“Looking for a juicy [hostile] takeover candidate? A new report suggests you might find one at Fifth Third Bancorp, the utility holding company Ameren or ConAgra Foods,” which shows you how much that new report knows. Oh, you’re gonna launch a hostile takeover on a regional bank? A regulated utility holding company? Let me know how that works out for you. Over here, I’m going to focus my imaginary hostile takeover activity on industries where hostile takeovers are, y’know, possible.1
That’s from this DealBook piece about a company called RotaryGallop and a report they’ve done on corporate vulnerability to activists. The report can be yours for $4,550, and needless to say I admire their gumption for creating a measure of corporate vulnerability that seems shall we say somewhat unmoored from reality, writing it down in a report, charging companies $4,550 to read it, and getting it written up in the Times. Still:
To prepare the report, Rotary Gallop analyzed share ownership at 459 of the companies in the Standard & Poor’s 500-stock index — essentially all those without multiple share classes. It then used December 2012 share ownership data from FactSet Research to model millions of hypothetical up-or-down votes, determining which shareholders could most often tip the outcome one way or the other. …
The calculations were straightforward, but complex, [RotaryGallop CEO Travis] Dirks said.
“You could sit down and do it yourself for a very small company,” he said. “The difficulty is that the number of permutations you have to deal with are two to the number of shareholders — it just blows up very fast.”
Dell — which faces competing bids from management, Carl C. Icahn and the Blackstone Group — has some 900 shareholders that have publicly disclosed their ownership stakes, so the potential vote possibilities would number 10 followed by 271 zeros, Mr. Dirks said.
By comparison, the atoms in the known universe are believed to number about 10 followed by 80 zeros.
As far as I can tell – and I confess I haven’t shelled out the $4,550 – RotaryGallop assumes that each shareholder has a 50/50 chance of voting either way, and then Monte Carlos out the “a priori odds” of a big shareholder winning a vote, that is, the odds that that shareholder will pick up enough 50-50 voters to bring you to a majority. It’s like Nate Silver but without the polling data. The more you own, and the less concentrated the remaining ownership is, the more likely that the random chips fall your way, and the more likely you are to win whatever the thing that you want to win. Which I think was a takeover? Or for that matter a proposal to split the CEO and chairman roles, or that Japanese toilet thing. The content is irrelevant.
You could … quibble with that? And with other things? One might, for instance, raise an eyebrow at the main underlying assumption, which is basically that every shareholder is equally likely to vote yes or no on every proposal.2 Or you could note that it takes no account of the fact that a lot of shareholders abstain in shareholder votes.3 Or that it relies on disclosed shareholder positions, which both (1) ignores undisclosed shareholder positions (private individuals, etc.) and (2) doesn’t reflect activity subsequent to the start of an activist fight (arbitrageurs, etc.). So for instance it vastly understates the complexity of the Dell situation: there are way more than 900 shareholders, and some of the disclosed shareholders have changed their positions while other undisclosed ones have moved in, so the potential vote possibilities number not only more than the number of atoms in the universe but more than the number of times that comparison has been used to impress people with the bigness of a number.
But those quibbles aren’t entirely fair. When these guys say “so-and-so has an 80% a priori probability of winning a proxy fight with Company X,” they don’t mean that so-and-so has an 80% probability of winning a proxy fight with Company X. They mean that they have quantified the concentration of Company X’s shareholder base in a certain pre-defined way, and that the number expressing that quantification is 80%. The number is pretty arbitrary but it corresponds in some loose way to some context-free probability of activist success. So there’s that.
I think I’ll allow it? People like numbers, man. Quantitative finance – and Nate Silver, I guess – has had an effect on companies similar to the CSI effect on juries: it’s led to an exaggerated belief in the possibility and usefulness of putting a number on uncertain human behavior. You sell companies enough swaps, with precise probability distributions of what their outcomes can look like, and they’ll start to expect similar math in more loosey-goosey areas.4 I have no doubt that plenty of raid-defense bankers have been asked the question “well, given our shareholder base, what is the probability that so-and-so could win a proxy fight? Give me a number” and said “oh, you know, maybe 40%?” Eventually that starts to feel insufficient.
And so in swoop RotaryGallop to fill the niche, applying the standard quantitative-finance toolkit – assume everything is i.i.d., run some Monte Carlo simulations, draw a graph – in a new domain. And they can give you an answer – one that is quantifiable and plausible enough to charge a lot of money for, but counterintuitive enough to make you feel like it’s worth the price. Everyone sleeps better at night. Except maybe Ameren. They probably weren’t sweating it to begin with.
1. I have a cranky banker’s assumption that hostile takeovers in heavily regulated industries are … uncommon. And that intuition seems to be right. A Bloomberg MA search for utilities hostile deals [Country: Target, US; Nature of Bid: Hostile, Unsolicited, Unsolicited to Hostile, Friendly to Hostile, Hostile to Friendly; Public Target; Utility Sector Target] reveals eight attempted deals in the last ten years, all terminated (and most merchant generators rather than regulated utilities). A similar search but for regional and super-regional banks reveals eight attempts, six of them failed, one completed (in 2003), and one pending (PacWest Bancorp and First California). Lazy Googling finds one electric or gas utility hostile takeover from 1960 to 1990, and one subsequent attempt, in 1996, as well as a Wikipedia claim that SunTrust’s hostile bid for Wachovia – sparked by a previously announced friendly transaction – was “the first hostile takeover attempt in the banking sector in many years.” It failed.
So as far as I can tell, in those two sectors combined, in the U.S., the rate of successful hostile takeovers is a little under one every ten years. But hey y’know good luck with that.
Incidentally the full top ten list of “Most vulnerable to activist investors” includes Wisconsin Energy, Fifth Third, Ameren, Regions Financial, and Con Ed – so five banks + utes out of ten names.
Dirks: Control is itself an objective number without assumptions beyond a shareholder-list / Cap-table. Control is literally the percent of all possible outcomes in which the result of the vote/election is decided by the shareholder’s vote. In the link you will find a second graph, in which I give the chances of winning a shareholder vote. Here there is an assumption that each shareholder is as likely to vote one way as another. There is no doubt it is a softer number, but I report it because people can understand what it means much faster than control.
Guy: You are making some very naive assumptions about independence … A good followup would be to go out and actually measure the distribution of voting outcomes rather than making a strong claim based on the most tractable assumptions you can find.
“Making a strong claim based on the most tractable assumptions you can find” of course describes [__]% of quantitative finance, you fill in the blank with your own preferences, the default is 50.
3. How would you improve the calculation to do that? “Well you could just say each shareholder is equally likely to vote yes, vote no, or abstain.” “But that’s nuts.” “True, but the assumption that each shareholder is equally likely to vote yes or vote no is almost equally nuts.” I suppose you could take the average number of abstentions and give each shareholder that chance of abstaining (and otherwise equal chances of voting yes or no?). Or like figure out who typically votes and who abstains. Seems hard.
4. When I was a banker I sometimes helped companies calculate the probability – for tax, accounting, etc. purposes – that an option would end up in the money, which is totally a respectable thing even though it’s almost as incoherent. (I mean. The risk-neutral probability of expiring in the money is a respectable thing. It’s just not what you want, when you’re asking for it. Nor is it necessarily what you get!)