Equity Vesting and Managerial Myopia Alex Edmans, LBS, Wharton, - - PowerPoint PPT Presentation

1

Alex Edmans, LBS, Wharton, NBER, CEPR, ECGI Vivian W. Fang, Minnesota Katharina A. Lewellen, Dartmouth Bristol-Manchester 3rd Annual Corporate Finance Conference, September 2014

Equity Vesting and Managerial Myopia

2

Motivation

n Myopia is believed to be a first-order issue

n Theories: Narayanan (1985), Stein (1988, 1989),

Bebchuk and Stole (1993) …

n Policy arguments: Porter (1992), Thurow (1993),

Zingales (2000) …

n But very difficult to document empirically

n In theory models, what matters is horizon of

• incentives. Max α[ωP + (1-ω)V]

n Standard measures of incentives quantify overall

sensitivity to stock price: α, not ω

3

Empirical Approach

n αωP is dollar value of CEO’s equity sales

n But actual equity sales are (a) endogenous (b)

potentially unpredictable

n Our approach: use scheduled vesting of equity

n Relevance: highly correlated with equity sales n Exclusion: driven by grants several years prior n Available post-2006 SEC rules. Short time series, so

Execucomp has little power

n Use Equilar

4

Empirical Specification

n ΔINVESTMENTt+1 = α +

β1NEWLYVESTINGt+1 + β2UNVESTEDADJt + β3ALREADYVESTEDt + γCONTROLt +

n NEWLYVESTING: $ change for 1% rise in price n Control for unvested and already-vested equity n Additional controls:

n Largely follow Asker, Farre-Mensa, and Ljungqvist (2013) n Investment opportunities: Qt, Qt+1, momentum, age, MV n Capacity to finance investment: cash, leverage, retained

earnings

n Other: ROA (measures both), salary, bonus n Firm FE, year FE, cluster standard errors at firm level

5

Related Literature

n Graham, Harvey and Rajgopal (2005) n Level of incentives and EM:

n Cheng and Warfield (2005), Bergstresser and Philippon

(2006), Peng and Roell (2008) vs. Erickson, Hanlon, and Maydew (2006)

n Vesting horizons

n Kole (1997) documents n Johnson, Ryan, and Tian (2009): corporate fraud n Gopalan et al. (2013) introduce “duration”

n Varies across industries n Linked to EM

n Ladika and Sautner (2013)

6

The Data

n Equilar 2006-10 covers Russell 3000 n Shares: “shares acquired on vesting of stock” n Options: uniquely identify each grant using

strike price and date

n Newly-vestingt+1 = Unvestedt + Newly-awardedt+1

– Unvestedt+1 gives number of vesting options

n Multiply by delta at start of t+1 and sum, to give

effective number of shares

n Multiply delta by stock price at start of t+1 to give

sensitivity: $/% incentives

n Sum with shares to give Newlyvesting

7

The Data (cont’d)

n Similarly calculate

n Alreadyvested n Unvested n Unvestedadj = Unvested – Newlyvesting

n Equitysold from Thomson Financial Insider

Trading

8

Dependent Variables

n ΔRD (scaled by total assets)

n Market can’t discern quality

n Cohen, Diether, and Malloy (2013): “the stock market

appears unable to distinguish between “good” and “bad” R&D investment”

n Bushee (1998), Bhojraj et al. (2009)

n ΔRDAD

n Chan, Lakonishok, and Sougiannis (2001)

n ΔCapex, ΔCapexall n ΔRDADCapex, ΔRDADCapexall

9

Predicted Equity Sales and Investment

1st: $1 increase in NEWLYVESTING -> 33c increase in EQUITYSOLD 2nd: IV increase associated with 0.25% fall in ΔRD vs. average of 4.6%; equates to $2.2 million

(1) (2.1) (2.2) (2.3) (2.4) (2.5) (2.6) Dependent Variables EQUITY_ SOLDt ΔRDt ΔRDADt ΔCAPEXt ΔRDAD_ CAPEXt ΔCAPEX _ ALLt ΔRDAD_ CAPEXALLt NEWLYVESTINGt 0.328*** (0.034) FIT_ EQUITYSOLDt

• 0.942*
• 1.192*
• 0.625
• 2.154**
• 4.252**
• 6.564**

(0.553) (0.635) (0.585) (1.083) (1.918) (2.631) UNVESTEDADJt-1

• 0.022
• 0.054
• 0.078
• 0.013
• 0.139

0.422 0.337 (0.025) (0.073) (0.089) (0.123) (0.193) (0.492) (0.593) VESTEDt-1 0.018*** 0.013 0.020 0.050** 0.074** 0.098* 0.136* (0.002) (0.014) (0.016) (0.023) (0.033) (0.059) (0.078) Controls, year FE, firm FE Yes Yes Yes Yes Yes Yes Yes Observations 6,730 6,730 6,730 6,730 6,730 6,730 6,730 Adjusted R2 (R2) 0.421 0.354 0.359 0.304 0.343 0.159 0.138

10

Robustness Checks / Additional Analyses

n Reduced-form on NEWLYVESTING directly n Performance-based vesting (Bettis et al.

(2010)) not a concern if price-based, is a concern if earnings-based

n Stronger for options, for which PBV is very rare

n Using delta of 0.7 for all options or assuming

all options are ATM

n Controlling for duration or vega n But cannot make strong claims about

causality or efficiency

11

Earnings Announcements

12

Earnings Announcements (cont’d)

(1) (2.1) (2.2) (2.3) Dependent Variables EQUITY_SOLDt BEATq BEAT_ BELOW1q BEAT_ABOVE1q NEWLYVESTINGt 0.451*** (0.015) FIT_ EQUITYSOLDt 10.829** 14.596***

• 1.760

(4.863) (5.576) (4.541) Controls, year FE, industry FE Yes Yes Yes Yes Observations 17,173 17,173 17,173 17,173 Wald-statistics 2.70 11.60 2.52 p-value 0.10 <0.01 0.11

Similar results using CUTANDBEAT: dummy = 1 if firm beats the forecast but would have missed it if R&D same as previous year

13

Earnings Announcement Returns

n Stein (1989): market rationally takes into

account managers’ myopic tendencies and discounts announced earnings

n Alternatively, market may not discount, as

n Lacks information on managers’ incentives n Is inefficient (von Lilienfeld-Toal and Ruenzi

(2014))

n Earnings surprises suggest that analysts don’t

take myopic incentives into account

14

Earnings Announcements (cont’d)

(1) (2.1) (2.2) Dependent Variables EQUITYSOLDt CARq (-1, +1) NEWLYVESTINGt 0.420*** (0.026) FIT_ EQUITYSOLDt 76.350** 44.418 (29.820) (30.145) DIFq 0.329 (0.290) BEATq 6.327*** (0.200) Industry Fixed Effects Yes Yes Yes Observations 18,686 18,686 18,686 Adjusted R2 (R2) 0.306 0.007 0.088

15

Conclusion

n NVE is negatively related to R&D, advertising,

capex; narrowly beating earnings

n Frydman and Jenter (2010):

n “Compensation arrangements are the endogenous

• utcome of a complex process … this makes it

extremely difficult to interpret any observed correlation between executive pay and firm

• utcomes as evidence of a causal relationship”

16

Other Consequences of Vesting Equity

n Edmans, Goncalves-Pinto, Wang, and Xu

(2014): “Strategic News Releases in Equity Vesting Months”

n Why is news important?

n Real decision makers base decisions on news (or

stock prices affected by news): Bond, Edmans, and Goldstein (2012)

n Reduces information asymmetry among investors

(cf. Regulation FD)

n News is not mechanically triggered by events,

but a strategic decision by the CEO

17

Equity Vesting and Equity Sales

18

News Releases in Vesting Month

n

Column 3: firms release 5% more discretionary news in vesting month

19

Positivity of News Releases

20

Abnormal Returns to News Releases

n 16-day CAR of 28 bps to discretionary news

in VM

n $14,504 applied to average CEO equity vesting of

$5.18m

n Meulbroek (1992): median gain to illegal

insider trading of $17,628 ($33,968 in 2007 dollars)

n Martha Stewart avoided losses of $45,673

when she sold ImClone shares in 2001

21

Economic Significance, Returns

22

Economic Significance, Volume

n CEO’s equity sale (on a sale day) is

n 6.2% of the average daily volume n 0.165% of shares outstanding

n A discretionary news item generates

abnormal trading volume of 0.32% of shares

• utstanding

23

Time From News Until CEO’s First Sale

n Median time to full sale is 7 days for DN

released in vesting months

n Consistent with short-lived return and volume

increase