Page 1 of 2 Fraud and the subprime bubble
By George Pugh
The subprime crisis is ending in massive losses, bare balance sheets and wilted
capitalizations: a true rite of spring. Early theories blamed the crash on
corporate greed, bad product design, poor analytics, rating agency failures,
and fraud, in various combinations.
The US Federal Reserve Board is now asking for tighter regulation of the
brokerage industry, while in truth fraud started with the mortgage originators,
which helped cause the collapse of the market, while the firms that are now
being blamed are actually the victims rather than the perpetrators of the
fraud.
The Sarbanes Oxley Act of 2002, with its supporting regulations, was supposed
to curb fraud at the company level by improving
internal controls and emphasizing management responsibility for accurate
financial reporting: in that aspect it is a signal failure.
The subprime mortgage-based CDO
Germany's President Horst Koehler has blamed the subprime crisis on highly
leveraged investments in complex financial instruments. Actually the subprime
instruments were the result of increased investor sophistication. The firms
held product inventories not for speculation but rather as a service to their
customers for collateralized debt obligations (CDOs) and to provide general
liquidity in a counter-party market.
CDOs were first issued in the late 1980s by the now defunct Drexel Burnham
Lambert of junk-bond fame. The CDO is an entity that holds collateral and sells
cash flows, with varying seniorities organized into tranches broken down by
asset-quality risk. They are popular because they can be large, meet the demand
for cash-flow only products and are more profitable for the issuer than
traditional debt instruments: they are a legitimate product filling an actual
demand rather than being a speculative tool.
Banks have long issued CDOs collateralized by mortgage-backed securities (MBS)
containing prime mortgages, that is, in the US, ones eligible for Fannie Mae,
Freddie Mac and Ginnie Mae guarantees. Subprime mortgages have no such
protection, so for that lack alone they are at once more risky because of the
range of borrower credit quality.
Many have terms that give the borrower the ability to pick a payment size -
full (principle and interest), interest only or some mix of the two. In
addition, many are adjustable-rate mortgages (ARMs) that switch from a fixed to
variable rate after the first two years, called the "2-28" option. The variable
rate is set at say 5% over the 12-month London interbank offered rate after the
initial fixed period. There is generally a very high cap, say 15% on the rate.
Developing the common risk profile needed to create CDOs is both expensive and
difficult for these assets because of the wide range of terms and uninsured
asset risk (mortgages).
CDOs are designed to provide separate cash flows, risk defined by the tranche
from which each is drawn. The complicating factor is that the underlying
instruments, MBSs funded by subprime mortgages, do not have easily estimated
cash flows. [1]
Doing the math
Despite initial reactions in the news, the firms in question managed their
risks extremely well, considering the unanticipated element of fraud, which
thanks to the Sarbanes Oxley reforms they had no rational expectation of
encountering. The best single assessment of the firms' response quality is
found in "Observations on Risk Management Practices during the Recent Market
Turbulence" by the Senior Supervisors Group of the Bank of International
Settlements, dated March 6, 2008. Though the study did not include the whole
universe of firms involved, it covered 11 of the largest banking and security
firms, along with an additional five present at the round table discussion,
which were all there to the end.
This report outlines supervisors'
observations on the risk-management practices that may have enabled some firms
to weather the financial market turmoil better than others from summer 2007
through year-end 2007. Specifically, supervisors focused on practices related
to the following:
The role of senior management oversight in assessing and responding to the
changing risk landscape;
The effectiveness of market- and credit-risk management practices in
understanding and managing the risks in retained or traded exposures, and
The effectiveness of each firm's liquidity-risk management practices in
assessing its vulnerability to that risk in a stressed environment and taking
appropriate action. [2]
The key was management quality, which in this case turned on the
ability to properly allocate capital and a solid understanding of market
conditions.
Firms that managed their funding liquidity needs more
successfully through year-end 2007 encouraged individual business lines to
assess and communicate their likely needs for funding to the treasury function
and to price those internal claims on liquidity appropriately in light of
actual market conditions. [3]
These efforts were strongly
related to intellectual flexibility and timely decisions. The more-successful
firms were willing to revisit their assumptions and sometimes revert to
simpler, risk-reducing measures.
In light of some firms' uncertainty
about the accuracy of assumptions underlying their risk measures during the
current period of turbulence, several firms cited the usefulness of revisiting
simple notional limits to highlight potential concentrations of risk. [4]
These firms faired better than those that relied on assumptions that might not
have been true. Some, by holding to their initial assumptions, finally lost
sight of their total risk and did not do as well as others for that reason.
There were other problems that required adjustment, which related to the tools
used. One of the more important was related to value at risk (VaR):
Nevertheless,
firms indicated that VaR, as a backward-looking measure of risk dependent on
historical data, may never fully capture severe shocks that exceed recent or
historical norms. ... Most of these exceptions were generated by much higher
market volatility and realized asset price correlation than the historical data
series implied. [5]
Data quality and an over-reliance on
historical activity are always problematic. The data may have been smoothed by
removing outliers, or not have taken in enough data to include extremes. Put
another way, VaR data does not necessary reflect the current situation because
of the number of prior periods used (generally five years) hiding the true
level of volatility at the time.
The firms also had to cope with the fact there was a confusion between the risk
of CDOs and those of debt instruments, which were assumed to be equivalent and
weren't.
In general, the construction of CDOs tends to make them more
sensitive to systematic shocks. In contrast, highly rated corporate debt
issuances tend to be more sensitive to "idiosyncratic" risk, or risks
associated with characteristics specific to the corporation that issued the
debt. [6]
The second problem is more profound: firms made
assumptions about their CDOs that simply were not true. The only thing they
have in common is the rating. Corporate debt is unique, being tied to one set
of unique assets. The CDOs use fungible collateral, making them risky as a
class. That is, if a corporation has problems, its debt value and rating may
fall, but others in the class may not: it is company specific. For CDOs, if
anything happens to the perception of the collateral, the shock will be
universal. Management simply could not or would not see this simple difference
and acted as if it didn’t exist because their models did not take the
difference into account or did not recognize it.
Though firms varied in how effectively the used their risk-management tools,
good management simply used very similar ones better. In fact, considering that
the tools handled the fraudulent collateral (mortgages) which was in no way
anticipated, it can be said that any additional regulation at this level would
be nugatory, except as political display. Mortgage fraud at this level was the
sign of serious problems not of their making.
The subprime mortgage as an asset
The subprime CDO was a remarkable new instrument. It was not until 2001 that
any CDO pricing became cost effective, thanks to the adoption of Gaussian
cupola models developed by David Xi. The models are simpler and cheaper to use
than the Monte Carlo simulation method. The math was by no means suspect and it
will be of note that the tool itself came under no criticism. [7]
What made subprime-backed MBSs more problematic was the lack of guarantees
concerning the underlying mortgage obligations by the US government. In this
instance, the firms decided to purchase ratings for these instruments from the
rating agencies. It must be noted that Moody's and the other rating companies
were doing their risk assessment on the originator documents, and not on the
underlying facts represented. Moody's structured-finance group grew to account
for about 43% of Moody's revenue in 2006, up from 28% in 1998. By 2006, the
firm had more revenue from structured finance - US$881 million - than its
entire revenue had been in 2001. [8]
More to the point, Moody's in particular was way overcommitted to the line of
business and was subject to customer pressure very early on.
Consider a
Bank of America mortgage deal in early 2001. As in most such deals, the vast
majority of the securities based on the pool of mortgages would be rated
triple-A. The question was how big a chunk would be rated lower - paying a
higher interest rate and bearing the brunt of any defaults that occurred.
A rating committee at Moody's voted to require that the issuer put about 4.25%
of the deal's value in the lower-rated section, to provide extra protection for
buyers of the top-rated section. But after Bank of America complained and said
it might go with a different rating firm, Moody's reduced the size of the
lower-rated chunk slightly, saving the issuer some interest costs, according to
people with knowledge of the matter. [9]
The problem was
twofold: first, Moody's regularized its business before the subprime market in
all its forms really began to grow, and they were rating the terms and more
importantly the component tranches rather than the mortgages that backed the
MBSs. Lastly, there was fraud, and Brian Clarkson, the architect of Moody's
role in this market, offers the following:
We knew that there was
fraud. We may have thought it was X; [it turns out] it was X to the 10th power.
We knew the risks were increasing, so we increased the protection. It was
completely dwarfed. We were preparing for a rainstorm and it was a tsunami. We
saw the increased risk, but we didn't see what appears to be an 18-month period
where anything went. I hate going through this because it sounds defensive, but
the fact is that there were people who were supposed to be doing due diligence
on this who just didn't do it. [10]
The problem was that they
were simply not there to find fraud nor were they looking for it, simply
because that was not what they
Head
Office: Unit B, 16/F, Li Dong Building, No. 9 Li Yuen Street East,
Central, Hong Kong Thailand Bureau:
11/13 Petchkasem Road, Hua Hin, Prachuab Kirikhan, Thailand 77110
