Did Liquidity Providers Become Liquidity Seekers? Evidence from the CDS-Bond Basis During the 2008 Financial Crisis

Transcription

1 Did Liquidity Providers Become Liquidity Seekers? Evidence from the CDS-Bond Basis During the 2008 Financial Crisis Jaewon Choi 1 Or Shachar 2 1 University of Illinois at Urbana-Champaign 2 Federal Reserve Bank of NY June, 2014 The views presented here are the authors, and are not representative of the views of the Federal Reserve Bank of New York or of the Federal Reserve System

2 Motivation Breakdown of no-arbitrage relation between the CDS and corporate bonds during the 2008 financial crisis CDS-bond basis became large and negative, following the Lehman collapse Arbitrage opportunities left unexploited; Could have bought cheap bonds and hedge with a long position in CDS

3 Motivation Breakdown of no-arbitrage relation between the CDS and corporate bonds during the 2008 financial crisis CDS-bond basis became large and negative, following the Lehman collapse Arbitrage opportunities left unexploited; Could have bought cheap bonds and hedge with a long position in CDS Deleveraging by bond dealers pointed out as a major reason for the breakdown Dealer banks shrank bond positions due to limited funding Deleveraging caused downward selling pressure Dealers could have been demanding liquidity, while their job was to provide liquidity

4 CDS-Bond Basis (CDS - Bond Spreads) The negative basis implies bonds are cheap relative to CDS AAA Non AAA BS LEH bps

5 Did Dealers Demand Liquidity? Dealers are supposed to absorb liquidity demand when there is excessive selling/buying Leaning against the wind (Weil 2007)

6 Did Dealers Demand Liquidity? Dealers are supposed to absorb liquidity demand when there is excessive selling/buying Leaning against the wind (Weil 2007) Dealers (typically investment banks) are highly levered and are subject to funding liquidity risk (Brunnermeier and Pedersen 2009) Would demand liquidity instead of providing, given funding liquidity shocks

7 Did Dealers Demand Liquidity? Dealers are supposed to absorb liquidity demand when there is excessive selling/buying Leaning against the wind (Weil 2007) Dealers (typically investment banks) are highly levered and are subject to funding liquidity risk (Brunnermeier and Pedersen 2009) Would demand liquidity instead of providing, given funding liquidity shocks Anecdotal evidence that dealers demanded liquidity, following the Lehman Brothers collapse Mitchell and Pulvino (2011): dealers unwound long corporate bond positions and drove significant underpricing

8 Aggregate Bond Holdings by Dealers 250 FRBNY BS LEH $BLN

9 This Paper Were dealers seeking liquidity during the crisis? Did the deleveraging cause the negative basis? Who or what drove the basis then?

10 This Paper Were dealers seeking liquidity during the crisis? Did the deleveraging cause the negative basis? Who or what drove the basis then? Implications for the Volcker Rule Prop trading by dealers are blamed for market instability Regulators try to rein in dealers ability to take risk Duffie (2012): Other levered players, e.g. hedge funds, will fill the gap.

11 This Paper Were dealers seeking liquidity during the crisis? Did the deleveraging cause the negative basis? Who or what drove the basis then? Implications for the Volcker Rule Prop trading by dealers are blamed for market instability Regulators try to rein in dealers ability to take risk Duffie (2012): Other levered players, e.g. hedge funds, will fill the gap. Relates to literature regarding the impact of CDS on the real side Bolton and Oehmke (2011), Saretto and Tookes (2012), Subrahmanyam, Tang, and Wang (2012). Our paper shows that derivatives can disrupt the cash market when arbitrageurs exit

12 Main Results I Contrary to the widespread perception, corporate bond dealers, including prop trading desks, provided liquidity following the Lehman crisis Increased bond holdings in response to liquidity demand Traded against price pressure especially when bond prices are depressed Neither mutual funds nor insurance companies were the major liquidity demanders

13 Main Results II Unwinding of basis trades by non-dealer arbitrageurs (e.g. hedge funds) is responsible for the large negative basis Hedge funds unwound long CDS positions There was no large liquidity demand for bonds without CDS vs. for bonds with CDS Bonds with CDS experienced 8% bigger drop in prices after the Lehman collapse Bonds with maturity close to five years and large negative basis experienced a bigger price fall

14 Data Corporate Bond Mergent FISD: Terms and conditions Enhanced TRACE: Transaction level data NAIC: Insurance company transactions

15 Data Corporate Bond Mergent FISD: Terms and conditions Enhanced TRACE: Transaction level data NAIC: Insurance company transactions Credit Default Swap Markit: End-of-day quotes from major dealers DTCC: CDS volumes for 42 financial firms

16 Data Corporate Bond Mergent FISD: Terms and conditions Enhanced TRACE: Transaction level data NAIC: Insurance company transactions Credit Default Swap Markit: End-of-day quotes from major dealers DTCC: CDS volumes for 42 financial firms Sample period from July 2007 to June 2009 Crisis 1 (Pre-Lehman): July 1, 2007 to Sep 13, 2008 Crisis 2 (Post-Lehman): Sep 14, 2008 to Feb 28, 2009 Crisis 3 (Recovery): Mar 1, 2009 to June 30, 2009

17 Calculating CDS-Bond Basis Basis = CDS Spread Bond Spread Negative basis means bonds are cheap relative to CDS and vice versa.

18 Calculating CDS-Bond Basis Basis = CDS Spread Bond Spread Negative basis means bonds are cheap relative to CDS and vice versa. Calculate the bond spread based on the par-equivalent credit spread methodology of J.P. Morgan Step 1. Get CDS-implied survival probability curve Step 2. Get bond-implied survival probability curve S bond = S cds + ε s.t. ε = arg min { PV (S bond Market Price of Bond) 2}

19 Did Corporate Bond Dealers Delever during the Crisis? Corporate Bonds Aggregate Bonds BS LEH $BLN(Corporate) $BLN(Aggregate)

20 Dealers Liquidity Provision during the Financial Crisis Liquidity provision is captured by negative association of trades with price changes Liquidity providers trade against price dislocation (Brunnermeier and Pedersen 2009)

21 Dealers Liquidity Provision during the Financial Crisis Liquidity provision is captured by negative association of trades with price changes Liquidity providers trade against price dislocation (Brunnermeier and Pedersen 2009) q(bond, t) = c 1 + β 1 basis(t) + ctrls + ε 1t q(cds, t) = c 2 + β 2 basis(t) + ctrls + ε 2t basis t = p CDS,t p CS,t q(cds) and q(bond): CDS and Bond buy order imbalance of dealers Negative signs on β 1 and β 2 indicates liquidity provision by dealers.

22 Liquidity Provision by Dealers Crisis 1 Crisis 2 Crisis 3 q(bond) q(cds) q(bond) q(cds) q(bond) q(cds) basis(t) -0.15*** *** *** 0.00 (-5.27) (-0.94) (-4.22) (-0.20) (-6.13) (0.51) basis(t 1) * ** ** (0.92) (1.91) (0.30) (2.11) (-0.27) (2.27) vix t * -0.06*** (-1.38) (0.41) (1.40) (0.62) (1.82) (-4.51) ois t 0.21*** * 1.22*** (3.25) (-0.31) (1.96) (6.05) (1.36) (-1.36) ret dealer,t -0.46*** 1.26*** (-2.95) (2.59) (0.10) (-0.02) (0.62) (-1.05) R 2 0.6% 0.5% 0.6% 2.4% 0.9% 2.7% N 63,258 39,681 21,543 13,071 27,643 16,389

23 Stabilizing vs. Destabilizing Liquidity Seeking When the basis becomes more negative (price falls further down) P t Zero Basis P t+1

24 Stabilizing vs. Destabilizing Liquidity Seeking When the basis becomes more negative (price falls further down) P t Zero Basis P t+1 Seller: Destabilizing Liquidity Seeker Pushing prices further down by selling underpriced securities. Disrupting the market. Buyer: Stabilizing Liquidity Provider Engaging in arbitrage trades

25 Stabilizing vs. Destabilizing Liquidity Seeking When the basis becomes more positive (price rises further up) P t+1 P t Zero Basis

26 Stabilizing vs. Destabilizing Liquidity Seeking When the basis becomes more positive (price rises further up) P t+1 P t Zero Basis Buyer: Destabilizing Liquidity Seeker Pushing prices further up by buying overpriced securities. Disrupting the market Seller: Stabilizing Liquidity Provider Engaging in arbitrage trades

27 Stabilizing vs. Destabilizing Liquidity Seeking Were dealers a destabilizing liquidity seeker or a stabilizing arbitrageur?

28 Stabilizing vs. Destabilizing Liquidity Seeking Were dealers a destabilizing liquidity seeker or a stabilizing arbitrageur? When basis < 0, for example (bonds are underpriced): q(bond, t) = β 1 basis(t) 1 basis(t)>0 }{{} When Basis Narrows + β 2 basis(t) 1 basis(t)<0 + ε 1t }{{} When Basis Widens β 2 > 0: Destabilizing liquidity seeker β 2 < 0: Stabilizing arbitrageur

29 Stabilizing vs. Destabilizing Liquidity Seeking No evidence for destablizing liquidity seeking in the bond market. Dealers are stabilizing arbitrageurs, while clients are destabilizing. Crisis 1 Crisis 2 Crisis 3 basis > 0 basis < 0 basis > 0 basis < 0 basis > 0 basis < 0 Bond Bond Bond Bond Bond Bond basis 1 basis>0-0.27*** -0.07** * -0.09*** -0.06*** (-7.15) (-2.51) (-0.46) (-1.69) (-5.70) (-4.05) basis 1 basis<0-0.26*** -0.15*** -0.14*** -0.07*** -0.09*** -0.07*** (-5.53) (-4.47) (-2.89) (-4.87) (-3.98) (-6.12) R N 26,077 37,181 4,015 17,528 7,544 20,099

30 Summary Thus Far Dealers increased corporate bond inventory after the Lehman collapse

31 Summary Thus Far Dealers increased corporate bond inventory after the Lehman collapse Dealers (including prop traders in investment banks) provided liquidity No evidence that dealers demanded liquidity and disrupted the corporate bond market

32 Summary Thus Far Dealers increased corporate bond inventory after the Lehman collapse Dealers (including prop traders in investment banks) provided liquidity No evidence that dealers demanded liquidity and disrupted the corporate bond market Who might have driven the basis? Who was seeking liquidity?

33 Summary Thus Far Dealers increased corporate bond inventory after the Lehman collapse Dealers (including prop traders in investment banks) provided liquidity No evidence that dealers demanded liquidity and disrupted the corporate bond market Who might have driven the basis? Who was seeking liquidity? Deleveraging by highly levered, non-investment bank, investors Most likely by the unwinding of basis arbitrage trading Bought corporate bonds through repo financing and hedged with CDS Deterioration in funding situation causes unwinding Massive selling pressure on the corporate bond market

34 Dealer Inventory for CDS vs. Non-CDS Bonds 10 5 BS LEH 0 5 $BLN CDS Exists CDS Does Not Exist

35 Mutual Fund Holdings in CDS vs. Non-CDS Bonds 1200 CDS Exists CDS Does Not Exist BS LEH Aggregate Holdings by MFs ($BLN) Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q4 2009

36 Hedge Fund and Dealer s CDS Positions 60 BS LEH $BLN Dealers Hedge Funds Insurance Companies

37 Unwinding of Basis Arbitrage Trades Unwinding of basis trading causes bond prices to decrease dramatically. Bond prices will fall more for bonds with CDS available Bond prices will fall more for bonds with high basis trading activity

38 Unwinding of Basis Arbitrage Trades Unwinding of basis trading causes bond prices to decrease dramatically. Bond prices will fall more for bonds with CDS available Bond prices will fall more for bonds with high basis trading activity Measure of CDS availability Based on the presence of CDS quotes in the Markit database Measure of basis arbitrage activity How close bond maturity is to five year at the Lehman default. Five-year CDS is the easiest to arbitrage. How negative the basis is before the Lehman Brothers default

39 Cumulative Returns for CDS vs. Non-CDS Bonds BS LEH Cumulative Returns(%) CDS Exists CDS Does Not Exist

40 Did Unwinding of Basis Arbitrage Drive the Negative Basis? Ret t = c 1 + β 1 CDS YES + β 2 basis(aug) Mat5Y(Aug) + ctrls + ɛ t Ret t: Corporate bond return CDS YES : One if the bond has CDS traded anytime before the Lehman bankruptcy. Zero otherwise basis(aug): CDS-bond basis at the end of August 2008 Mat5Y(Aug): One if the bond s maturity is between 4.5 years and 5.5 years at the end of August Zero otherwise.

41 Did Unwinding of Basis Arbitrage Drive the Negative Basis? Ret t = c 1 + β 1 CDS YES + β 2 basis(aug) Mat5Y(Aug) + ctrls + ɛ t Ret t: Corporate bond return CDS YES : One if the bond has CDS traded anytime before the Lehman bankruptcy. Zero otherwise basis(aug): CDS-bond basis at the end of August 2008 Mat5Y(Aug): One if the bond s maturity is between 4.5 years and 5.5 years at the end of August Zero otherwise. Predictions: β 1 < 0: Bond returns more negative if CDS available β 2 > 0: Bond returns more negative if the basis is negative and the cloest maturity is five year

42 Bond Returns after Lehman Brothers Collapse Bond Return CDS YES -0.02** -0.02* -0.04** (-2.39) (-1.69) (-2.27) basis(aug) Mat5Y(Aug) 1.33** 1.57** 0.23 (2.27) (2.17) (0.25) TTM t (1.26) (0.30) VIX t 0.00*** 0.00*** (10.77) (8.92) ILLIQ1 t (0.55) (0.43) ILLIQ2 t 0.01* 0.00 (1.73) (0.67) LEV t (-1.35) VOL t (-1.50) R % 12.0% 12.3% 37.1% N 1,960 1,933 1,933 1,466 Rating Dummy Yes Yes Yes Yes

43 Month-by-Month Bond Price Change Bond Return Aug Sep Oct Nov Dec CDS YES *** (-0.42) (-2.89) (-0.71) (-0.37) (-0.16) basis(aug) Mat5Y(Aug) -0.30** 4.25*** -3.38* (-2.21) (2.87) (-1.79) (0.70) (-0.34) R % 24.1% 41.9% 42.5% 10.4% N Rating Dummy Yes Yes Yes Yes Yes

44 Conclusion No evidence that dealers de-leveraging is associated with lack of liquidity provision on the part of dealers Evidence for the hypothesis that basis arbitrage trading is the driver for the big negative basis Shedding light on the ongoing debate about the possible effects of the Volcker Rules in the OTC markets