CDS Market Liquidity How Liquid is the CDS Market by Adreas Fulop and Laurence Lescourret CDS Liquidity by Ren-Raw Cehn, Franck Fabozzi and Ronald Sverdlove.

CDS Market Liquidity

How Liquid is the CDS Marketby Adreas Fulop and Laurence Lescourret

CDS Liquidity by Ren-Raw Cehn, Franck Fabozzi and Ronald

Sverdlove

Objectives

• CDS premiums contain credit risk AND liquidity problems => CDS premiums cannot be used as pure credit risk measures

• Analysis of CDS market liquidity for a better understanding of volatility and transaction costs on CDS market (FL2007) or for a better understanding of bond spreads (CFS2007)

• Hot topic these days• Can we get some insight concerning recent

credit event ?

Objectives

• CDS premiums contain credit risk AND liquidity problems => CDS premiums cannot be used as pure credit risk measures

• Analysis of CDS market liquidity for a better understanding of volatility and transaction costs on CDS market (FL2007) or for a better understanding of bond spreads (CFS2007)

• Hot topic these days• Can we get some insight concerning recent

credit event ?

Objectives

• CDS premiums contain credit risk AND liquidity problems => CDS premiums cannot be used as pure credit risk measures

• Analysis of CDS market liquidity for a better understanding of volatility and transaction costs on CDS market (FL2007) or for a better understanding of bond spreads (CFS2007)

• Hot topic these days• Can we get some insight concerning recent

credit event ?

Société Générale and BNP-Paribas

Bear Stearns

Models

• FL (2007)Microstructure model

of efficient price and quotes

Hasbrouck (2003)Analysis of intraday

volatility and transaction cost

Mostly descriptive

• CFS (2007)Reduced-form model

of credit risk with liquidity factors for the bonds and CDS markets

Bulher and Trapp (2006)

Data

• FL (2007)Intraday CDS bid and ask

quotes + trades from GFI stamped down to a minute

3 US + 1 European entities January 2004 – December 2006

Each day = 5 time periods5:30-7:30, 7:30-9:30,

9:30-14:30, 14:30-16:30 and 16:30-5:30 NY GMT

• CFS (2007)Intraday CDS bid and

ask quotes + trades from Creditex

February 2000 – April 2003

+ information about the companies from FISD data set

+ bond information from TRACE data set reduced to one observation per day

Data problems

• FL (2007)

Sample : take the last bid and ask (every minute)

Remove the joint observation with negative or null bid-ask spread

• CFS (2007)Remove the repeating

entries and bad data points

Interpolate bid and ask to end up with joint bid and ask observations

Remove the observation with negative or null bid-ask spread

Data problems

• FL (2007)

Sample : take the last bid and ask (every minute)

Remove the joint observation with negative or null bid-ask spread

• CFS (2007)Remove the repeating

entries and bad data points

Interpolate bid and ask to end up with joint bid and ask observations

Remove the observation with negative or null bid-ask spread

Is there such corrections ?

Data problems

• FL (2007)

Sample : take the last bid and ask (every minute)

Remove the joint observation with negative or null bid-ask spread

• CFS (2007)Remove the repeating

entries and bad data points

Interpolate bid and ask to end up with joint bid and ask observations

Remove the observation with negative or null bid-ask spread

Data problems

• FL (2007)

Sample : take the last bid and ask (every minute)

Remove the joint observation with negative or null bid-ask spread

• CFS (2007)Remove the repeating

entries and bad data points

Interpolate bid and ask to end up with joint bid and ask observations

Remove the observation with negative or null bid-ask spreadHow often that happens ?

Model and Estimation

• FL (2007)Hasbrouck (2003) model for efficient price

mt=mt-1+qt-jj+ut

Here efficient log spreadmi

=mi-1+mi-imi-1

+f(i-1,i) (ui+NiZi)

Then A=M+C, where C is the cost of market making

Filtering + MC EM algorithm to account for time varying volatility, jumps and data errors

Model and Estimation

• CFS (2007)Estimation of hazard rate and liquidity

factor based on mid-CDS quotes and ask CDS quotes.

Fixed rate corporate bond pricing formula with or without liquidity impact => yield to maturity

Estimation of a one-factor model for liquidity

Results

• FL (2007)Lost in the tables and

graphs, missing explanations and interpretations

(ex: J-shaped pattern for volatility parameter table 3, no comment fig 1-14, no title fig 7-10)

• CFS (2007)Counterintuitive

results: Fig. 11 as mentioned by the authors.

The relations described do not show (ex: increasing a with rating Fig 12), we do not have standard deviations.

FL (2007)

FL (2007)J-Shaped pattern

FL (2007)J-Shaped pattern

Picks up during off hours

CFS (2007)

CFS (2007)

Increasing in rating for the industrial sector

CFS (2007)

Increasing in rating for the industrial sector

Increasing in rating decreasing in rating for CORP, flat for FI

CFS (2007)Hazard Liquidity

No relation

The larger the firm the more liquid the premium

Results• FL (2007)Bid ask spreads and

roundtrip cost are not lower than their counterparts in the corporate bond markets

Framework allows for data errors, price discreteness and jumps

Volatility is low and transaction costs are higher when trading is thinner.

• CFS (2007)Large bid-ask spreads in

CDS quotes can affect the estimation of the liqui-dity spreads of bonds

Liquidity risk is idiosyncratic

Liquidity is positively related to credit risk

Liquidity premium is uncorrelated to credit risk

Comments

• Very interesting and still a lot to do to understand CDS markets.

• Data are a real problems. • Not sure long sample can be used.

Comments

FL (2007)• Microstructure

model + filtering and MC EM estimation is interesting.

• Stable period. Can serve as a benchmark to analyze the recent evolution.

CFS (2007)• Data treatment in

not neutral.• Since liquidity is

increasing, could be more interesting to do the same analysis in 2003, 2004, …,2007

SEARS

Last comments

• Normal liquidity ?• Cross effects ?• Link with daily liquidity ?

Daily Data