You are on page 1of 1

capital 40 THEEDGE mal aysia   |  may 31, 2010

derivatives world

Lessons from the Abacus


Fabulous Fab” Fabrice (Issue 790, Jan 25) and “The insur- followed by the mezzanine, senior ties for the Abacus deal,
Tourre and Goldman ance that did not protect” (Issue and “Super Senior Tranche” at the the 50% to 100% tranche
Sachs have been hog- 788, Jan 11). top [see Chart 1]. by was hardly a super sen-
ging the news all month. Perhaps In the case of a synthetic CDO, Jasvin Josen ior tranche despite being
it is time to present some facts and the pool of securities is not owned The Abacus deal awarded an AAA grade by
technical features of the synthetic by the protection buyer seeking From the many, many pieces of the Rating Agencies.
CDO so that we better understand credit protection. The transaction literature attempting to study the protection and long the risk. Then Summing up the above, funded
these world-changing events. appears more like a speculation by Goldman case, I find the best ex- Goldman bought protection from CDOs should be a preference even
CDOs are like bonds whose in- the protection buyer and the pro- planation at the following link: ACA Capital (then an arm of ABN when a senior tranche appears to
come payments and principal re- tection seller, expecting a fall and http://www.interfluidity.com/ Amro) on the 50% to 100% tranche be safe with an AAA rating.
payments are dependent on a pool increase in value of the securities v2/814.html. of the same CDO. Here, Goldman was
of securities such as bonds, loans respectively. The gist of the transaction: Gold- long protection and short the risk. On Portfolio loss distributions
and so on. For the purpose of this The uniqueness of the CDO is man sold protection to Paulson, a a net basis, Goldman was effectively and correlation
article, I have skipped the basic de- founded in its ability to allow in- hedge fund, on the 45% to 100% exposed to the 45% to 50% tranche of We know that CDO tranches carry dif-
scription of the CDO and suggest vestors with different credit risk tranche of a synthetic CDO refer- the CDO as a protection seller. ferent degrees of default risk. How the
interested readers refer to “Under- appetites to invest in appropriate encing a portfolio of RMBS (resi- A week after the deal was closed, default risk varies for each tranche
standing credit default swaps: CDS tranches accordingly. The riskiest dential mortgage backed securities). Goldman wrote down most of the depends on how correlated the un-
spawns the even murkier CDO” equity tranche takes the first loss In other words, Goldman was short value of the CDO tranche in its derlying securities are. This can be
books. Not long after that, the un- illustrated with a portfolio loss distri-
derlying CDO mortgage loans start- bution, which shows the percentage
ed defaulting and the defaults rap- loss probability of the portfolio.
idly crawled up beyond the 45% When the portfolio is fully di-
Chart 1 – A typical CDO structure tranche, triggering massive pay- versified and correlation is close
outs payable by ACA to Goldman, to zero, we can expect a skewed
$ and Goldman to Paulson. bell curve. As correlation increases,
10 90%-100% Super Senior Tranche The Securities and Exchange the curve becomes monotonically
Commission charge on Goldman decreasing. At this stage, the prob-
15 75%-90% is on ethical grounds, where the ability of larger losses increase and
Senior Tranches investment bank is said to have at the same time, the probability of
$100m 25 50%-75%
acted on instructions of Paulson, smaller losses also increases. For a
who was eager to bet on a decline of very high level of correlation, the
Portfolio of loans (credit cards,
subprime loans. Goldman allegedly distribution becomes U-shaped.
mortgages, etc) 25 25%-50% allowed Paulson to play a major role Here, at maximum default correla-
Junior/MezzanineTranches in the asset-picking process, which tion, all the probability is gathered
15 10-25% caused the underlying securities of at the two ends of the distribution.
the CDO to comprise mainly close- The portfolio either all survives
10 0-10% Equity Tranche to-default home loans. or it all defaults. It behaves like a
The Abacus CDO only had senior single asset.
tranches that begin at around the The Abacus reference portfolio
100
45% mark. The equity tranche did was made up of residential mort-
not exist. In retrospect, with such gages in four areas, where three
risky loans, it is very unlikely Gold-were just next to each other — Cali-
man or any other party wanted to fornia, Nevada and Arizona. All the
take on the default risk there. borrowers were of subprime quality,
Chart 2 : Funded and unfunded transactions From a technical perspective, meaning that they either had poor
there are three interesting aspects credit quality or were refinancing
Unfunded Funded to this: their houses based on a previous
Credit Default Swap Credit Linked Note (CLN) • Funded and unfunded CDOs have almost-default case. The loans in-
Investors (PS) different counterparty risk pro- advertently were highly correlated
$100m files with each other.
Protection premium • Correlation and portfolio loss A senior tranche only looks safe
Buyer(PB) SPV CLN $100m distributions are crucial in un- assuming imperfect correlation.
(InvBank) derstanding the risks a CDO car- With correlation being almost per-
payment upon premium
ries fect in this case, the losses very
default PS PB At maturity or upon
• Pricing of a CDO is centred on quickly crept up to the 45% level and
credit event:
modelling the dependent default started to crawl its way through.
structure of underlying assets What we can take from the above
$100m-(losses on default)
in the reference portfolio. is that parties to a CDO must be
able to know the loss distribution
Funded and unfunded CDOs and the impact of correlation to ap-
In most cases, CDOs are funded. A preciate the extent of default risk
funded transaction is just another they are faced with.
Chart 3: Portfolio Loss Distribution for a large portfolio   way of stating that the counterparty
at 0%, 20% and 95% correlation risk of the protection seller (PS) is What next
taken care of. The protection buyer In the heat of the CDO era, investors
(PB) is exposed to a huge counter- were so eager to collect attractive
Probability(%)
party risk, if upon default of the coupons that they overlooked what
corr=0 CDO (tranche), the PS is unable to they were actually going into. No-
60 corr =20% pay the PB. So at inception, the PB body questioned the credit quality
55 corr =95% pays the notional of the CDO (if of the underlying securities, relying
100% funded) to the PB in the form purely on the supercilious grading
50 of a Credit Linked Note, a kind of awarded by the rating agencies who
45 a bond with the embedded CDO. If also believed that defaults were re-
40 default occurs, the PB will keep the mote in the good times.
losses on default and return the How were they pricing the CDO
35
rest to the PS. in their books? Surely the parties
30 In the Goldman case, apparently could see that the tranches were
25 the trade between Goldman and not worth the value? This will be
20
ACA was unfunded. As the defaults reviewed in the next article.
occurred, Goldman was put at risk
15 in the ability of getting a payout Jasvin Josen is a specialist in
10 from ACA. developing methodologies for
5 It is quite common to have un- the valuation of various credit
funded super senior tranches (say products. She has over 10 years’
85% to 100%) as the tranche is almost experience in investment
0 10 20 30 40 50 60 never expected to default. However, banking and the financial
Loss(%) with hindsight, having known the industry in Europe and Asia.
quality of the underlying securi- Comments: jasvin@gmail.com.

You might also like