages of subprime mortgages, in particular,
had higher rates of default than previous years and created
a bleak prospect for 2008 given that some $250bn worth of
these were due to reset in this period (IMF, 2008). These
actual and anticipated levels of default far exceeded the
assumptions of default risk that had been modelled into
CDOs based on subprime mortgages. Further, and critically,
in a declining housing market modelled assumptions of
recovery rates post-default and the degree of correlation
between defaults had also been drastically underestimated
(Ashraft & Schuermann, 2008; Chomsisengphet & Penning-
ton-Cross, 2006; Ryan, 2008; SEC, 2008a). This was the
shock that passed through into ?nancial markets and, sig-
ni?cantly, it was the non-human agents, the models and
accounting, that acted as the messengers. In this context
we can only touch on some of the major shocks of the per-
iod in order to give a ?avour of the nature of the over?ows.
In mid-June 2007, two of Bear Sterns’ mortgage related
hedge funds collapsed. They had been used to house sub-
prime RMBS and CDOs, many of which were structured
using the worst performing 2006 vintage mortgages. Dur-
ing the ?rst four months of 2007 one fund lost 23% of its
value. This was compounded by its borrowings, which at
$6bn were ten times larger that its $600m capital (Econo-
mist, 2007). As a result, the funds were unable to meet calls
from their counterparties for additional margin or
collateral.
In July the rating agencies, who were now in possession
of new default data, started to re-rate CDOs and there fol-
lowed a huge number of downgrades, often by multiple
notches. Moody’s for example, downgraded 252 AAA rated
CDOs sold in 2006/2007, or 22% of the 1155 deals issued in
that period with a combined value of $12bn; the average
magnitude of each downgrade was eight notches. The
meaning for market participants was ambiguous – was
the default prospect so much worse or had the original rat-
ings been faulty, and by implication were the modelling
processes upon which they had relied themselves funda-
mentally ?awed? The volume and severity of these down-
grades pushed market prices even lower, especially for
junior tranche, subprime CDOs. Downgrades on the senior
tranches raised fears that ‘rating sensitive’ investors might
be forced to sell assets.
In August, the world of SIVs began to unravel when
Sachsen LB announced that German savings banks had
provided a Eur 17bn credit facility given the dif?culties it
was facing in re?nancing short term commercial paper
upon which it depended, raising concerns about asset sales
in this $350bn segment of the market. Sachsen LB was
immediately bought to avoid default because of its invest-
ments in US subprime assets. Later in August, London
based SIV Cheyne Finance reported losses and the possible
sale of assets because of similar problems related to short
term money market re?nancing. In October JP Morgan,
Bank of America and Citigroup announced plans for an
$80bn bailout for their SIVs. In November HSBC announced
a bailout of two SIVs of $45bn, whilst in December Citi-
group announced a bailout of $49bn for its six SIVs.
More shocks came when the large investment banks
started reporting large write downs in September 2007.
Between then and early 2008 Merrill Lynch reported
write-downs of $24.5bn, Citigroup of $22bn, UBS $18bn,
HSBC $10bn and Morgan Stanley $9.5bn. The scale of these
losses fed huge falls in their stock market capitalisation,
creating an urgent need either for de-leveraging and credit
rationing or for new capital injections. As seriously, given
the dependence of the synthetic CDO on credit default
swaps, this fed into the spreads in their debt instruments
(Felsenheimer & Gisdakis, 2008, p. 123). These concerns
in relation to CDSs were heightened when in December
and January the ratings agencies began to downgrade their
ratings on the monoline insurers. Their exposure to struc-
tured ?nance was only 30% of their business but hugely
leveraged and losses had begun to eat very rapidly into
their capital. More seriously the downgrades put at risk
the entire $2.4trn of structured ?nance, municipal bonds
and securities that they had guaranteed. The near collapse
of AIG, and the backlash from the collapse of Lehman
Brothers re?ects the dangers of over?ow to the vast CDS
market.
862 J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867
In the light of our above discussion of framing these
over?ows can be understood since, as Callon asserts, ini-
tially they arose from the ways that subprime defaults
fed rapidly through the networks that CDOs had created.
Most obviously, increased defaults often wiped out the
equity and mezzanine tranches and began to eat into what
had been rated as apparently risk remote senior tranches.
Having believed that CDOs allowed them to pass on risks
to others, banks and hedge funds found themselves incur-
ring large and immediate losses that, in turn, created the
need for re-capitalisation. But as described earlier, the syn-
thetic versions of CDOs had also created new ‘market’ and
‘counterparty’ risks that had not been assessed as part of
the rating process. Even without defaults, falling prices
on indices created unrealised mark-to-market losses,
whilst margin calls for increased collateral, or trigger
mechanisms in the highly leveraged CDOs, forced ?re sales,
which only further depressed prices in a potentially self
feeding vicious circle. The reliance of SIVs on short term
commercial paper that could not now be re?nanced cre-
ated an immediate connection to money markets, forcing
up spreads and feeding the growing liquidity crisis. Liquid-
ity facilities offered by the sponsors of such vehicles, as
well as considerations of ‘reputational risk’, brought these
until then invisible liabilities back onto balance sheets fur-
ther squeezing capital, etc. To each of these channels for
over?owing must be added the multiplier effects of lever-
age. Now, however, it was losses rather than pro?ts that
were being multiplied.
A loss of faith in calculating tools
Although such post-hoc tracing of the channels along
which markets over?owed is possibly helpful, it is as
important to observe that, as these over?ows appeared in
reported accounting losses, they were both on a shocking
scale, and for most people completely mysterious. Like
blind men touching different parts of an elephant, or like
a pixelated TV screen, there was no sight of what was gen-
erating these effects. Instead, accounting losses signalled
only that previously reliable frames were no longer pro-
ducing pro?ts but not why. Entity accounting provides part
of the explanation for this since systemic effects were only
visible in the ways these touched an individual institution.
However, it was also the case that some effects were com-
ing from entities that had not even been reported on – the
off balance sheet SPVs and SIVs that were now suddenly
appearing on the balance sheets of their sponsoring banks –
as well as the effects on valuations of the still invisible
but vast set of counterparty risks held in the OTC CDS mar-
ket. That losses could appear unexpectedly in these ways
told market participants that what they had previously ta-
ken as a complete frame had been no such thing. Even
within entity accounting, say in the investment banks,
the reported aggregated data initially did not allow inves-
tors to see suf?cient detail of exposures in relation to par-
ticular activities and asset classes. The variable timing and
‘jumps’ in the values of write downs from quarter to quar-
ter were also shocking, and there was particular investor
concern about level 3, mark-to-model valuations. Investors
feared this could be being used to minimise reported
losses, whilst for entities it was often simply a necessity gi-
ven the collapse of liquid markets (see IMF, 2008, chap. 2).
When all was going well models could be taken as evi-
dence and embodiments of the rationality so prized by ?-
nance, and accounting as no more than a neutral
observer of the fruits of this rationality; both models and
accounting were treated as reliable ‘intermediaries’. With
the crisis, however, both models and accounting came to
be seen as unreliable just as they were most needed; they
were now viewed as possibly distorting ‘mediators’. This
collapse of belief in what had been seen as reliable and
dependable ways of knowing upon which all still de-
pended, was one of the routes through which panic en-
tered the scene and this, in turn, served only to heighten
a different kind of self interest. Freud’s de?nition of panic
seems relevant here: ‘a panic arises if a group becomes dis-
integrated. . .each individual is solicitous only on his own
account, and without any consideration for the rest’ (cited
in Weick, 1993).
Such panic can involve an intensi?cation of calculation
intended to defend institutional and individual self interest
from market turmoil. Unintentionally, it feeds a paranoia
that then tries to run ahead of current facts; not just sub-
prime but the entire market for securitisation has effec-
tively closed; spreads on CDO and CDS indices have
blown out in anticipation of future losses and, perhaps
most consequentially, money markets have frozen as insti-
tutions seek to preserve their own liquidity in the now cer-
tain knowledge that other institutions with whom they
formerly traded face counterparty risks of which they too
are ignorant. Whilst seemingly rational from an individual,
institutional point of view, such extreme self defence is, of
course, self ful?lling (Ghoshal & Moran, 1996; Roberts,
2001). As with the anticipation of future pro?ts, the ra-
tional attempt to protect the institution from future losses
can unintentionally serve to bring these into reality. At the
time of writing, despite governments’ attempts to breathe
life back into inter-bank lending by injecting billions of
dollars into markets, homo economicus still remains in a
state of ‘shocked disbelief’ and is largely closed for
business.
Accounting for self interest
In the above we have followed Callon’s insistence that
self interest should be seen, not as a given and de?ning fea-
ture of the ‘individual’ as it is in economics and ?nance, but
rather as something that is actively constructed. We have
then sought to trace, if only in sketch form, the intense la-
bour of framing/disentanglement through which CDO
products and markets were created and elaborated, only
then to over?ow to produce the current crisis. Within this
account we have also followed Callon’s insistence on the
importance of ‘calculating tools’ as agents or ‘mediators’
within these processes, and, in particular, have sought to
trace some of the effects of mathematical models and
accounting in producing the greed and then fear that have
characterised the rise and fall of the CDO market. Latour’s
distinction between neutral ‘intermediaries’ and distorting
‘mediators’ possibly seems obscure, but is re?ected in the
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 863
intense, post-crisis debates as to the possibly distorting
and misleading effects of both models and fair value.
In relation to models we have already mentioned some
of the erroneous assumptions that have subsequently
come to light. The SEC subsequently summarised these
concisely:
The performance history of the type of subprime mort-
gages that dominated many of the RMBS portfo-
lios. . .had been very short. Further, the performance
history that did exist occurred under very benign eco-
nomic conditions. These conditions included: consistent
high economic growth, interest rates at historic lows,
very low volatility in interest rates and a period where
housing prices increased consistently year over year’
(SEC, 2008a, p. 35).
With the wisdom of hindsight it is obvious that a model
can only be as good as the assumptions that are built into
it, and in this case, assumptions about default rates, and
the way these might feed into a wider housing downturn
that radically changes default correlations, were simply
wrong.
Whilst these wrong assumptions were at the heart of
the distortions that had unwittingly been built into the
models, other investigations suggested more pervasive
problems in the reliance on models. The CRMPG, for exam-
ple, noted a range of problems in risk management pro-
cesses which they suggest ‘struggled to keep up with the
complexities of product design and development’ in a
way that made their hedging strategies ineffective. They
also noted that the reliance of large integrated ?nancial
intermediaries on Value at Risk metrics masked the effects
of ‘correlations between exposures’ both within institu-
tions and in their counterparties. It was also argued that
models simply cannot predict low probability ‘extreme
events’ (Taleb, 2007).
The CRMPG concluded that ‘risk management profes-
sionals and senior management must recognise the limita-
tions of mathematical models’ suggesting that
‘incremental analytical detail must not be allowed to over-
whelm users of the data’ (2008, p. 83). There was arguably
no need to rely on sophisticated models in order to antici-
pate the possibility of highly correlated defaults on sub-
prime mortgages; a moment of thought would have been
suf?cient. Possibly the reassurance offered by complex
modelling had taken the place of thought. But when the
models proved to have been misleading, this in turn led to
‘a collapse in con?dence in a very broad range of structured
product ratings and a collapse in liquidity for such products’
(CRMPG, 2008, p. 53). The dilemma here, however, is intense
for, as Millo and MacKenzie (2008) have recently observed,
the markets simply could not have functioned without the
models and, in this sense, their organisational usefulness
was as signi?cant as their accuracy.
Post the onset of the crisis similar doubts have been
raised about the effects of fair value accounting. Supporters
argued that blaming fair value was like ‘shooting the mes-
senger’ or ‘going to a doctor for a diagnosis and then blam-
ing him for telling you are sick’ (Morgan, 2008). The SEC
investigation reported that, if anything, investors wanted
more transparency (SEC, 2008b, pp. 139–156). Opponents
suggested that the application of fair value was exacerbat-
ing market instability by applying the valuations arising
from sales in abnormal market conditions across all portfo-
lios. Rather than seek to resolve this either/or debate, the
important point here is to recognise that accounting was
indeed an agent in its own right – creating con?dence
(unwarranted at times) through the ‘transparency’ it cre-
ated as well as triggering decisions that fed the crisis. What
we want to observe here is the possibly ‘hyperreal’ interac-
tion of models and fair value accounting.
Macintosh, Shearer, Thornton, and Welker (2000) fol-
lowing Baudrillard de?ne ‘hyperreality’ as ‘the current con-
dition of postmodernity where simulacra are no longer
associated with any real referent and where signs, images,
and models circulate, detached from any real material ob-
jects’ (2000, p. 14). The capacity of CDO structures to ‘dis-
entangle’ risk from the underlying assets and make it
tradeable arguably opened the door to such hyperreality,
with models and accounting then creating a mutually
self-referencing hall of mirrors that was only shattered
when real defaults and correlations started to occur. As
we have seen, the price of a CDO tranche was initially mod-
el derived through the calculation of future cash ?ows on
RMBS, based on multiple assumptions including antici-
pated default and recovery rates and correlation. Subse-
quent valuation was then either model derived for
illiquid products, or marked-to-market against a compara-
ble product or indices. The apparently immediate focus of
fair value on the current ‘exit value’ thereby possibly
masked the way that the price re?ected only the modelled
assumptions of future pro?tability. In this way the success
of the CDO seems, at least with the wisdom of hindsight,
to have been based on no more than the modelled anticipa-
tion of pro?tability – and of course that is indeed precisely
how risk came to have a price. But then real pro?ts and
front loaded bonuses were taken on this basis, creating
incentives for further growth. For example, recent reports
suggest that between 2002 and 2008, the $76bn in net
pro?t of the ?ve largest US investment banks was dwarfed
by $190bn in bonuses, whilst 2008 data suggests that com-
bined net losses of $25.3bn nevertheless yielded $26bn in
bonuses (Lucchetti & Karnitschnig, 2009).
Far from signalling a return to reality, the dynamics of
the credit crisis retain a similarly hyperreal quality, as con-
cerns for market and liquidity risk drove indexed values
below those implied by underlying cash-?ows. As the
IMF fearfully anticipated in October 2007:
A small loss in value can force funds to sell large
amounts of assets as liquidations to meet margin calls
and, simultaneously, their redemptions, increase. Such
‘?re sales’ could lead to a vicious circle of forced sales,
as the widening of spreads forces hedge funds and oth-
ers who mark portfolios to market to post losses, possi-
bly sparking investor withdrawals and further forced
sales’ (2007, p. 20).
What must ?rst be observed in this anticipation is the
repeated use of the word ‘forced’. There is nothing natural
at work here but rather, in the terms of our preceding dis-
864 J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867
cussion, it is the framing of CDO deals – collateral require-
ments, liquidity facilities, regulations, accounting stan-
dards, etc. – that are the ‘forces’ at work, as well as panic
driven calculation. But equally important is the IMF’s rec-
ognition that such a defence of self interest might itself
then produce a vicious self feeding circle. Six months on,
the IMF starkly concluded that: ‘although structured ?-
nance can be bene?cial by allowing risks to be diversi?ed,
some complex and multi-layered products added little eco-
nomic value to the ?nancial system’ (IMF, 2008, p. 54).
Conclusions – the leverage of ‘Moral Hazard’
In seeking to account for self interest in the current
credit crisis this paper has asked two different but related
questions. The ?rst concerns accounting’s role in producing
the self, and the calculation of ‘self’ interest, and here we
must observe that accounting, and associated modelling
processes and practices, have been indispensable allies
for both individuals and institutions. Accounting in its
capacity to recognise (disentangle) pro?ts within a stream
of transactions has provided the motive, means and rea-
lised gains (and now losses) for all concerned. In a recent
speech Christopher Cox, the chairman of the SEC argued
that ‘accounting standards should not be viewed as a ?scal
policy tool to stimulate or moderate economic growth, but
rather as a means of producing neutral and objective mea-
surement of the ?nancial performance of public compa-
nies’ (2008). Such a view re?ects accounting’s public
image as no more than a neutral ‘intermediary’ that simply
passes on the truth to investors and markets. In this paper,
however, we have insisted that accounting and modelling
should be understood as agents in their own right – as
‘mediators’ that in their actual effects fed both the illusions
of rationality (greed) that fuelled market growth, and com-
pounded fear and panic as ‘credit risk’ escaped its pro?t-
able framing and became systemic (FSA, 2009).
This takes us back to the second aspect of accounting for
self interest that we have explored here; Greenspan’s
‘shocked disbelief’ as to how self interest could have failed
so dramatically. The key paradox of the current credit crisis
is that it has been created by the very processes that,
according to ?nance theory and those who marketed, secu-
ritised and tranched CDO products, were supposed to bet-
ter manage risk to the bene?t of all. The paradox is intense
– vast amounts of money have been earned in the process
of creating products and markets that in the end them-
selves ‘manufactured’ risk both for ?nancial markets and
for the ‘real’ economy (Beck, 1992; Giddens, 1999). In this
particular manifestation of the ‘risk society’ the proffered
solution for better risk management – the CDO – itself be-
came the source of risk. The problem of Greenspan’s ‘self
interest’ was not that it was insuf?ciently calculating, but
rather that it was fundamentally confused as to the real
nature of interests which, as Callon insists, and the current
crisis violently illustrates, always lie between rather than
within selves or institutions.
Here we can observe both the absolute limit and
responsibility of accounting, for in framing and possibly
hermetically sealing the calculative mind-set, it effects an
illusory separation between one entity and another, and
between personal agency and market dynamics. Or to re-
turn to Callon’s language, accounting is key in reproducing
a sense of self as ‘isolated – too isolated – and autonomous –
too autonomous’, such that it is largely blind and/or
indifferent to the unintended ‘side effects’ of the calculated
pursuit of self interest. Long before accounting numbers
reached investors for their decision making they were
informing the ever more single minded pursuit of pro?t
by participants, leveraged by pay structures that offered
immediate rewards for the trading of longer term risks.
At an individual level, fair value accounting merely mim-
icked the deal culture that it fed whilst encouraging indi-
viduals to be indifferent to the wider processes of which
they were a part. At an institutional level, accounting’s en-
tity focus encouraged the now obviously erroneous belief
that within ?nancial markets capital could feed endlessly
upon itself in a permanently expanding universe of pro?t-
able deals.
As noted earlier, the striking difference between a tradi-
tional lending relationship and the ‘originate to distribute’
model was the complex web of inter-institutional net-
works and relationships (inter-ests) that the latter created.
Signi?cantly, these inter-ests were rendered largely invisi-
ble by accounting. As losses started to ?ow so too did
blame and, on the back of this, there followed a host of
investigations which sought to go beyond the numbers to
the relational processes that had generated them. Through
these investigations it became evident that not only pro?ts
but also ‘moral hazard’ had been leveraged through the
network of market relations. Whilst in traditional dyadic
lending relationships the calculation of self interest in-
cluded a concern for the other’s ability to pay, with the
new model such concern was apparently judged to be
unnecessary. Evidence has emerged of both predatory bor-
rowing and lending practices (Ornstein, Tallman, & Hola-
han, 2006; Shiller, 2008). Signi?cantly, Mian and Su?
(2008) found that the incidence of default on subprime
mortgages was higher where these were securitised, sug-
gesting that with the originate to distribute model there
was little perceived incentive to monitor or appraise the
credit risk of individual borrowers since the risk was al-
most immediately passed on, earning an upfront commis-
sion. Similar ‘moral hazard’ applied in the banks and
investment banks since they too planed to pass on the risk
to other investors. Hedge funds were also, it appeared,
working on extraordinary fee structures earned in part
through short selling practices that arguably further feed
market volatility (Lo, 2008, chap. 10).
The rating agencies were also discovered, after the
event, to be caught up in all sorts of con?icts of interest
and poor organisational practices; they were short staffed
and relied on historical data of corporate bond defaults to
rate products that had no history. One staff email obtained
by the SEC noted concerns that the ?rm’s model did not
capture half of the deal’s risk, and that ‘it could be struc-
tured by cows and we would rate it’ (SEC, 2008a). Coffee
(2008) further observed that rating agency dependence
on a handful of originators and ‘the issuer pays’ fee struc-
ture fundamentally compromised their assumed indepen-
dence in the rating process. Finally, we should mention
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 865
regulation itself, which in the context of ?nancial markets
seems to take for granted the thought that all regulation
will itself furnish new opportunities for regulatory arbi-
trage. As Power has recently observed, regulatory style
has come to rely increasingly on the ‘self governing capac-
ities of organizations’, and this has set up a tension be-
tween public conformity with the normative ideal
embodied in audited compliance, and practice where par-
ticipants ‘realise they may be participating in the costly
construction of an illusion or fantasy of control’ (2007, p.
199). In the context of the credit crisis, reassurance seems
to have been drawn from the routines of risk management,
which nevertheless failed to identify and manage substan-
tive sources of institutional and inter-institutional risk.
So the tension that becomes visible here is between in-
ter-dependencies (inter-ests) that were systemic and the
calculations of individual and institutional ‘self’ interest
that were seemingly indifferent to these. The real hazard,
however, was not just the lack of care exercised in the bor-
rowing, lending, tranching, rating, leveraging, insuring and
buying of these ‘pass through’ products. It was also that, in
reality, they turned out to be anything but ‘pass through’
since the unanticipated consequences of this carelessness
then revisited agents at every point in the networks. The
error of ‘self’ interest lies in the way in which it imagines
the self as separate and separable from others. Its calcula-
tion conceives of others only in instrumental terms, and so
it remains blind or simply indifferent to the consequences
of its conduct for others. The moment of panic signi?es the
rediscovery of the self as vulnerable and dependent, but
this reality is instantly ?ed from in a renewed and often
self-defeating attempt to defend the self from others.
Whilst there is now an intense effort to de?ne new trans-
parencies that will effectively re-frame the pursuit of self
interest, arguably what are also needed are new forms of
accounting that make interdependencies visible, and new
‘intelligent’ processes of accountability that secure the
relational nature of interests.
Acknowledgements
We would like to thank Tyrone Carlin, Graeme Dean,
David Johnstone, Stewart Jones, Jan Mouritsen and Sue
Newberry for their comments on earlier drafts of this
paper.
References
Adrian, T., & Shin, H. (2008a). Liquidity and leverage. Federal Reserve Bank
of New York staff reports no. 328 (May).
Adrian, T., & Shin, H. (2008b). Liquidity and ?nancial contagion. Financial
stability review. Banque de France (11 February).
Ashraft, A. & Schuermann, T. (2008). Understanding the securitization of
subprime mortgage credit. Federal Reserve Bank of New York staff
reports no. 318 (March).
Augar, P. (2006). The greed merchants: How the investment banks played the
free market game. Penguin: Harmondsworth.
Beck, U. (1992). Risk society: Towards a new modernity. New Delhi: Sage.
Blackburn, R. (2008). The subprime crisis. New Left Review, March/April,
63–106.
Callon, M. (1986). Some elements of a sociology of translation:
Domestication of the scallops and the ?shermen of St. Brieuc Bay.
In J. Law (Ed.), Power, action and belief: A new sociology of knowledge?
(pp 196–223). London: Routledge.
Callon, M. (1998). The laws of the markets. Sociological review monograph.
Oxford: Blackwell.
Callon, M., Millo, Y., & Muniesa, F. (2007). Market devices. Sociological
review monograph. Oxford: Blackwell.
Callon, M., & Muniesa, F. (2005). Economic markets as calculative
collective devices. Organization Studies, 26(8), 1229–1250.
Chomsisengphet, S., & Pennington-Cross, A. (2006). The evolution of the
subprime mortgage market. Federal Reserve Bank of St. Louis review
(January/February).
Coffee, J. (2008). The role and impact of credit rating agencies on the
subprime credit markets. In Testimony before the Senate Banking
Committee (September 26).
Counterparty Risk Management Policy Group III (2008). Containing
systemic risk: The road to reform (August).
Cox, C. (2008). In Remarks before the AICPA national conference on current
SEC and PCAOB developments (December). <http://www.sec.gov/news/
speech/2008/spch120808cc.htm>.
Das, S. (2006). Traders, guns & money. London: Prentice Hall.
Economist (2007). Bearish turns – The subprime meltdown continued (June
23).
Engelen, E., Ertuk, I., Froud, J., Leaver, A., & Williams, K., (2008). Financial
innovation: Frame, conjuncture and bricolage. CRESC working paper
no. 59.
Felsenheimer, J., & Gisdakis, P. (2008). Credit crises: From tainted loans to
global ?nancial meltdown. Weinham: Wiley-Vch.
Fender, I., & Kiff, J. (2004). CDO rating methodology: Some thoughts on
model risk and its implications. Bank for International settlements
working paper no. 163 (November).
Financial Accounting Standards Board (FASB) (2006). SFAS 157. Fair value
measurements (September).
FSA (Financial Services Authority) (2009). The Turner Review: A regulatory
response to the global banking crisis, March. <http://www.fsa.gov.uk-
pubs-other-turner_review.pdf>.
Ghoshal, S., & Moran, P. (1996). Bad for practice. A critique of
transaction cost theory. Academy of Management Review, 21(1),
13–47.
Giddens, A. (1999). Risk and responsibility. The Modern Law Review, 62(1),
1–10.
Greenspan, A. (2005). Risk transfer and ?nancial stability. In Remarks at
the Federal Reserve Bank of Chicago’s 41st annual conference on bank
structure , May 5, Chicago, IL.
Greenspan, A. (2008). Transcript of testimony to House
Oversight Committee. <http://oversight.house.gov/documents/
20081024163819.pdf>.
IMF (2007). Global ?nancial stability report: Financial market turbulence –
Causes, consequences and policies (October). <http://www.imf.org-
External-Pubs-FT-GFSR-2007-02-pdf-text.pdf>.
IMF (2008). Global ?nancial stability report: Containing systemic risks and
restoring ?nancial soundness (April). <http://www.imf.org-external-
pubs-ft-gfsr-2008–1-pdf-text.pdf>.
International Accounting Standards Board (2005). IAS 39. Financial
instruments: Recognition and measurement.
Joint Forum on Credit Risk Transfer (2004). Bank for International
Settlements (March).
Jones, S., & Peat, M. (2008). Credit derivatives: Current practices and
controversies. In S. Jones & D. Henscher (Eds.), Advances in credit risk
modelling and corporate bankruptcy prediction (pp. 207–241). New
York: Cambridge University Press.
Latour, B. (2005). Reassembling the social: An introduction to actor-network-
theory. Oxford: Clarendon.
Lo, A. (2008). Hedge funds: An analytic perspective. Princeton NJ: Princeton
University Press.
Lucchetti, A., & Karnitschnig, M. (2009). Wall street workers brace for
more pay shrinkage. Wall Street Journal (February 2).
Macintosh, N., Shearer, T., Thornton, D., & Welker, M. (2000). Accounting
as simulacrum and hyperreality: Perspectives on income and capital.
Accounting, Organizations and Society, 25(1), 13–50.
MacKenzie, D. (2006). An engine, not a camera: How ?nancial models shape
markets. Cambridge MA: MIT Press.
MacKenzie, D., Muniesa, F., & Siu, L. (2007). Do economists make markets?
On the performativity of economics. Princeton: Princeton University
Press.
Markowitz, H. (1959). Portfolio selection: Ef?cient diversi?cation of
investments. New York: John Wiley and Sons.
Mian, A., & Su?, A., (2008). The consequences of mortgage credit expansion:
Evidence from the 2007 mortgage default crisis (May). <http://ssrn.com/
abstract=1072304>.
Miller, P. (1998). The margins of accounting. European Accounting Review,
7, 605–621.
866 J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867
Miller, P., & O’Leary, T. (2007). Mediating instruments and making
markets: Capital budgeting, science and the economy. Accounting,
Organizations and Society, 32(7/8), 701–734.
Millo, Y., & MacKenzie, D. (2008). The usefulness of inaccurate models:
Towards an understanding of the emergence of ?nancial risk
management. Accounting, Organizations and Society, doi:10.1016/
j.aos.2008.10.002.
Morgan, J.P. (2008). Shooting the messenger. Global Equity Research
(September).
Morgan, M., & Morrison, M. (1999). Models as mediators: Perspectives on
natural and social science. Cambridge: Cambridge University Press.
Morris, C. (2008). The trillion dollar meltdown. New York: Perseus Books.
Mortgage Bankers Association (2008). Delinquencies and foreclosures
increase in latest MBA national deliquency survey (March 3). <http://
www.mortgagebankers.org/NewsandMedia/PressCenter/
60619.htm>.
Ornstein, S., Tallman, D., & Holahan, J. (2006). Predatory lending and the
secondary market. Journal of Structured Finance, Fall, 54–64.
Pleven, L., & Craig, S. (2008). Deal fees under ?re amid mortgage crisis –
Guaranteed rewards of bankers, middlemen are in the spotlight. Wall
Street Journal (January 17).
Power, M. (2007). Organized uncertainty: Designing a world of risk
management. Oxford: Oxford University Press.
Roberts, J. (2001). Trust and control in Anglo-American systems of
corporate governance. The individualising and socialising effects of
processes of accountability. Human Relations, 54(12), 1547–1572.
Ryan, S. (2008). Accounting in and for the subprime crisis. Stern School of
Business, New York University. <http://ssrn.com/abstract=1115323>.
SEC (2008a). Summary report of issues identi?ed in the commission staff’s
examination of select credit rating agencies (July).
SEC (2008b). Report and recommendations pursuant to section 133 of the
emergency economic stabilization act of 2008: Study of mark-to-market
accounting (December).
SIFMA (2007). Securities Industry and Financial Markets Association: Global
CDO Market Issuance Data. <http://archives1.sifma.org/assets/?les/
SIFMA_CDOIssuanceData2007q1.pdf>.
Shiller, R. (2008). The subprime solution: How today’s global ?nancial crisis
happened and what to do about it. Princeton: Princeton University
Press.
Taleb, N. (2007). The black swan: The impact of the highly improbable. New
York: Random House.
Weick, K. (1993). The collapse of sense-making in organisations:
The Mann Gulch disaster. Administrative Science Quarterly, 38,
628–652.
Whittington, G. (2008). Fair value and the IASB/FASB conceptual
framework project: An alternative view. Abacus, 44(2), 139–168.
J. Roberts, M. Jones / Accounting, Organizations and Society 34 (2009) 856–867 867
doc_875233860.pdf