You are on page 1of 28

PRESENTATION TO

Northwestern University
Did Market Structure Contribute to
the Recent Financial Crisis?
Ravi Mattu

(CDS Bond) Basis Post the Lehman Bankruptcy:


Role of Market Structure

What happened to the basis?


Historical behavior of the basis and volatility during the crisis

Plausible explanations
Issues with LIBOR
Risky counterparties and impact of default correlations
Market structure? Was the deleveraging in corporate bonds triggered
by the basis?
Flows in the derivative market

A look at other Forward/Futures Markets


FX, Equities, Mortgages, Treasury Futures

U.S. Investment Grade and High Yield Corporate Bond Spreads: Option
Adjusted Spreads Over Treasuriesa

2500

2000

Breakeven
Spreads Implied
by default rates
from 1932-1935

bp/year

1500

OAS (bp)
1236

1000

500
236

0
Jun-89 Jun-91 Jun-93 Jun-95 Jun-97 Jun-99 Jun-01 Jun-03 Jun-05 Jun-07 Jun-09
Investment Grade

High Yield

Breakeven Spreads IG

Breakeven Spreads HY

Source: Barclays Capital, Moodys for default rates in the 1932-1935 period, we assume recovery rates of 20%
and that losses were equally distributed over this time period.

Average for the Barclays Capital Indices.

9/12/2008

327

Corp
HY
815

12/16/2008

608

1971

10/8/2009

218

771

Date

Corp IG

Definition of the (CDS Bond) Basis

Basis = CDS Spread Par Priced Asset Swap Spread


(over the Swap curve) of bond
Adjustments

Calculate CDS spread for a default swap with maturity matched to the cash
bond by interpolating the par CDS curve.

The Par Priced-Asset Swap Spread represents the Spread over LIBOR that
would equal the risk-free present value of the coupon stream of a cash bond
plus the current difference between par and the price of the bond:
n

i =1

i =1

par price + c * DFi = (Li + s )* DFi


Where DFi is the risk free discount factor for time i, Li is the LIBOR rate for
time i, c is the risky bond coupon and s is the par priced asset swap spread.

Average Daily Asset Swap Spread and (BondCDS) Basis: for A-rated Bonds
500

400

Date

bp/year

300

200

100

-100
1/2/2007

7/2/2007

1/2/2008

7/2/2008

A-Rated Par Asset Swap Spread

1/2/2009

A-Rated (Bond-CDS) Basis

Source: J.P.Morgan

7/2/2009

Par asset
Swap
Spread

(Bond-CDS)
Basis

9/12/2008

244bp

54bp

12/16/2008

427

282

10/8/2009

162

51

Average Daily (BondCDS) Basis: by Rating

900

(Bond-CDS) Basis

800

Date
700
600

BBB

BB

9/12/2008

54bp

105bp

126bp

12/16/2008

282

388

760

10/8/2009

51

100

123

bp/year

500
400
300
200
100
0
-100
-200
1/2/2007

7/2/2007

1/2/2008

7/2/2008

A-Rated

BBB-Rated

1/2/2009
BB-Rated

Source: J.P.Morgan
6

7/2/2009

Libor As Well As Swap Spreads Were Contaminated


Libor is not a true transaction rate. Banks are asked at what rate they
perceive they can raise rates.

Swap spreads have become extremely low in the 10-year sector and have
been negative in the 30-year sector due to demand to receive fixed.

On-the-run and Off-the-run 10-year Treasuries Par Asset Swap Spreads


80
60
40

Spreads over LIBOR

bp/year

20

Date

0
-20
-40
-60
-80
-100
1/2/2007

7/2/2007

1/2/2008

7/2/2008

On-the-run

1/2/2009

Off-the-run

Source: J.P.Morgan
7

7/2/2009

On-the-run

Off-the-run

9/12/08

-61bp

-38bp

12/16/08

-17

56

10/08/09

-14

Impact of Counterparty Credit Risk on CDS Spreads

Both parties in a CDS are exposed to counterparty risk.


However, the exposure is asymmetric with the buyer of protection having
more exposure to the counterparty than the seller.
Bond spread = CDS spread + Counterparty credit adjustment + Liquidity

To estimate the adjustment required calculate the implied default probability


of both the reference entity (on which the CDS is based) and the financial
counterparty (the seller of protection) from the cash bond spreads.

For an assumed default correlation calculate the joint probability of both


the reference entity and the counterparty defaulting.

Calculate counterparty credit adjustment from (see Appendix) and back


out liquidity compensation. Subtract it from bond spread and iterate
again using Bond Spread less liquidity premium to calculate default
probability in step 1.

Calculating Counterparty Credit Cost Adjustment Numerical Example


Reference Entity: A-rated Industrial
CDS protection seller: A-rated financial
Assumed recovery rates: Counterparty = 25%, Reference entity = 35%
Assume, default correlation = 0.2 (3 to 5 year maturity)
Date: November 3, 2008

Financials spread = 446 bp/treasuries

A-rated Industrial spread = 602 bp/treasuries

A-rated CDS Bond basis = -279 bp

Probability of default (4yr) for Financials = 15%


Probability of default (4yr) for Industrial = 31%
Joint probability of default = 8%
Counterparty credit costs = 35 bp16
The example shown should not be relied upon as representative of an actual investment. No representation is being made that
any account will or is likely to achieve profits or losses similar to those being shown.
9

Sensitivity of Counterparty Credit Costs to Spread Levels


Default Correlation = 0.2

Counterparty Bond Spreads (bp)

Reference Entity Bond Spreads


300

400

500

600

700

800

900

200

13

17

20

23

26

29

400

19

28

36

44

52

60

68

500

23

33

43

53

63

73

83

600

26

38

49

61

72

85

97

800

30

45

59

74

89

105

122

900

32

47

63

79

97

115

134

1000

33

50

67

85

104

124

145

Par Asset Swap Spread is assumed to be 200bp above CDS spread.


If reference entity bond spread is 400 bp and the spread on the seller of protection is 500bp, then the Counterparty credit adjustment is
33bp and the liquidity premium is 167bp.
Source: Citadel Investment Group
The analysis shown above is performed using Citadel proprietary models. The estimates and projections used in this presentation rely
on numerous assumptions that forecast market conditions. These assumptions can be materially inaccurate. Further, regulatory
changes can cause the models used to be materially inaccurate. The example shown should not be relied upon as representative of an
actual investment. No representation is being made that any account will or is likely to achieve profits or losses similar to those being
shown.
10

Sensitivity of Counterparty Credit Costs to Spread Levels


Default Correlation = 0.4

Counterparty Bond Spreads (bp)

Reference Entity Bond Spreads


300

400

500

600

700

800

900

200

19

27

33

37

38

38

39

400

41

58

72

86

98

110

122

500

49

69

86

102

117

132

147

600

56

78

97

116

134

151

169

800

66

92

116

139

161

184

206

900

62

98

123

148

173

198

223

1000

57

103

130

157

184

211

238

Source: Citadel Investment Group

The analysis shown above is performed using Citadel proprietary models. The estimates and projections used in this presentation rely
on numerous assumptions that forecast market conditions. These assumptions can be materially inaccurate. Further, regulatory
changes can cause the models used to be materially inaccurate. The example shown should not be relied upon as representative of an
actual investment. No representation is being made that any account will or is likely to achieve profits or losses similar to those being
shown.
11

Does Counterparty Risk Explain Movements in the (Bond-CDS) Basis?


Counterparty Credit Adjustment Versus Average Basis in A-Rated Industrials
Default Correlation between Reference Entity and Protection seller = 0.2
300
250
200
150
100
50
0
-50
-100
Jan-07

May-07

Sep-07

Jan-08

May-08

Basis (bp)

Sep-08

Jan-09

May-09

Counterparty Credit Adjustment

Source: Citadel Investment Group and J.P. Morgan


The analysis shown above is performed using Citadel proprietary models. The estimates and projections used in this presentation rely
on numerous assumptions that forecast market conditions. These assumptions can be materially inaccurate. Further, regulatory
changes can cause the models used to be materially inaccurate. The example shown should not be relied upon as representative of an
actual investment. No representation is being made that any account will or is likely to achieve profits or losses similar to those being
shown.
12

Sep-09

Sensitivity of Counterparty Credit Cost Adjustment to Default Correlation

350
300
250
200
150
100
50
0
Jan-07

May-07

Sep-07

Jan-08

May-08

Sep-08

Jan-09

May-09

Sep-09

0% Correlation

20% Correlation

40% Correlation

Max Possible Correlation

Source: Citadel Investment Group and JP Morgan

The analysis shown above is performed using Citadel proprietary models. The estimates and projections used in this presentation rely
on numerous assumptions that forecast market conditions. These assumptions can be materially inaccurate. Further, regulatory
changes can cause the models used to be materially inaccurate. The example shown should not be relied upon as representative of an
actual investment. No representation is being made that any account will or is likely to achieve profits or losses similar to those being
shown.
13

Maximum Possible Counterparty Credit Cost Adjustment Versus Average


(Bond-CDS) basis on A-Rated Industrials
350
300
250
200
150
100
50
0
-50
-100
Jan-07

May-07

Sep-07

Jan-08

Basis (bp)

May-08

Sep-08

Jan-09

May-09

Counterparty Credit Adjustment

Source: Citadel Investment Group and J.P. Morgan

The analysis shown above is performed using Citadel proprietary models. The estimates and projections used in this presentation rely
on numerous assumptions that forecast market conditions. These assumptions can be materially inaccurate. Further, regulatory
changes can cause the models used to be materially inaccurate. The example shown should not be relied upon as representative of an
actual investment. No representation is being made that any account will or is likely to achieve profits or losses similar to those being
shown.
14

Sep-09

Historical Default Correlations


Five and Ten Year Default Correlation
by Initial Moodys Ratings (1970 to 1993)
Five Year
A

Baa

Ba

.01

Baa

.01

Baa

.04

.03

.15

.06

.07

.25

Ten Year
B

Baa

Ba

.02

.29

.01

.04

.02

.08

.09

.06

.17

.38

Source: Default Correlation and Credit Analysis, Douglas J. Lucas, The Journal of Fixed Income, March 1995.

15

Scale of Dealer Deleveraging in Corporate Bonds over 2007 and 2008

600

Peak (7/18/07)
Total Inventory: $524BN
for Clients: $243BN

Just prior to Lehm an Bankruptcy (9/10/09)


Total Dealer Inventory: $340BN
for Clients: $180BN

500

$ Billions

400

(9/16/09)
Total Inventory: $215BN
for Clients: $91BN

300
200
100
0
01/05 05/05 09/05 01/06 05/06 09/06 01/07 05/07 09/07 01/08 05/08 09/08 01/09 05/09 09/09
Date
Repo for Clients

Dealer Positions

Source: Primary Dealer Survey, Federal Reserve Bank of New York


16

Scale of Deleveraging Relative to Peak Levels

120%

% of Peak Positions

100%
80%
60%
40%
20%
0%
07/01

07/02

07/03

07/04

07/05

07/06

07/07

07/08

Date
Repo for Clients

Dealer Positions

Source: Federal Reserve Bank of New York


17

07/09

Why did Dealers Have Large Positions in Corporate Bonds?

Many Clients took credit risk exposure through single name CDS or
structured credit tranches.

Dealers could not offset the hedges by buying protection in the CDS
market and, therefore, were buying corporate bonds (cash).

These cash positions became extremely hard to finance during the crisis.

18

Change in Haircut or Initial Margin, April 2007 versus August 2008

April 2007

August 2008

U.S. Treasuries

0.25

3.0

Investment-grade Bonds

0-3

8-12

High-yield Bonds

10-15

25-40

Equities

15

20

Investment grade CDS

Senior leveraged loans

10-12

15-20

Mezzanine leveraged loans

18-25

35+

Prime MBS

2-4

10-20

ABS

3-5

50-60

Source: Financial Stress and Deleveraging, IMF Global Financial Stability Report, October 2008, Page 42
19

Market Structure and Dislocation in other Forward/Futures Markets

Product

Exchange/OTC

Clearing Mechanism

Credit Derivatives

OTC

Bilateral till recently

U.S. Agency Mortgages OTC

Multilateral, MBSCC

Foreign Exchange

Largely OTC

Multilateral

Equities Futures

Exchange (CME)

Multilateral

20

Implied Financing of Forwards/Futures during the Crisis: Agency Mortgage


Current Production Coupon
(Implied spread over LIBOR for financing between
front month delivery and back month delivery)

bp

Spread of TBA Mortgage Financing to LIBOR (Front Month)


300
200
100
0
-100
-200
-300
-400
-500
-600
Jan-07

May-07

Sep-07

Jan-08

May-08 Sep-08

Source: JP Morgan

21

Jan-09

May-09 Sep-09

Implied Spread in Financing Deposits in Various Currencies


(Euros, Sterling, Yen) through 6 month U.S. Dollars
Investing in Sterling

300

300

200

200

bp/year

bp/year

Investing in Euro

100
0
-100
Jan-08

100
0

May-08

Sep-08

Jan-09

May-09

Sep-09

May-09

Sep-09

-100
Jan-08

May-08

Sep-08

Investing in Yen
100

bp/year

50
0
-50
-100
Jan-08

May-08

Sep-08

Jan-09

Source: Citadel Investment Group and JP Morgan

The analysis shown above is performed using Citadel proprietary models. The estimates and projections used in this presentation
rely on numerous assumptions that forecast market conditions. These assumptions can be materially inaccurate. Further,
regulatory changes can cause the models used to be materially inaccurate. The example shown should not be relied upon as
representative of an actual investment. No representation is being made that any account will or is likely to achieve profits or
losses similar to those being shown.
22

Jan-09

May-09

Sep-09

Implied Financing Rate in S&P 500 Futures (90 days Constant Maturity) Versus
3 months LIBOR

60

bp/year

40
20
0
-20
-40
-60
1/2/2008

5/2/2008

9/2/2008

1/2/2009

Source: JP Morgan

23

5/2/2009

9/2/2009

Appendix

24

Citadel Methodology for Calculating Counterparty Credit - Derivations


Step1: Calculating the marginal 4 year default probability
First, the equivalent continuous cash spread is deduced from the observed one (Semi - Annual) :
CashSpread Observed
exp(CashSpread cont ) = 1 +

Next, the 1yr default probability is computed :


Pr( default )1 yr =

exp(CashSpread cont ) 1
exp(CashSpread cont ) (1 R )

where R is the recovery rate of the entity.


The N - year default probability is calculated using the following relationship :

Prob Survivial (1 yr ) = 1 Prob Default (1 yr )


Prob Survival ( N yr ) = (1 Prob Default (1 yr ) ) N
1

Prob Default ( N yr ) = 1 (1 Prob Default (1 yr ) ) N

25

Citadel Methodology for Calculating Counterparty Credit - Derivations


Step2: Calculating the 4 year Joint default probability

Let A be the event that the counterparty defaults, and B be the event that the
reference entity defaults. Also, define IA and IB as the indicator functions of A
and B respectively.

Following these definitions, we have:

( A, B) =

Cov( I A , I B )
[I A I B ] [I A ][I B ]
=
var(I A ) var(I B )
var(I A ) var(I B )

But I A and I A are Bernoulli random variables :

[I A ] = PA , var(I A ) = PA (1 PA ), [I B ] = P (B ), var(I B ) = PB (1 PB )
And [I A I B ] = PAB (Joint Probability)
2

PAB = A, B

PA (1 PA ) PB (1 PB ) + PA PB

26

Citadel Methodology for Calculating Counterparty Credit - Derivations

Step3: Backing-out the counterparty credit from the 4yr joint


default probability
The following correction factor is calculated to correct for the counterparty credit cost
embedded in the observed CDS spread* :

CF =

s
s

P
1 rc
=
P
1 c

2 Pr

P
+ rc

where
3

Prc = Joint probability of both entities defaulting in 4 years


Pr = 4yr default probability of the reference entity
Pc = 4yr default probability of the copunterparty
s = observed CDS spread
s = CDS spread assuming no counterparty default risk

Counterparty Credit = s - s = s 1

CF

Pc
P
1
+ rc
2
3
Counterparty Credit = s
1 Prc
2 Pr

* Source : " Valuing Credit Default Swaps II" , John Hull and Alan White, Journal of Derivatives, Vol. 8, No 3, Spring 2001.

27

This presentation reflects the analysis and views of certain members of Citadel Investment Group, LLCs Fixed Income Team. No recipient should interpret
this presentation to represent the general views of Citadel Investment Group, LLC or its affiliates (together Citadel) or its personnel. Facts, analysis and
views presented in this presentation have not been reviewed by, and may not reflect information known to, other Citadel professionals.
This presentation is based upon information that Citadel considers to be reliable, but Citadel does not warrant to its completeness, accuracy or adequacy and
it should not be relied upon as such.
Assumptions, opinions, views and estimates constitute the Fixed Income Teams judgment as of the date of this presentation and are subject to change
without notice and without any duty to update. Citadel is not responsible for any errors or omissions contained in this presentation and accepts no liability
whatsoever for any direct or consequential loss arising from your use of this presentation or its contents.
The analysis in this presentation was based on a number of quantitative and qualitative estimations and assumptions. These assumptions and estimations
were subjectively determined and may not be accurate. Different estimations and assumptions would have produced materially different results. As a result,
the analysis in this presentation has many inherent limitations and no representation is being made that the analysis will translate into actual outcomes.
This material is not intended as an offer or solicitation for the purchase or sale of any security or other financial instrument.
Nothing in this presentation constitutes investment, legal, accounting or tax advice or a representation that any investment or strategy is suitable or
appropriate to your individual circumstances or otherwise constitutes a personal recommendation to you.
This presentation is not directed to, or intended for distribution to or use by, any person or entity who is a citizen or resident of or located in any locality, state,
country or other jurisdiction where such distribution, publication, availability or use would be contrary to law or regulation or which would subject Citadel to
any registration or licensing requirement within such jurisdiction.
None of the material, nor its content, nor any copy of it, may be altered in any way, transmitted to, copied or distributed to any other party, without the prior
express written permission of Citadel.
Citadel may issue other presentations or materials that are inconsistent with, or reach different conclusions from, the information presented in this
presentation. Those presentations and materials may reflect the different assumptions, views and analytical methods of the individuals who prepared them
and Citadel is under no obligation to ensure that such other presentations or materials are brought to the attention of any recipient of this presentation.
Citadel is involved in many strategies that relate to the asset classes mentioned in this presentation. Citadels businesses may make investment decisions
that are inconsistent with the views, opinions or conclusions expressed in this presentation.
All trademarks, service marks and logos used in this presentation are trademarks or service marks or registered trademarks or service marks of Citadel.
Copyright 2009. All rights reserved.

28