Does the Tail Wag the Dog? The Effect of Credit Default Swaps on Credit Risk Marti G. Subrahmanyam

Stern School of Business New York University

Dragon Yongjun Tang

School of Economics and Finance, The University of Hong Kong

Sarah Qian Wang

Warwick Business School, University of Warwick

For presentation at the IGIDR Silver Jubilee International Conference on “Development: Successes and Challenges in Achieving Economic, Social and Sustainable Progress” Bombay, December 1-3, 2012

Outline

•The Credit Default Swap Market •Literature and Hypotheses •Data and Empirical Methods •CDS Trading and Credit Risk: Empirical Results •CDS Trading and Credit Risk: Controls for Endogeneity •Mechanisms for CDS Trading to affect Credit Rsik •Summary and Conclusions 2

CDS and Bankruptcy: Examples

• CIT Group got into financial difficulty in 2009 – Bankruptcy recovery rate 68.125% – Restructuring exchange offer 82.5% in July 2009 – Many creditors voted for restructuring – Biggest creditor: Bank of America, rumored CDS protection buyer – Goldman made loan to CIT in June 2008, bought CDS in January 09 – Filed for Chapter 11 in November 2009 • YRC got into financial difficulty in 2009 – Restructuring plan supported by many creditors – Brigade Capital, hedge fund with rumored CDS protection, tried to hold-out the restructuring – Workers threatened to protest in front of hedge fund offices – Goldman stopped making the market for YRC CDS – Reached restructuring agreement in 2011

Credit Default Swaps (CDS)

• Cash settlement versus physical delivery • Buyer is shorting credit; seller is taking on credit risk • Protection buyer does not have to hold the bonds • Bond issuer may not know of the existence of these contracts Protection Buyer Quarterly CDS premium In the event of default: Deliver acceptable bond Reimbursement of par value Protection Seller Reference Entity 4

Credit Default Swaps (CDS)

• Insurance-like contracts on losses from credit events • Headline news re: AIG in ‘08-’09, and often mentioned in the popular press • CDS is a tool for credit risk transfer, allows shorting of credit: – Detaches the economic interest from the voting power of creditors – Has the potential to change the behavior of lenders and borrowers – Permits the creation of “ empty creditors”!

– Changes the debtor-creditor landscape significantly

Global Notional Size (ISDA Survey)

CDS: Savior or Evil?

• Greenspan, 98-05: CDS are “ extraordinarily useful ” • Soros, 2009: CDS are “ instrument(s) of destruction ” and should be banned • Stulz (2010) asks for a “better understanding of CDS” • Duffie (2010): “don’t throw the baby out with the bathwater” • U.S. Dodd-Frank regulates CDS • E.U. ban on “naked CDS” • China and India launched on-shore CDS trading • Typical recent comment:

“Credit default swaps are derivatives on credit that look like insurance products. They are so specific they may deserve specific treatment and additional regulatory requirements or perhaps a better harmonisation of the prudential and the traditional market conduct regulation” – October 2012, AMFI, French Securities Regulator.

Our Study

• Empirically examines the impact of CDS trading on credit risk – We find that CDS trading increases credit risk – The relationship seems causal difference from IV/2SLS, Heckman treatment effects, propensity score matching, and difference-in – Impact is associated with amount of CDS outstanding and the type of CDS contract – Number of lenders increases and credit risk increases number of lenders with the • Studies mechanisms for the CDS trading effect – Increased leverage/higher debt cost – Tougher “empty creditors” – Creditor coordination failure

CDS Literature

• CDS pricing, in relation to bond spreads – Longstaff, Mithal, and Neis (2005) – Blanco, Brennan, and Marsh (2005) – Nashikkar, Subrahmanyam and Mahanti (2011) • CDS mechanics, including regulation • CDS implications for firm and users 9

Prior Studies: Real Effects

• Acharya and Johnson (2007): insider trading in CDS • Implications of CDS trading on CDS users: – Duffee and Zhou (2001), Fung, Wen, and Zhang (2012) • Implications of CDS trading on reference entities: – Ashcraft and Santos (2009): CDS increases borrowing cost for risky firms – Saretto and Tookes (2012): CDS firms are able to maintain higher leverage and longer debt maturity • Equilibrium model of Bolton and Oehmke (2011)

Hypotheses

•

H1

: Credit risk increases after CDS introduction •

H2

: Credit risk increases with the amount of CDS outstanding •

H3

: CDS effect is more severe for firms with a greater proportion of CDS to debt outstanding and CDS that excludes restructuring as credit event •

H4

: Number of lenders increases after CDS introduction and credit risk increases with the number of lenders

Data

• CDS transactions 1997-2009 from CreditTrade and GFI (cross-checked with Markit) – 901 CDS introduction for N.A. corporates • Bankruptcy data from New Generation Research, Altman, FISD, UCLA LoPucki, Moody’s – 1628 bankruptcies; 60 with CDS • Firm accounting and financial data from CRSP and Compustat • Bond trading data from TRACE; Analyst data from I/B/E/S; Bank data from Call Reports; LPC DealScan

Variables and Methodology

• Proportional Hazard model (Bharath and Shumway (2008)) – –

CDS Active

CDS Firm

: Indicator for CDS introduction and after : Indicator for firms with CDS at any time – Controls: size, leverage, volatility, stock return, profitability – Modeling bankruptcy filing and rating downgrade

CDS Introduction and Rating Change: A Preliminary Look

Baseline Results

Selection/Endogeneity in CDS Trading

• Potential selection and endogeneity in CDS trading: – CDS firms are fundamentally different from non-CDS firms – Firms are selected into CDS trading – It is possible that investors expect the increase in bankruptcy risk for a firm and initiate CDS trading on it – Omitted variables drive both CDS trading and bankruptcy • To control for endogeneity, we use: – Adjustment for initial credit risk (Distance-to-Default) – Instrumental variables (IV)/2SLS – Heckman selection model for treatment effects – Propensity score matching – Difference-in-difference – Continuous CDS exposure/contract type measures

Instruments for CDS Trading

• Lender FX Hedging • Lender Tier 1 Capital Ratio • Predicting the probability of CDS Trading 17

IV Estimation Result

18

Heckman Treatment Effects Model

Propensity Score Matching

Difference-in-Difference Results

• Matching firms: three CDS prediction models, three matching criteria • Leverage finding is consistent with Saretto and Tookes (2012)

The Other Side of the CDS Effect: Exposure Matters

Summary of Findings so far

• CDS trading is positively related to credit risk • Relationship is robust to controlling for selection and endogeneity in CDS trading • The amount of CDS is also positively associated with bankruptcy risk in both directions

Understanding the Mechanisms

• Potential mechanisms: – Increased leverage/higher debt cost (fundamentals change) • Monitoring reduction by lenders and increased risk-taking by firms (fundamentals change) – Tougher “empty creditors” • “Empty creditor”: creditors whose exposures are hedged (or over hedged) with CDS while nominally they are still lenders. (unobservable, may only take one) • Bolton and Oehmke (2011 RFS) model with CDS seller aware of this • Consistent with: Bankruptcy risk increases with the amount of CDS outstanding • An additional test: ex post restructuring – Creditor coordination failure – Feedback effect from CDS/Market manipulation – Information to increase downside risk

Quote from WSJ, March 6, 2012

Thus, bondholders who also hold credit default swaps are likely —in a strange bit of legal gymnastics —to hold out from the exchange, but at the same time vote to give Greece the authority to drive them into it. "We can't risk getting exchanged and CDS not triggering," said one creditor who holds both Greek bonds and credit-default swaps, and is voting that way.

25

“Empty Creditor” Hypothesis

26

Restructuring as a Credit Event

CDS Increase the Lender Base

• The

ex ante

effect of CDS trading • Relationship bank may have reputation concerns • Lead bank is the delegated monitor • Other banks may find it attractive to lend • CDS trading encourages lending, but then lender coordination is more difficult, resulting in potential failure

Change in Number of Lenders

Lender Coordination Failure

Summary and Conclusions

• • Bankruptcy risk increases after CDS trading – Effect seems causal and works in both directions – Finding consistent with “empty creditor” hypothesis – Creditor coordination failure is worsened after CDS • CDS trading affects banking relationships – Easier access to credit but tougher lenders for renegotiation –

post

active CDS markets – Welfare effects of CDS needs more research (tradeoff of increase in risk with better access to debt capital)

Our view

: CDS help high quality firms and hurt low quality firms, and may aid market discovery of the credit risk of firms

A Related Study

• “

Credit Default Swaps and Corporate Cash Holdings

”, by Subrahmanyam, Tang, and Wang, working paper