The Effect of Liquidity on Governance Alex Edmans

The Effect of Liquidity on Governance Alex Edmans, LBS, Wharton, NBER, CEPR, and ECGI Vivian Fang, University of Minnesota Emanuel Zur, University of Maryland June 19 th, 2013, WFA 1

Background and Research Question v Policy debate: does stock market liquidity improve or weaken corporate governance? v Policies in the 1930 s promoted market liquidity. v Some worries enhanced liquidity hinders governance. 2

Theoretical Framework v Traditional theories: Shleifer and Vishny (1986), Burkart, Gromb, and Panunzi (1997), Kahn and Winton (1998), Bolton and von Thadden (1998) v Under these theories, blockholders govern through “voice” (intervention) “Voice-B” ØLiquidity results in “no governance” (1) by allowing cut and run: Coffee (1991), Bhide (1993) and (2)“Voice-G” by stimulating insider trading: Maug (2002) ØLiquidity encourages “voice” (1) by providing camouflage: Maug (1998) and (2) by increasing price informativeness Faure-Grimaud and Gromb (2004) 3

Theoretical Framework (Cont’d) v Recent theories: Admati and Pfleiderer (2009), Edmans and Manso (2011) v Under these theories, “exit” (threat of exit) is a governance mechanism in itself “Exit” ØLiquidity strengthens “exit” by (1) encouraging block formation: Edmans (2009) and (2) by stimulating information acquisition: Edmans (2009), Edmans and Manso (2011) 4

Empirical Challenges v Many blockholders do not engage in “voice”. v Diversification reqts for mutual funds; “Prudent man” rules for pension funds (Del Guercio(1996)); Conflicts of interest (Agrawal (2011)) v We focus on hedge fund activists, as they have v Few conflicts and legal restrictions v The full “menu” to choose from: “Voice”, “Exit”, or “No Governance” v High performance-based pay: Clifford and Lindsey (2011) 5

Empirical Challenges (Cont’d) v The threat of exit or threat of voice also exerts governance. v Parrino, Sias, and Starks (2003): actual exit v Norli, Ostergaard, and Schindele (2009): actual voice v Lack of actual exit or voice does not necessarily mean no governance v We use Schedule 13 D (active) and 13 G (passive) filings v Capture monitoring intent rather than instances of actual governance v Filings accurately represent the true monitoring intent Ø 13 G: Legally prohibited from activism Ø 13 D: Onerous filing reqts / Target hostility / Worse credit / Reputation 6

Empirical Challenges (Cont’d) v Liquidity may be endogenous v Reverse causality/ Omitted control variables v We address the endogeneity issue v Reverse causality is less of a concern v We study future governance events (Schedule 13 filings) v Governance characteristics (future or contemporaneous) can be sticky v Decimalization as an exogenous shock to liquidity v Minimum tick size reduced from 1/16 dollar to 1 cent in early 2001 v Bid-ask spreads fell substantially market wide (Bessembinder 2003) 7

Main Findings 1) Liquidity increases the likelihood of block acquisition 2) Conditional on block formation, liquidity encourages 13 G over 13 D 3) A 13 G filing represents a governance mechanism Ø A positive market reaction, a positive holding-period return, and an improvement in operating performance; all stronger in more liquid firms Ø 1) and 2) are stronger in firms with higher managerial sensitivity to price 4) Unconditional effect of liquidity on 13 D filings is positive 8

Contribution v Provide insight into the policy and theoretical debate v Literature on the effect of liquidity on firm outcomes v Firm value: Fang, Noe, and Tice (2009), Bharath, Jayaraman, and Nagar (2013) v Voice: Norli, Ostergaard, and Schindele (2009), Gerken (2009) v Literature on the role of hedge funds in governance v Brav et al. (2008), Clifford (2008), Greenwood and Schor (2009), Klein and Zur (2009, 2011), Boyson and Mooradian (2011) 9

Sample Selection v Data sources v Factiva: 223 activist HFs for 1995 -2010 (Brav et al. (2008)) v SEC’s EDGAR: Initial 13 D/13 G schedules v Filed upon acquiring 5% if intend (do not intend) to engage in activism v CRSP: Daily price, return, and trading volume v Compustat: Firm-year financial data v Execu. Comp: Equity incentives v We end up with v 88, 742 firm-year obs. between 1995 and 2010 (Full sample) v 1, 135 firm-year obs. have initial 13 D/13 G filings (HF sample) 10

Regressors v Measuring liquidity v Amihud (2002) ratio; Fong, Holden, and Trzcinka (2011) measure v -1 × natural logarithm of the two illiquidity measures LIQAM= −ln(AMRATIO); LIQFHT= − ln(FHT) v Higher values of LIQAM and LIQFHT correspond to higher liquidity v Control Variables: Firm characteristics following Brav, Jiang, and Kim (2010); year and industry fixed effects 11

Effect of Liquidity on Block Formation Voice-G H 1: Stock liquidity increases the likelihood that a hedge fund acquires a block. Dependent Variables LIQAMt if post =1 decimalization; LIQFHT 0 otherwise t DECIMAL Controls Year and Industry FE Number of Obs. Used Pseudo R 2 0. 5% v. s. 1. 3% unconditional (2) probability Voice-B Exit – (1) (3) (4) (5) (6) BLOCKt+1 (=1 if 13 D Filing or 13 G Filing; 0 if no block acquisition) 0. 079*** 0. 171*** (0. 013) (0. 021) 3. 975*** 3. 902*** (0. 747) (1. 064) 0. 299*** 0. 544*** (0. 024) (0. 064) Included Included 88, 742 88, 742 0. 003 0. 052 0. 003 0. 046 0. 013 0. 044 12

Effect of Liquidity on Block Formation (Cont’d) v DECIMAL may capture confounding events around 2001 v Tick size reduction should have a bigger impact on stocks with lower price. v We split sample by LOWPRC. DECIMAL is only significant in the subsample with LOWPRC=1. v Narrow testing window helps to focus on the Δ we intend to capture. v We show Δ in liquidity from t-1 to t+1 positively predicts the block formation in t+2, with t indicating the decimalization year. 13

Effect of Liquidity on Governance Mechanisms Voice-G H 2: Conditional upon acquiring a block, stock liquidity reduces the likelihood that the hedge fund files a 13 D rather than a 13 G. Dependent Variables LIQAMt LIQFHTt DECIMAL Controls Year and Industry FE Number of Obs. Used Pseudo R 2 (1) -0. 152*** (0. 046) 1, 135 0. 007 7% v. s. 43% unconditional (2) probability Voice-B (3) (4) (5) 13 Dvs 13 Gt+1 (=1 if 13 D Filing; 0 if 13 G Filing) -0. 169*** (0. 064) -4. 047* -6. 662** (2. 456) (3. 260) -0. 295*** (0. 084) Included 1, 135 0. 096 0. 002 0. 092 0. 008 DECIMAL is again only significant in the subsample with LOWPRC=1. 14 Exit (6) -0. 492** (0. 236) Included 1, 135 0. 087

Is 13 G Filing a Governance Event? Voice-G Voice-B Exit H 3 a: A 13 G filing leads to a positive event-study return, particularly among liquid firms. 3 -day CAR over CRSP valueweighted index Testing CAR_VW(-1, +1)>0 Number of Obs. Used (1) Pooling (2) (3) (4) (5) Low LIQAM High LIQAM Low LIQFHT High LIQFHT 0. 008*** (0. 002) 0. 004 (0. 004) 0. 012*** (0. 003) 630 315 315 • Unlikely to be explained by undervaluation • Multivariate regression results are consistent. Switching from below. LIQAM-median subsample to above increases CAR_VW by 1. 7%. • Results are similar if using 3 -day CAR over CRSP equal-weighted index. 15

Is 13 G Filing a Governance Event? (Cont’d) Voice-G Voice-B Exit H 3 b: A 13 G filing leads to a positive holding-period return, particularly among liquid firms. Holding period raw return over CRSP valueweighted index Testing HOLDINGRET_VW>0 Number of Obs. Used • (1) Pooling (2) (3) (4) (5) Low LIQAM High LIQAM Low LIQFHT High LIQFHT 0. 053*** (0. 017) 0. 015 (0. 026) 0. 092*** (0. 022) 0. 019 (0. 025) 0. 088*** (0. 023) 523 262 261 Results are similar if using holding period raw return over CRSP equal-weighted index. 16

Is 13 G Filing a Governance Event? (Cont’d) Change in operating H 3 c: A 13 G filing leads to an increase in operating performance, performance from particularly among liquid firms. year t-1 to t+1 Low LIQAM Subsample ∆EBITDA/ASSET ∆CFO/ASSET High LIQAM Subsample ∆EBITDA/ASSET ∆CFO/ASSET Low LIQFHT Subsample ∆EBITDA/ASSET ∆CFO/ASSET High LIQFHT Subsample ∆EBITDA/ASSET ∆CFO/ASSET 13 G firms Control firms -0. 021 -0. 007 0. 011 -0. 003 -0. 025 -0. 021 0. 015 0. 011 Voice-G Di. D estimator -0. 017 -0. 019 -0. 022 -0. 019 -0. 021 -0. 020 -0. 018 17 Voice-B T-statistics of Di. D -0. 004 0. 012 0. 033** 0. 016 -0. 004 0. 000 0. 033*** 0. 029** Exit -0. 35 1. 08 2. 55 1. 33 -0. 38 -0. 02 2. 67 2. 43

Effect of Liquidity on Block Formation: Role of WPS Voice-G H 4 a: The effect of liquidity on the probability of block acquisition is stronger in firms with higher managerial sensitivity to price. (1) (2) Dependent Variables LIQAMt ×WPSt LIQFHTt ×WPSt Scaled wealth performance DECIMAL sensitivity (WPS) of Edmans, Gabaix, and Landier (2009) DECIMAL×WPS t = Total Delta / Annual Pay WPSt Controls, Year and Industry FE Number of Obs. Used Pseudo R 2 – Voice. B – (3) BLOCKt+1 (=1 if 13 D Filing or 13 G Filing; 0 if no block acquisition) 0. 180* (0. 101) * 0. 019 (0. 010) * 8. 326 (5. 042) ** 0. 049 (0. 021) 0. 508*** (0. 079) 1. 480* (0. 816) * ** 0. 002 0. 020 -0. 534 (0. 001) (0. 009) (0. 588) Included 24, 645 0. 087 0. 086 18 Exit

Effect of Liquidity on Governance: Role of WPS Voice-G H 4 b: The effect of liquidity on the probability of filing choices is stronger in firms with higher managerial sensitivity to price. (1) (2) A dummy to indicate Dependent Variables LIQAMt above-median-WPS LIQAMt ×HIGHWPSt LIQFHTt ×HIGHWPSt DECIMAL×HIGHWPSt Controls, Year and Industry FE Number of Obs. Used Pseudo R 2 0. 722 (0. 927) -2. 390* (1. 298) 0. 017 (0. 171) Included 322 0. 161 – Voice. B – 13 Dvs 13 Gt+1 (=1 if 13 D Filing; 0 if 13 G Filing) 7. 337 (11. 494) -38. 281* (22. 928) -0. 009 (0. 188) Included 322 0. 157 19 (3) 0. 852 (0. 751) -0. 854* (0. 463) 0. 373 (0. 509) Included 322 0. 156 Exit

Effect of Liquidity on HF Activism Liquidity increases the likelihood that a HF acquires a block (H 1); Conditional on BA, liquidity increases the likelihood of 13 G over 13 D (H 2). Which effect dominates? Voice-G H 5: Stock liquidity increases the unconditional likelihood that a 0. 14% v. s. 0. 6% hedge fund files Schedule 13 D. Dependent Variables LIQAMt LIQFHTt DECIMAL Controls Year and Industry FE Number of Obs. Used Pseudo R 2 Voice-B Exit – unconditional (1) (2) (3) probability 13 DFILINGt+1 (=1 if 13 D Filing; 0 if 13 G Filing or no block acquisition) 0. 103*** (0. 026) *** 3. 851 (1. 435) 0. 309*** (0. 088) Included Included 88, 742 0. 040 0. 038 0. 036 20

Conclusion 1) Liquidity increases the likelihood of block acquisition 2) Conditional on block formation, liquidity encourages 13 G over 13 D 3) A 13 G filing represents a governance mechanism Ø A positive market reaction, a positive holding-period return, and an improvement in operating performance; all stronger in more liquid firms Ø 1) and 2) are stronger in firms with higher managerial sensitivity to price 4) Unconditional effect of liquidity on 13 D filings is positive Liquidity improves blockholder governance 21