Management Always Wins the Close Votes

One of the advantages of guest blogging vs. regular blogging is that you can talk about your own work for a high percentage of your posts.

Today,  I'm going to take advantage of my handy stock of posting topics and start talking about what I'm working on, which for the last few months has been corporate voting. (Click here for a draft.) My research takes as its starting point a point developed by Jason Snyder (paper available here) in the context of voting in democratic elections-- if you examine a small interval of possible vote percentages received, then the number of votes in the first half of that interval should be approximately the same as the number of votes in the second half of the interval. For example, suppose you are looking at all U.S. House elections where the winner received between 60 and 61 percent of the total vote and suppose that there 400 elections in this interval (i.e. where the winner got between 60 and 61 percent of the total vote). We would expect there to be approximately the same number of votes, 200, (with allowances for random variation) between 60 and 60.5 percent as between 60.5 percent and 61 percent (which should also have something around 200).

This is exactly what you find in the vast majority of votes received intervals that you look at. There is one exception to this-- when the difference between the first half of the interval and the second half of the interval is a practically important one. The obvious example of this is around 50% (or some similar plurality). Very close votes should effectively be a coin flip (think of Florida in 2000), yet Snyder finds that incumbents wins very close votes more often than they "should"-- more often than half the time.

Snyder's paper and others in the burgeoning economic literature on corruption and irregularities got me to thinking-- how does management of publicly traded companies do in very close votes?

For example, most stock option plans need to be approved by the majority of shares voting. Again, without perfect information very close votes on executive compensation plans should be a coin flip. I find that this is decidedly not the case. Management wins these close elections almost all the time, to a degree that should occur by chance less than 1 in 1 billion times. For example, management receives between 49.5 and 50% of the vote (a close loss) just two times in the data set, while management receives between 50 and 50.5% of the vote ( a close win) 18 times. The ratios are similar for wider intervals.

The next question is why management does so well. There are a few reasons, some more innocuous than others. First, management gets to choose the the issues that come before shareholders, so its no surprise that they win more often than they lose. The problem with this explanation is that plans must be put on the ballot many days before the polls close, when its hard to believe that management has such a great sense of the level of support (down to a half percentage point).

Second, management campaigns harder in close votes, and can spend corporate money on campaigning. In addition, management gets updates on how the vote is progressing that help it determine how many votes it needs. Here too, this would explain a good deal of asymmetry, but many votes come in at the last minute, and, even after the vote is over, there is some uncertainty about what votes should and should not count (think hanging chads). (Marcel Kahan and Ed Rock are currently working on a paper that examines the "hanging chads" of corporate voting, detailing all the ways in which the corporate voting system is flawed and filled with "noise".) How can management be so good at winning what should be a process filled with uncertainty?

Other explanations, for which I have little or no hard evidence, suggest greater shenanigans. For example, management could observe that it is about to lose a close vote and call up a large no-voting shareholder and give them inducements to change their vote, such as money or future business. (This is alleged to have happened in the HP-Compaq merger vote.) Alternatively, management or its agents, which have a large degree of control over the vote counting process, could "cook the books" by selectively counting the votes or keeping the polls open until the moment when they get just over 50%. This explains why management would be so good at winning really close votes, though I must again emphasize that I have no hard evidence for the existence of such activities.    

I discuss what this means for policy in the paper. Some obvious recommentations include more standardized and transparent vote counting procedures and changes to the availability of real time information about the vote. Even if there are no overtly pernicious activities going on, such changes would increase confidence in the corporate voting system.