Does Majority Voting Improve Board Accountability? Stephen Choi, Jill Fisch, Marcel Kahan, and Edward Rock Forthcoming __ U. Chi. L. Rev. __ (2016)

Director elections Traditional decision rule for director elections in US corporations: plurality vote (PVR): candidates with most votes are elected. The critique of PVR: Most director elections are uncontested if # of candidates = # of open positions then candidate can win with 1 “for” vote (despite widespread opposition). The alternative: majority voting (MVR): to be elected, director must receive a majority of votes cast.

The Recent Rise of Majority Voting 2005 only 9 of the S&P 100 companies used majority voting in director elections The Carpenter Pension Funds submitted 12 shareholder proposals in the 2004 proxy season January 2005 reporter Louis Lavelle published an article in Business Week By January 2014, almost 90% of S&P 500 companies had a majority voting standard and/or a director resignation policy Advocates: ISS: “A majority vote standard transforms the director election process from a symbolic gesture to a meaningful voice for shareholders.” CII: “Majority voting ensures that shareowners’ votes count and makes directors more accountable to the shareowners they represent.”

Plurality v. Majority Vote Regime (S&P 1500) Based on our sample of SP1500 companies. 2 more left later

What to make of this development? Does MVR make a meaningful difference in director accountability? Or is this an example of symbolic corporate governance – a paper tiger? The incidence of majority withholds under either rule is very small Directors at PVR companies are 20 times more likely to receive majority withhold than directors at MVR companies But even in failed elections, directors rarely depart Is this because MVR company directors behave better? MVR causes directors to behave better? Something else is going on? In 2007-2013: -24,000 director nominees in S & P 1500 companies with MVR -Only 8 failed to received a majority -And, of those 8, only 3 left the board promptly.

Potential Explanations “Selection”: companies whose directors behave “well” are more likely to adopt MVR “Causal”: a change in behavior occurs after a company shifts to MVR leading to more for votes “Deterrence”: directors at MVR companies behave “better” (or at least more shareholder-friendly) than directors at PVR companies “Electioneering”: MVR companies may engage in more campaigning (including lobbying ISS) because the stakes are higher “Shareholder restraint”: shareholders in PVR companies are more likely to vote against directors (to induce changes in policy) than in MVR companies (where a no vote may disrupt governance)

Dataset Uncontested director elections for S&P 1500 companies for the years 2007 through 2013 64,933 company-director election observations Overall – 37.3% of the elections were subject to MVR ISS issued withhold recommendations for 6.6% of director nominees

Plurality v. Majority Vote Regime Early Late Based on our sample of S&P 1500 companies

Key Tests Hazard Model – Tests Selection Fixed Effects – Tests Causation Difference in Difference – Tests Causation Direct Test of Bad Behavior (Attendless75 and Issue Proposal Implementation) – Tests Causation Controlling for Offense—What happens to the Withhold Vote at the 50% threshold – Weak test of electioneering and/or SH restraint But we seem to find evidence of causation in direct test of bad behavior!!!

Key Results MVR firms are different – evidence of selection effect In particular, early adopters of MVR were more shareholder-responsive than PVR firms Adoption of MVR increased director accountability After adopting MVR, directors of adopting firms had greater shareholder support and behaved better (firm fixed effects and matching tests)

The Voting Threshold Most of our tests use a 30% threshold Reasons to believe that a 30% withhold vote is a significant signal of investor concern As indicated above, there are very few elections in which a director receives less than 50% of the votes cast Even a 30% withhold vote is relatively rare Final tests look at 50% threshold because only there is the difference in voting rule potentially binding, although, we noted above, the impact of a failed MVR election is unclear

The Selection Hypothesis: Are companies that adopted majority voting different from those that did not?

Table 3: Average of Prior 1st and 2nd Years (Measured in 2011) Variable Did Not Switch Switched to MVR   N Mean p-value Average ISS WH Rec 827 0.153 167 0.101 0.011 Any Director Received an ISS WH Rec 0.418 0.329 0.032 Any Director Received >30% WH Vote 826 0.252 0.180 0.046 Switching companies had a better prior record of director success – fewer WH recommendations and fewer problem votes So shareholders are not targeting the firms where they are unhappy with directors as a way to actually get the bad directors off the boards Firms adopting MVR are different – the switching firms had a better prior record of director success

Table 4: Hazard Model for Switch to MVR The difference is concentrated in the early adopters (compared to everyone else) Table 4: Hazard Model for Switch to MVR These models are full sample   Model 1 Model 2 Model 3 Avg ISS WH Rec Prior 2 Yrs 0.628+ Any ISS WH Rec Prior 2 Yrs 0.818* High WH Vote Prior 2 Yrs 0.869 Delaware 1.240* 1.246* 1.250* PPill 0.844 0.835+ 0.827+ ClassBd 1.023 1.003 1.015 CumVote 0.592* 0.589* 0.588* Top5AbRet 0.529** 0.528** 0.530** Bot5AbRet 1.415 1.433 1.418 ln(Mktcap) 1.596** 1.600** 1.592** Insthold 1.699+ 1.671+ 1.716+ CharterAmend 0.654* 0.649** 0.653* N 4693 4668 Pseudo R2 0.042 0.041 Significant for early adopters Significant for late adopters Note – it is the ISS withhold rec that seems to matter, not the actual vote Del and pill significant for early adopters only (and sign switches when we go to late adopters although we lose significance) Top 5 AB only significant for late adopters Exponentiated coefficients; + p < 0.10, * p < 0.05, ** p < 0.01

The Causation Hypothesis: The Effects of Majority Voting on Subsequent Electoral Success

Table 5 Panel A: OLS Firm Fixed Effects   Model 1 All OBS Model 2 Excludes 2 years after SH adoption of MVR resolution Model 3 Excludes 2 years after SH adoption of MVR resolution + year switched to MVR Model 4 Excludes firms that never adopted MVR Model 5 Only firms that adopted MVR after SH adoption of MVR resolution Whvote30 MVR -0.0259** -0.0219** -0.0252** -0.0278** -0.102** Insthold -0.00429 -0.0123 -0.0117 0.0101 -0.023 Top5AbComp 0.00197 0.00106 0.00501 0.00637 0.0242 ln(Mktcap) -0.00814 -0.00716 -0.00747 -0.0105 -0.0164 SDret 0.113 0.238 0.24 -0.111 0.651 Top5AbRet -0.0132 -0.0136 -0.0148 0.00881 -0.0241 Bot5AbRet 0.0223 0.0222 0.0236 0.0109 0.0456 Constant 0.0999+ 0.0917 0.0905 0.124+ Firm Fixed Effects Yes Year Fixed Effects N 44592 43201 39507 25354 3175 adj. R2 0.202 0.196 0.099 0.148 Any way you cut it, firms that adopt MVR subsequently have a lower incidence of a votehigh withhold Firms that adopt MVR have a lower subsequent incidence of significant withhold votes

Table 5 Panel B: OLS Firm Fixed Effects, Early v. Non-Early Adopters   Model 1 All OBS Model 2 Excludes 2 years after SH adoption of MVR resolution Model 3 Excludes 2 years after SH adoption of MVR resolution + year switched to MVR Model 4 Excludes firms that never adopted MVR Model 5 Only firms that adopted MVR after SH adoption of MVR resolution Whvote30 EarlyMVR -0.0149* -0.0175** -0.0203** -0.00783 -0.0306 LateMVR -0.0308** -0.0239** -0.0285** -0.0367** -0.129** This effect seems to be concentrated in the late adopters 10% 1% 5% The effect seems to be concentrated in the late adopters NOTE: Coefficients for other independent variables omitted from above table.

Table 5 Panel C, Ranges   Model 1 Model 2 Model 3 Whvote ≥ 10% & Whvote < 30% Whvote ≥ 30% & Whvote < 40% Whvote ≥ 40% & Whvote < 50% EarlyMVR -0.0065 -0.00326 -0.00712* LateMVR -0.00547 -0.0180** -0.00748* 10% NOTE: Coefficients for other independent variables omitted from above table. Changes in the ≥ 30% & < 40% Whvote range are more consistent with Deterrence-Causation as opposed to Electioneering or SH Restraint-Causation Hypotheses.

Table 5 Panel D Matched Sample Difference in Differences Models   Model 1 All OBS Model 2 Excludes 2 years after SH adoption of MVR resolution Model 3 Excludes the year that the MVR adopter switched to MVR Model 4 Only firms that adopted MVR after SH adoption of MVR resolution Whvote30 MVR Adopter -0.327 -0.307 -0.293 0.935 Post-MVR Switch 0.137 0.134 0.0887 -0.00558 MVR Adopter x Post-MVR Switch -1.862** -1.841** -2.001** -2.290+ This is a very crude matched sample Firm that switched had lower withhold votes after the switch than their match, relative to their withhold votes before the switch A crude match (market cap and 2 digit SIC) Switching firms had lower WH votes after the switch than their match NOTE: Coefficients for other independent variables omitted from above table.

Table 5 Panel E Matched Sample Early v. Late Adopters   Model 1 Model 2 Model 3 Model 4 Whvote30 MVR Adopter EARLY -1.337** -1.191** -1.151** 0.0803 Post-MVR Switch EARLY 0.213 0.239 0.113 0.699 MVR Adopter EARLY x Post-MVR Switch EARLY -0.735 -0.972+ -1.409* -1.463 MVR Adopter LATE -0.260 -0.246 -0.239 0.937 Post-MVR Switch LATE 0.0497 0.000363 0.0456 -2.955* MVR Adopter LATE x Post-MVR Switch LATE -2.084** -1.855** -1.650** -0.897

The Effect of the Majority Vote Rule on Primary Conduct

Table 6: Directors Who Failed to Attend 75% of Meetings   Failure to Attend/ All Nominees Failure to attend+ISS “for”/All nominees Plurality Vote Rule 0.606% 0.405% Majority Vote Rule 0.344% 0.113% Prob. Value 0.000 0.465 Majority Vote Rule (Early Adopter) 0.403% 0.124% Majority Vote Rule (Late Adopter) 0.212% 0.088% Prob. Value Difference Early and Late MVR 0.041 0.507 Directors of MVR firms have better attendance

Table 7: Failure to Attend, Regression Results   Model 1 Model 2 Model 3 Model 4 Attendless75 Attendless75 + ISS WH Rec MVR -0.00397* -0.00292* EarlyMVR -0.00609+ -0.00292 LateMVR -0.00304+ -0.00292+ NOTE: Coefficients for other independent variables omitted from above table. We find no evidence of an effect on failure to adopt a shareholder proposal

The Electioneering and Shareholder Restraint Hypotheses: Majority Withhold Votes Given Primary Conduct http://www.davispolk.com/briefing/corporategovernance/broadridge-explains-its-current-policies-interim-vote-tallies-issuers/

Broader potential effects of MVR MVR firms may lobby shareholders more in close elections MVR firms may lobby ISS more in close elections ISS may evaluate MVR firms differently Shareholders may be more restrained at MVR firms

Fraction of Directors that Received a Majority Withhold Vote Outcome Table 8 Panel A   Plurality Voting Regime Majority Voting Regime N Fraction of Directors that Received a Majority Withhold Vote Outcome p-value ISS WH Rec 3454 0.072 787 0.010 0.000 Attendless75 219 0.137 64 0.016 0.006 Attendless75 + ISS WH Rec 146 0.205 21 0.048 0.082 IP NO 589 0.051 461 0.002 IP NO + ISS WH REC 254 0.118 51 0.020 0.034 Directors of MVR firms who commit an “offense” are less likely to have a high withhold vote

Table 8 Panel B: Majority Withhold Vote Outcome, Firm Fixed Effects Regressions   Model 1 Model 2 Model 3 Model 4 Model 5 Whvote50 MVR -0.00197 -0.00418* -0.00425* -0.00395* -0.00402* ISS WH Rec 0.0697** MVR x ISS WH Rec -0.0652** Attendless75 0.140** MVR x Attendless75 -0.122** Attendless75 + ISS WH Rec 0.204** MVR x Attendless75 + ISS WH Rec -0.146* IP NO 0.0339* MVR x IP NO -0.0280+ IP NO + ISS WH Rec 0.0960** MVR x IP NO + ISS WH Rec -0.0658 Early Adopters: Interaction terms significant at 1% and 5% levels in all 5 models Late Adopters: Interaction terms not significant except in Model 1 , even after controlling for endogeneity through firm fixed effects, given similar conduct, majority voting rule companies have a lower likelihood of receiving a majority withhold vote for than plurality voting rule companies. Given similar conduct, MVR directors are less likely to receive a majority withhold vote

Conclusions and Future Work We find some support for all four hypotheses Importantly, we find some support that is inconsistent with a pure selection story Early adopters are different from later adopters Implications How does corporate governance reform happen? Do investors target leaders or laggards? How do we know if corporate governance reform matters? We may have to separate early adopters from late adopters Early studies may be misleading Potential application of these results to other empirical work – proxy access, separating Chair and CEO, etc. Possible international governance issues – board structure, board diversity