Sunday, September 4, 2011


While the CDO valuation project was underway, Goldman senior executive Thomas
Montag asked Daniel Sparks for an estimate of how much the firm would need to write down the
value of its CDO assets. Mr. Sparks responded that “the base case from traders is down [$]382
[million].” He also wrote: “I think we should take the write-down, but market [the CDO
securities] at much higher levels.”1574 Another Goldman senior executive, Harvey Schwartz,
expressed concern about selling clients CDO securities at one price and then immediately
devaluing them: “[D]on’t think we can trade this with our clients andf [sic] then mark them down
dramatically the next day.”1575 At the same time, Goldman’s Chief Credit Risk Officer Craig
Broderick told his staff to anticipate deep markdowns and highlighted the need to identify clients
that might suffer financial difficulty if Goldman devalued their CDO securities and demanded
they post more cash collateral.
On May 20, 2007, the Gameplan results were summarized in an internal presentation.1576
It projected that Goldman would have to take from $248 to $440 million in writedowns on unsold
CDO securities and warehouse assets, making it clear to Goldman executives that its CDO assets
were losing value rapidly.1577 In several drafts of the presentation, the Mortgage Department had
also written that Goldman’s CDOs were expected “to underperform,” but that statement was
removed from the final presentation given to senior executives.1578
Targeted Sales. The Gameplan also recommended a two-pronged approach to selling the
firm’s remaining CDO assets. First, it recommended transfer of the CDO warehouse assets to the
Mortgage Department’s SPG Trading Desk for further valuation and sale. Second, it
recommended that the Mortgage Department use a “targeted” approach to sell off the existing and
5/19/2007 “Mortgages CDO Origination – Retained Positions 1579 & Warehouse Collateral, May 2007,” Goldman
presentation, GS MBS-E-010951926; 5/20/2007 email from Lee Alexander to Daniel Sparks, Donald Mullen, Lester
Brafman, Michael Kaprelian, “Viniar Presentation - Updated,” GS MBS-E-010965211 (attached file “Mortgages
V4.ppt,” “Mortgages Department, May 2007,” GS MBS-E-010965212).
1580 See discussions of Hudson, Anderson, Timberwolf, and Abacus sales efforts, below.
new securities from Goldman-originated CDOs, naming four hedge funds as the primary targets
and providing a list of another 35 clients as secondary targets.1579
At the same time, Mr. Sparks named David Lehman, a commercial mortgage backed
securities trader, as the new head of the CDO Origination Desk. Shortly thereafter, Goldman
dismantled the CDO Origination Desk and moved all remaining CDO securities to the SPG
Trading Desk, where he was based. The SPG Trading Desk, which was a secondary trading desk
and had little experience with underwriting, assumed responsibility for marketing the remaining
unsold Goldman-originated CDO securities. The SPG Trading Desk’s lack of underwriting
experience meant that it was less familiar with the obligations of underwriters and placement
agents to disclose all material adverse interests to potential investors.
The SPG Trading Desk worked with Goldman’s sales force to market the CDO securities.
Goldman employed “hard sell” tactics, repeatedly urging its sales force to sell the CDO securities
and target clients with limited CDO familiarity.1580 After trying the Gameplan’s “targeted” client
approach during May, June, and July 2007, the Mortgage Department switched back to issuing
sales directives or “axes” to its entire sales force, including sales offices abroad. Axes on CDOs
generally went out at weekly or monthly intervals, identified specific CDO securities as top sales
priorities, and offered additional financial incentives for selling them. Despite the CDOs’
declining value, the sales force succeeded in selling some of the CDO securities, primarily to
clients in Europe, Asia, Australia, and the Middle East, but was unable to sell all of them.
The four CDOs that the Subcommittee examined illustrate a variety of conflict of interest
issues related to how Goldman designed, marketed, and administered them.
Hudson Mezzanine 2006-1. Hudson Mezzanine 2006-1 (Hudson 1) was a $2 billion
synthetic CDO comprised of $1.2 billion in ABX assets from Goldman’s own inventory, and $800
million in single name CDS contracts on subprime RMBS and CDO securities that Goldman
wanted to short. It was called a “mezzanine” CDO, because the referenced RMBS securities
carried the riskier credit ratings of BBB or BBB-. Goldman used the CDO to transfer the risk
associated with its ABX assets to investors that bought Hudson 1 securities. Goldman also took
100% of the short side of the CDO, which meant that it would profit if any of the Hudson
securities lost value. In addition, Goldman exercised complete control over the CDO by playing
virtually every key role in its establishment and administration, including the roles of underwriter,
initial purchaser of the issued securities, senior swap counterparty, credit protection buyer,
collateral put provider, and liquidation agent.
10/2006 Goldman Sachs report, “Hudson Mezzanine Funding, 1581 2006-1, LTD.,” at 4, GS MBS-E-009546963,
Hearing Exhibit 4/27-87.
1582 Id.
1583 For more information on this pricing mismatch, see the Hudson discussion, below.
1584 12/3/2006 Hudson Mezzanine 2006-1, LTD. Offering Circular, GS MBS- E-021821196 at 251 (emphasis
Goldman began marketing Hudson 1 securities in October 2006, soliciting clients to buy
Hudson securities. It did not fully disclose to potential investors material facts related to
Goldman’s investment interests, the source of the CDO’s assets, and their pricing. The Hudson 1
marketing materials stated prominently, for example, that Goldman’s interests were “aligned” with
investors, because Goldman was buying a portion of the Hudson 1 equity tranche.1581 In its
marketing materials, Goldman did not mention that it was also shorting all $2 billion of Hudson’s
assets – an investment that far outweighed its $6 million equity share and which was directly
adverse to the interests of prospective investors. In addition, the marketing materials stated that
Hudson 1’s assets were “sourced from the Street” and that it was “not a balance sheet CDO.”1582
However, $1.2 billion of the Hudson assets had been selected solely to transfer risk from ABX
assets in Goldman’s own inventory.
Goldman also did not disclose in the materials that it had priced the assets without using
any actual third party sales. The absence of arm’s length pricing was significant, because the
Hudson CDO was designed to short the ABX Index using single name RMBS securities, and there
was a pricing mismatch between the two types of assets.1583 Goldman not only determined the
pricing for the RMBS securities purchased by Hudson 1, but retained the profit from the pricing
differential. The marketing materials did not inform investors of Goldman’s role in the pricing, the
pricing methodology used, or the gain it afforded to Goldman. In addition, the marketing materials
stated Hudson 1 was “not a balance sheet” CDO, without disclosing that Hudson had been
designed from its inception to remove substantial risk from Goldman’s balance sheet.
The Hudson 1 Offering Circular contained language that may have also misled investors
about Goldman’s true investment interest in the CDO. The Offering Circular stated:
“[Goldman Sachs International] and/or any of its affiliates may invest and/or deal, for their
own respective accounts for which they have investment discretion, in securities or in other
interests in the Reference Entities, in obligations of the Reference Entities or in the obligors
in respect of any Reference Obligations or Collateral Securities ... , or in credit default
swaps ... , total return swaps or other instruments enabling credit and/or other risks to be
traded that are linked to one or more Investments.”1584
This provision seems to inform investors that Goldman “may invest” for its own account in the
CDO’s securities, reference obligations, or CDS contracts, while withholding the fact that, by the
time the Offering Circular had been drafted, Goldman had already determined to take 100% of the
short position in the CDO, an investment which was directly adverse to the interests of Hudson
securities investors.
2/6/2008 email from John Pearce to Michael P 1585 etrick, HUD-CDO-00005146.
1586 2/7/2008 email from John Pearce to David Lehman, HUD-CDO-00005147.
1587 In April 2007, the month after Anderson issued its securities, New Century announced it would have to restate
its earnings. It declared bankruptcy soon after. For more information about New Century, see Chapter IV, section
1588 See, e.g., 3/2007 Goldman internal email, GS MBS-E-002146861, Hearing Exhibit 4/27-77 (“I recommend
putting back 26% of the pool....if possible.”). See also 2/2/2007 email from Matthew Nichols to Kevin Gasvoda, GS
MBS-E-005556331 (“NC is running a 10% drop rate [due diligence drop] at ~6 points / drop and 4% EPD rate at
close to 20 points.”). 2/8/2007 email from John Cassidy to Joseph Ozment, others, GS MBS-E-002045021 (“Given
the current state of the company I am no longer comfortable with the practice of taking loans with trailing docs . . .
that we need in order to conduct compliance testing. . .”).
Once Hudson issued its securities, Goldman placed a priority on selling them, and delayed
the issuance of a CDO on behalf of another client in order to facilitate Hudson sales. Goldman
sales representatives reported that clients expressed skepticism regarding the quality of the Hudson
assets, but Goldman continued to promote the sale of the CDO.
In 2008, as Hudson’s assets lost value and received rating downgrades from the credit
rating agencies, Goldman, in its role as liquidation agent, was tasked with selling those assets to
limit losses to the long investors. The major Hudson investor, Morgan Stanley, pressed Goldman
to do just that. Goldman, however, delayed selling the assets for months. As the assets dropped in
value, Goldman’s short position increased in value. Morgan Stanley’s representative reported to a
colleague that when Goldman rejected the firm’s request to sell the poorly performing Hudson
assets, “I broke my phone.” 1585 He also sent an email to the head of Goldman’s CDO Desk saying:
“[O]ne day I hope I get the real reason why you are doing this to me.”1586 Morgan Stanley lost
nearly $960 million on its Hudson investment.
Anderson Mezzanine 2007-1. Anderson Mezzanine 2007-1 (Anderson) was another
synthetic CDO referencing BBB and BBB- rated subprime RMBS securities. It was issued in
March 2007. Among other roles, Goldman served as the CDO’s placement agent, initial purchaser,
collateral put provider, and liquidation agent. Goldman hired another firm, a New York hedge
fund founded by former Goldman employees, GSC Partners, to act as the collateral manager.
Goldman took a short position on approximately 40% of the $305 million in assets underlying
Anderson referenced a number of poor quality assets. Those assets had been selected by
GSC Partners, with the approval of Goldman. Over 45% of the referenced subprime RMBS
securities contained mortgages originated by New Century, a subprime lender known within the
industry, including Goldman, for issuing poor quality loans and which was experiencing financial
problems while Anderson was being structured and marketed.1587 Inside Goldman, staff were
aware of New Century’s problems and were taking action to return substantial numbers of
substandard loans purchased from New Century and demand repayment for them.1588 Other assets
in the Anderson CDO were also performing poorly, and at one point, Goldman personnel estimated
3/16/2007 GSI Risk Committee Memo, “GSI Warehousing for Structured 1589 Product CDOs,” GS MBS-E-
1590 3/12/2007 email from Robert Black to Matthew Bieber and others, GS MBS-E-000898037.
1591 See Goldman response to Subcommittee QFR at PSI_QFR_GS0223 and PSI_QFR_GS0235.
1592 Among other assets, Goldman sold $15 million in certain Abacus securities to the Timberwolf CDO. Goldman
held 100% of the short side of the relevant Abacus CDO. Its sale of the Abacus securities meant that Goldman held
the short side of those assets, while Timberwolf investors took the long side.
its warehouse assets had fallen in value by $22 million.1589 Due to the asset quality problems, the
Mortgage Department head, Daniel Sparks, initially decided to cancel Anderson, but later changed
his mind and decided to market the CDO as quickly as possible, using the $305 million in assets
already in its warehouse account, rather than wait to accumulate all of the $500 million in assets
initially planned.
When Anderson issued its securities in March 2007, Goldman placed a high priority on
selling them, even delaying another CDO – Abacus 2007-AC1 which was being organized at the
request of the Paulson hedge fund – to allow its sales force to concentrate on promoting Anderson.
Potential investors raised questions about the quality of its underlying assets, especially the New
Century loans, and Goldman provided its sales representatives with talking points to dispel
concerns about the New Century assets. When one client asked how Goldman had gotten
“comfortable” with the New Century loans, Goldman did not disclose to the client its own negative
views of New Century loans or that it had 40% of the short side of the CDO.
Goldman marketed Anderson securities to a number of its clients, including pension funds,
and recommended using Anderson securities as collateral security in other CDOs.1590 In the end,
Goldman sold only $102 million or about one third of the Anderson securities.1591 Seven months
after the securities were issued, they suffered their first credit rating downgrade. Currently, all of
the Anderson securities have been reduced to junk status, and the Anderson investors have lost
virtually their entire investments.
Timberwolf I CDO. Timberwolf I was a $1 billion hybrid CDO2 transaction that
referenced single-A rated securities from other CDOs. Those CDO securities referenced, in turn,
RMBS securities carrying lower credit ratings, primarily BBB. Altogether, Timberwolf referenced
56 unique CDO securities that had over 4,500 unique underlying assets. Goldman served as the
CDO’s placement agent, initial purchaser, collateral put provider, and liquidation agent. It also
hired a hedge fund with former Goldman employees, Greywolf Capital Management, to act as the
collateral manager. Greywolf selected the CDO’s assets, with Goldman’s approval. Goldman took
a short position on approximately 36% of the $1 billion in assets underlying Timberwolf.1592
Timberwolf’s securities began losing value almost as soon as they were purchased. In
February 2007, Goldman’s Mortgage Department head told a senior executive that Timberwolf
was one of two deals “to worry about.” He also wrote that the assets in the Timberwolf warehouse
account had declined so much in value that they had already exhausted Greywolf’s responsibility to
pay a portion of any warehouse losses, and any additional losses would be Goldman’s exclusive
2/26/2007 email between Tom Montag and Daniel S 1593 parks, GS MBS-E-019164806.
1594 3/2007 Goldman internal email chain, GS MBS-E-001800634.
1595 See, e.g., 6/2007-8/2007 Goldman internal emails, “Timberwolf Sales Efforts,” Hearing Exhibit 4/27-166;
3/2007 Goldman internal email, “GS Syndicate Structured Product CDO Axes,” Hearing Exhibit 4/27-100.
1596 4/19/2007 email chain between Daniel Sparks and Bunty Bohra, GS MBS-E-010539324, Hearing Exhibit 4/27-
102 (Mr. Bohra responds, “[w]e have done that with Timberwolf already.”).
1597 5/14/2007 email from Edwin Chin, GS MBS-E-012553986.
1598 6/2007 Goldman internal email to Daniel Sparks, Hearing Exhibit 4/27-105.
obligation.1593 Goldman rushed Timberwolf to market, and it closed on March 27, 2007,
approximately six weeks ahead of schedule.1594 Almost as soon as the Timberwolf securities were
issued, they too began to lose value.
Despite doubts about its performance and asset quality, Goldman engaged in an aggressive
campaign to sell the Timberwolf securities. As part of its tactics, Mr. Lehman instructed Goldman
personnel not to provide written information to investors about how Goldman was valuing or
pricing the Timberwolf securities, and its sales force offered no additional assistance to potential
investors trying to evaluate the 4,500 underlying assets. Mr. Sparks and Mr. Lehman sent out
numerous sales directives or “axes” to the Goldman sales force, stressing that Timberwolf was a
priority for the firm.1595 In April, Mr. Sparks suggested issuing “ginormous” sales credits to any
salesperson who sold Timberwolf securities, only to find out that large sales credits had already
been offered.1596 In May, while Goldman was internally lowering the value of Timberwolf, it
continued to sell the securities at a much higher price than the company knew it was worth. At one
point, a member of the SPG Trading Desk issued an email to clients and investors, advising them
that the market was rebounding and the downturn was “already a distant memory.”1597 Goldman
also began targeting Timberwolf sales to “non-traditional” buyers and those with little CDO
familiarity, such as increasing its marketing efforts in Europe and Asia.
On June 18, 2007, Goldman sold $100 million worth of Timberwolf securities to an
Australian hedge fund, Basis Capital. Just 16 days later, on July 4, Goldman informed Basis
Capital that the securities had lost value, and it had to post additional cash collateral to secure its
CDS contract. On July 12, Goldman told Basis Capital that the value had dropped again, and still
more collateral needed to be posted. In less than a month, the value of Timberwolf had fallen by
$37.5 million. Basis Capital posted the additional capital, but soon after declared bankruptcy. On
June 1, Goldman Sachs sold $36 million in Timberwolf securities to a Korean life insurance
company, Hungkuk Life, that had little familiarity with the product. The head of the Korean sales
office said his office was willing to sell the company additional securities, if assured the office
would receive a 7% sales credit. Goldman agreed, and said “get ‘er done.” The sales office sold
another $56 million in Timberwolf securities to the life insurance company which paid $76 per
share when Goldman’s internal value for the security was $65. Within ten days of that sale,
Thomas Montag, a senior Goldman executive, sent an email to the Mortgage Department head,
Daniel Sparks, stating: “boy that timeberwof [Timberwolf] was one shitty deal.”1598 Despite that
comment, Goldman continued to market Timberwolf securities to its clients.
In synthetic CDOs, the cash proceeds from the sales of the CDO 1599 securities were used to purchase “collateral
debt securities.” Later, when cash was needed to make payments to a long or short party, those collateral securities
were sold, and the cash was used to make the payments. In the event the collateral securities used to pay the short
party (called “default swap collateral”) could not be sold for face (par) value, Goldman, the short party, absorbed the
loss, effectively serving as the default swap collateral put provider. For more information, see discussion of
Goldman’s actions taken while the collateral put provider of Timberwolf, below.
1600 9/17/2007 email from Matthew Bieber to Christopher Creed, GS MBS-E-000766370, Hearing Exhibit 4/27-106.
1601 6/2007 Goldman document, “CDO Platform Overview,” at 31, GS MBS-E-001918722 (“ABACUS is the
Goldman brand name for single-tranche CLN [credit linked note] issuances referencing portfolios comprised entirely
of structured products.”); 4/2006 Goldman presentation “Overview of Structured Products,” GS MBS-E-016067482.
1602 3/12/2007 Goldman internal memorandum to Mortgage Capital Committee, “ABACUS Transaction sponsored
by ACA,” GS MBS-E-002406025, Hearing Exhibit 4/27-118.
Goldman had also arranged for its subsidiary, Goldman Sachs International (GSI), to act as
both the primary CDS counterparty and the collateral put provider in Timberwolf.1599 Although
GSI received a fee for serving as the collateral put provider, GSI began to refuse to approve
Timberwolf’s purchase of new collateral securities whose values might decline below par value
and put Goldman at risk of having to make up the difference. Instead, GSI pressured Timberwolf
to keep its collateral in cash, even though an internal Goldman analysis had confirmed that cash
collateral produced lower returns for Timberwolf investors than collateral securities. When
Greywolf objected to this practice, Goldman backed down and allowed the purchase of a narrow
range of very safe, short term asset backed securities as collateral.
In the fall of 2007, a Goldman analyst provided executives with a price history for
Timberwolf A2 securities. It showed that, in five months, Timberwolf securities had lost 80% of
their value, falling from $94 in March to $15 in September. Upon receiving the pricing history, the
Timberwolf deal captain, Matthew Bieber, wrote that March 27 – the day Timberwolf issued its
securities – was “a day that will live in infamy.”1600 Timberwolf was liquidated in 2008.
Abacus 2007-AC1. Abacus 2007-AC1 was a $2 billion synthetic CDO that referenced
BBB rated mid and subprime RMBS securities issued in 2006 and early 2007. It was a static CDO,
meaning once selected, its reference obligations did not change. It was the last in a series of 16
Abacus CDOs that referenced primarily mortgage backed assets and were designed by Goldman.
Those Abacus CDOs were known as single tranche CDOs, structures pioneered by Goldman to
provide customized CDOs for clients interested in assuming a specific type and amount of
investment risk. They enabled the client to select the assets, the size of the investment, the amount
of subordination or cushion before the securities would be exposed to loss, and could be issued
with a single tranche.1601 The Abacus CDOs also enabled investors to short a selected group of
RMBS or CDO securities at the same time. Goldman used the Abacus CDOs not only to sell short
positions to investors, but also as a way for Goldman itself to short mortgage assets in bulk.1602
Abacus 2007-AC1 was the first and only Abacus transaction in which Goldman allowed a
third party client to essentially “rent” its CDO structure and play a direct, principal role in the
As collateral put provider, which it performed for a fee, Goldman did carry 1603 risk in the Abacus 2007-AC1
transaction. In addition, shortly before the Abacus 2007-AC1 transaction closed, Goldman agreed to take the long
side of a CDS contract on the performance of a small portion of the underlying assets when Paulson wanted to
increase its short position at the last minute. Goldman tried to find an investor to assume its small long position, but
was unable to do so.
1604 April 27, 2010 Subcommittee Hearing at 82.
1605 Subcommittee interview of Laura Schwartz (ACA) (4/23/2010). See also In the Matter of Abacus 2007-AC1
CDO, File No. HO-10911 (SEC), Statement of Laura Schwartz (January 21, 2010), ACA ABACUS 00004406 at
408. (hereinafter “Statement of Laura Schwartz”).
selection of the assets. Goldman did not itself intend to invest in the CDO.1603 Instead, it
functioned primarily as an agent, earning fees for its roles in structuring, underwriting, and
administering the CDO. Those roles included Goldman’s acting as the placement agent, collateral
securities selection agent, and collateral put provider. Unlike previous Abacus CDOs, Goldman
employed a third party to serve as the portfolio selection agent, essentially using that agent to
promote sales and mask the role of its client in the asset selection process.
Goldman originated Abacus 2007-AC1 in response to a request by Paulson & Co. Inc.
(“Paulson”), a hedge fund that was among Goldman’s largest customers for subprime mortgage
related assets. Paulson had a very negative view of the mortgage market, which was publicly
known, and wanted Goldman’s assistance in structuring a transaction that would allow it to take a
short position on a portfolio of subprime mortgage assets that it believed were likely to perform
poorly or fail. Goldman allowed Paulson to use the Abacus CDO for that purpose. In entering into
that arrangement with Paulson and simultaneously acting as the placement agent responsible for
marketing the Abacus securities to long investors, Goldman created a conflict of interest between
itself and the investors it would be soliciting to buy the Abacus securities.
Paulson established a set of criteria to select the reference assets for the Abacus CDO to
achieve its investment objective.1604 After establishing those parameters, Paulson worked with the
actual portfolio section agent to select the assets. Documents show that Paulson proposed,
substituted, rejected, and approved assets for the reference portfolio. Goldman was aware of
Paulson’s investment objective, the role it played in the selection of the reference assets, and the
fact that the selection process yielded a set of poor quality assets. Of the final set of 90 assets
referenced in the Abacus CDO portfolio, 49 had been initially proposed by Paulson. Yet Goldman
did not publicly disclose the central role played by Paulson in the asset selection process or the fact
that the economic interest held by an entity actively involved in the asset selection process was
adverse to the interest of investors who would be taking the long position.
ACA Management LLC, the company hired by Goldman to serve as the portfolio selection
agent, told the Subcommittee that, while it knew Paulson was involved, it was unaware of
Paulson’s true economic interest in the CDO. The ACA Managing Director who worked on the
Abacus transaction stated that ACA believed that Paulson was going to invest in the equity tranche
of the CDO, thus aligning its interests with those of ACA and other investors.1605 ACA and its
parent company both acquired long positions in the Abacus CDO as did a third investor. The
Abacus securities lost value soon after purchase. The three long investors together lost more than
3/12/2007 Goldman memorandum to Mortgage Capital Committee, 1606 “ABACUS Transaction sponsored by
ACA,” GS MBS-E-002406025-28, Hearing Exhibit 4/27-118.
1607 Securities and Exchange Commission v. Goldman, Sachs & Co. and Fabrice Tourre, Case No. 10-CV-3229,
(S.D.N.Y.), Consent of Goldman Sachs, (July 14, 2010), at 2.
$1 billion, while Paulson, the sole short investor, recorded a corresponding profit of about $1
billion. Today, the Abacus securities are worthless.
In addition to not disclosing the asset selection role and investment objective of the Paulson
hedge fund, Goldman did not disclose to investors how its own economic interest was aligned with
Paulson. In addition to accepting a sizable placement fee paid by Paulson for marketing the CDO
securities, Goldman had entered into a side arrangement with the hedge fund in which it would
receive additional fees from Paulson for arranging CDS contracts tied to the Abacus CDO that
included low premium payments falling within a specified range.1606 While those lower premium
payments would benefit Paulson by lowering its costs, and benefit Goldman by providing it with
additional fees, they would also reduce the amount of cash being paid into the CDO,
disadvantaging the very investors to whom Goldman was marketing the Abacus securities.
Goldman nevertheless entered into the arrangement, contrary to the interests of the long investors
in Abacus, and failed to disclose the existence of the fee arrangement in the Abacus marketing
On April 16, 2010, the SEC filed a complaint against Goldman and one of the lead
salesmen for the Abacus CDO, Fabrice Tourre, alleging they had failed to disclose material adverse
information to potential investors and committed securities fraud in violation of Section 17(a) of
the Securities Act of 1933 and Section 10(b) and Rule 10b-5 of the Securities Exchange Act of
1934. On July 14, 2010, Goldman reached a settlement with the SEC, admitting:
“[T]he marketing materials for the ABACUS 2007-AC1 transaction contained incomplete
information. In particular, it was a mistake for the Goldman marketing materials to state
that the reference portfolio was ‘selected by’ ACA Management LLC without disclosing
the role of Paulson & Co. Inc. in the portfolio selection process and that Paulson’s
economic interests were adverse to CDO investors.”1607
Goldman agreed to pay a $550 million fine.
The Hudson, Anderson, Timberwolf, and Abacus CDOs provide concrete details about how
Goldman designed, marketed, and administered mortgage related CDOs in 2006 and 2007. The
four CDOs also raise questions about whether Goldman complied with its obligations to offer
suitable investments that it believed would succeed, and provide full disclosure to investors of
material adverse interests. They also illustrate a variety of conflicts of interest in the CDO
transactions that Goldman resolved by placing its financial interests and favored clients before
those of its other clients.
9/26/2007 Michael J. Swenson Self-Review, GS-PSI-02396-401 at 398, Hearing E 1608 xhibit 4/27-55b. See also
12/14/2006 email from Daniel Sparks to Messrs. Montag and Ruzika, “Subprime risk meeting with Viniar/McMahon
Summary,” GS MBS-E-009726498, Hearing Exhibit 4/27-3 (“there will be very good opportunities as the market[]
goes into what is likely to be even greater distress”); 7/13/2006 email from Stuart Bernstein copied to Mr. Cohn, GS
MBS-E-016209254 (“he believes the REIT market is dead. We agreed . . . that as the market got worse, his
‘distressed’ expertise would be more (not less) interesting to investors”). See also Section 5(a)(iii) below regarding
Goldman executives’ negative views of the market for subprime mortgages and subprime mortgage backed
1609 9/19/2006 email chain between Joshua Birnbaum and Daniel Sparks, “ABX,” GS MBS-E-012683946.
Each of the four CDOs examined by the Subcommittee presents conflict of interest
concerns and elements of deception related to how information about the CDO was presented to
investors, including disclosures related to the relevant CDO’s asset selection process, the quality
and value of the CDO’s assets and securities, and the nature and size of Goldman’s proprietary
financial interests. The Subcommittee’s investigation raises questions regarding whether Goldman
complied with its obligations to disclose material information to investors, including its material
adverse interests, and to refrain from making investment recommendations that are unsuitable for
any investor by recommending financial instruments designed to lose value and perform poorly. A
key issue underlying much of this analysis is the structuring of and disclosures related to financial
instruments that enable an investment bank to bet against the very financial products it is selling to
(4) How Goldman Shorted the Subprime Mortgage Market
Having provided an overview of Goldman’s shorting activities and CDO activities in the
years leading up to the financial crisis, this next section of the Report provides detailed information
about how Goldman shorted the subprime mortgage market.
(a) Starting $6 Billion Net Long
By mid-2006, Goldman’s Mortgage Department had a predominantly pessimistic view of
the U.S. subprime mortgage market. According to Michael Swenson, head of the Mortgage
Department’s Structured Products Group: “[D]uring the early summer of 2006 it was clear that the
market fundamentals in subprime and the highly levered nature of CDOs [were] going to have a
very unhappy ending.”1608
$6 Billion Long. In mid-2006, Goldman held billions of dollars in long subprime
mortgage related securities, in particular the long side of CDS contracts referencing the ABX
Index. In September 2006, Mortgage Department head Daniel Sparks and his superior, Jonathan
Sobel, initiated a series of meetings with Mr. Swenson, head of the Structured Products Group
(SPG), and Mr. Birnbaum, the Mortgage Department’s top trader in ABX assets, to discuss the
Department’s long holdings.1609 In those meetings, they discussed whether the Asset Backed
Security (ABS) Trading Desk within SPG should get out of its existing positions or “doubledown.”
After the first meeting, Mr. Birnbaum emailed Mr. Swenson:
1610 Id.
1611 Goldman responses to Subcommittee QFRs at PSI_QFR_GS0239. For more information on Hudson, see
section C(5)(b)(ii)AA., below.
1612 3/10/2007 email to Daniel Sparks, “Mortgage presentation to the Board,” GS MBS-E-013323395, Hearing
Exhibit 4/27-17 (Mortgage Department was $6 billion net long at start of the quarter). 12/13/2006 Goldman email,
“Subprime Mortgage Risk,” Hearing Exhibit 4/27-2.
1613 The ABX Index actually consisted of five separate indices. Each index tracked a different set of RMBS
securities pulled from the 20 RMBS securitizations in the ABX basket. The sets varied according to their assigned
credit ratings. One index tracked the 20 RMBS securities with AAA ratings, another tracked the 20 RMBS
securities with AA ratings, and so on.
1614 Subcommittee interview of Joshua Birnbaum (10/1/2010); Subcommittee interview of Rajiv Kamilla
“Sobel and Sparks want to know if we should exit or double down. We double down if we
have a structured place to go with the risk. ... [W]e are going to sit down with the CDO
guys and talk about a deal.”1610
If the Department’s existing long positions could be transferred off SPG’s books by finding a
“structured place to go with the risk,” the ABS Trading Desk would then be free to “double down”
by taking on new positions and risk.
That same month, September 2006, the ABS and CDO Desks reached agreement on
constructing a new CDO to provide the ABS Desk with a “structured exit” from some of its
existing investments. The result was Hudson Mezzanine 2007-1, a CDO designed by Goldman to
transfer to Hudson investors the risk associated with $1.2 billion in net long ABX assets then in
Goldman’s inventory. The Hudson CDO was also designed to allow Goldman to short $800
million in RMBS securities to offset a portion of its long ABX assets.1611
In December 2006, even after the $2 billion Hudson CDO was constructed, the Mortgage
Department calculated that it still had a $6 billion net long position in subprime mortgage related
assets.1612 Goldman’s ABX holdings continued to be a major source of its long assets.
Goldman’s Long ABX Assets. In January 2006, Goldman, Deutsche Bank, and several
other Wall Street firms launched the ABX Index which, for the first time, allowed investors to use
standardized CDS contracts to invest in baskets of subprime RMBS securities. The ABX Index
measured the aggregate performance of a selected basket of 20 RMBS securitizations, producing a
single value that rose or fell over time in line with the performance of the underlying RMBS
securities.1613 Investors could enter into CDS contracts that used a particular ABX Index as the
“reference obligation,” without physically purchasing or holding any of the RMBS securities in the
underlying basket. Because the ABX Index itself was synthetic, and did not depend upon the
acquisition of large blocks of RMBS securities, it enabled an unlimited number of investors to
make unlimited bets on the performance of a group of subprime RMBS securities, using
standardized contracts that could be bought and sold. The ABX Index also made it economical for
investors to short subprime RMBS securities in bulk.1614
6/20/2006 email from David Lehman, “Mortara Nomination 1615 – ABX/CMBX Indices,” GS MBS-E-014038810
(attached file, “2006 Mike Mortara Award for Innovation Nomination Template for ABX & CMBX Indices,” GS
MBS-E-014038811 at 6 (hereinafter “Mortara Award Submission”)).
1616 Id. at 8.
1617 6/20/2006 email from “Equities and FICC Communications” to “All Equities and FICC,” Mike Mortara Award for
Innovation,” GS MBS-E-010879020 [original email].
1618 Mortara Award Submission at 2.
1619 Id. at 3-4. See also 8/30/2006 Fixed Income, Currency and Commodities Individual Review Book for Rajiv K.
Kamilla, at 20, GS-PSI-04064 (hereinafter “Kamilla 2006 Review”); 9/6/2007 Fixed Income, Currency and
Commodities Individual Review Book for Rajiv K. Kamilla, at 19, GS-PSI-04100 (hereinafter “Kamilla 2007
1620 Kamilla 2007 Review at 19. In 2007, Mr. Kamilla was named a Managing Partner at Goldman.
1621 Daniel Sparks, the Mortgage Department head, viewed Mr. Birnbaum as a talented trader, writing in October
2006: “Josh for EMD [Extended Managing Director] – he is an extraordinary commercial talent and a key franchise
driver. ... He will make us a lot of money.” 10/17/2006-10/18/2006 emails from Daniel Sparks, “3 things,” GS
MBS-E-010917469. Other Goldman senior executives also relied on his trading skills. See, e.g., 2/21/2007 email
from David Lehman, “ACA/Paulson Post,” GS MBS-E-003813259 (before approving the Abacus CDO, Mr.
Lehman asked Mr. Tourre to “[w]alk josh through the $, if that makes sense, let’s go”); 6/29/2007 email from David
Lehman, “ABS Update,” GS MBS-E-011187909 (during exceptionally bad trading day, Mr. Lehman asked Mr.
Swenson, “Is Josh in? Mr. Swenson replied “No he is in Spain – don’t worry I am fine.”).
In internal documents, Goldman described itself as “the leader and principal driver in the
creation of” the ABX Index.1615 In July 2006, Joshua Birnbaum, Rajiv Kamilla, David Lehman,
and Michael Swenson from the Mortgage Department nominated Goldman’s role in the creation of
both the ABX and CMBX – a similar index based on Commercial Mortgage Backed Securities –
for an internal Goldman award, called the “Mike Mortara Award for Innovation.”1616 That award
“recognize[d] the creative, forward-looking, and entrepreneurial contributions of an individual or
team” within the equities or fixed income divisions.1617 The Mortgage Department personnel wrote
that the new indices “enable[d] market participants to trade risk without ownership of the
underlying SP [structured product] security - thereby permitting market participants to efficiently
go short the risk of these securities.”1618 They also wrote that “Goldman Dominates Client Trading
Volume” with “an estimated 40% market share, and also “dominates the inter-dealer market.”1619
In 2007, Rajiv Kamilla, the ABS trader who spearheaded Goldman’s efforts to launch the ABX
Index, wrote that he “[c]ontinued to enhance our trading dominance in ... ABX indices.”1620
While Mr. Kamilla led Goldman’s efforts to develop the ABX Index, the firm’s day-to-day
ABX trading was conducted primarily by Joshua Birnbaum on the Mortgage Department’s
Structured Products Group (SPG) Trading Desk.1621 Mr. Birnbaum had a negative view of the
Subcommittee interview of Joshua Birnbaum (4/22/2010). 1622 Throughout 2006, Mr. Birnbaum was also working
to develop a new product for Goldman, a suite of home price derivatives that would track the Case/Schiller Home
Price Indices. Like the ABX Index, a derivative product tracking the Case/Schiller Indices would enable investors to
bet on the rise or fall in home prices in various U.S. markets, and provide another vehicle to short the mortgage
market. See 12/18/2006 New Product Memorandum, “Launch of US Property Derivatives Trading,” GS MBS-E-
013492538. In May 2006, Mr. Swenson forwarded an email to Mr. Birnbaum about another financial firm seeking
to develop the same type of home price derivatives as Mr. Birnbaum. A note with the email said: “FYI–The cat is
crawling out of the bag,” meaning that other firms were thinking about launching similar products. See 5/16/2006
email from Michael Swenson to Joshua Birnbaum, “Housing Futures and Options,” GS MBS-E-016087363.
Goldman offered the new U.S. property derivatives briefly in 2007, but they did not develop significant trading
volumes. Subcommittee interview of Joshua Birnbaum (10/1/2010).
1623 The ABX Index began trading in January 2006. Goldman’s position in the ABX Index was flat when the Index
debuted, but became long and grew considerably longer in the fall of 2006. See, e.g., 10/17/2006-10/18/2006 email
exchange between Daniel Sparks and Tom Montag, “3 things,” GS MBS-E-010917469; 3/1/2007 email from Daniel
Sparks, “Dinner,” GS MBS-E-002356757, Hearing Exhibit 4/27-143 (“Most of the synthetic flows were hedge funds
getting short and CDO vehicles getting long.”). Goldman executives also told the Subcommittee that the firm often
took the long side of ABX transactions in which hedge funds were going short. Subcommittee interview of Daniel
Sparks (4/15/2010); Subcommittee interview of David Viniar (4/13/2010). See also, e.g., 8/10/2006 email from
Goldman salesperson, “Paulson bookings,” GS MBS-E-012395893 (“GS sold 550mm protection on ABX.HE.A 06-
2 to Paulson”); 12/7/2006 email chain between Daniel Sparks and Tom Montag, “More thorough response,” GS
MBS-E-010931324 (Goldman had previously handled approximately $9 billion in transactions for Paulson);
Subcommittee interview of Michael Swenson (4/16/2010) (“Paulson [hedge fund] was the biggest counterparty on
day 1 [of the ABX Index] and throughout 2006”).
1624 Mr. Birnbaum’s acquisition of a large net long position in the ABX was viewed negatively by Goldman senior
executives. When Mr. Sparks recommended Mr. Birnbaum for a promotion to managing director in October 2006,
Mr. Montag responded: “Josh slipped a bit with the abx position etc. Is that appropriate.” Mr. Sparks replied:
“Josh ... has had a rough couple of months. But he has handled it very well and is a key person for our franchise. I
don’t think those 2 months should confuse his value to the firm.” 10/17/2006 email exchange between Daniel Sparks
and Tom Montag, “3 things,” GS MBS-E-010917469. Mr. Birnbaum was named a managing director later that
month, but Mr. Montag again raised the issue of his net long ABX position when Mr. Sparks sought to allow Mr.
Birnbaum to continue buying equity put options on companies with subprime exposure. Mr. Montag wrote:
“Unfortunately trader josh has not demonstrated a track record of controlling his position. ... Instead of these lousy
hedges he should just be selling his position.” 3/21/2007 email chain between Tom Montag and Daniel Sparks, GS
MBS-E-010629379, Hearing Exhibit 4/27-21. See also 2/5/2007 email from Richard Ruzika to Gary Cohn, “Are
you living Morgatages [sic],” GS MBS-E-016165784 (Mr. Ruzika: “You know and I know this position was allowed
to get too big – for the liquidity in the market, our infrastructure, and the ability of our traders. That statement would
be the same even if we had gotten the market direction correct –although the vultures would not be circling.”).
subprime mortgage market, and favored the firm’s building a net short position.1622 However,
during 2006, Goldman’s overall ABX position was net long, not net short.
Goldman was net long because, as a market maker that helped launch the ABX Index in
2006, it facilitated ABX trades for a number of clients, and many of those clients – primarily hedge

No comments:

Post a Comment