Executives of public companies tend to get a lot inside information about their companies in the course of their day. It would be weird if they didn't. This can make it hard for them to sell stock legally, since selling stock while you have inside information is illegal. And executives should be able to sell stock sometimes, because otherwise they'd have no money, except for their other money.1 So the law allows them to set up 10b5-1 plans, which are automated plans that they put in place when they don't have inside information, and that sell stock in the future according to fixed instructions. So you release earnings, and the next day -- when you're "clean" -- you enter a contract that says, "I will sell 10,000 shares on the first day of every month for the next two years, as long as the stock price is above $55."2

There is a lot of speculation about how these plans can be gamed, and it sure feels like they should be easy to game by setting up 10b5-1 plans and then changing or canceling them when you get inside information, but I think a lot of that speculation is overblown. It's harder than it looks to use a 10b5-1 plan as cover for making trading decisions based on undisclosed inside information.3

But the other way to game 10b5-1 plans is by making disclosure decisions based on your trading plan. So you hold back on disclosing good news until you're just about to sell under your 10b5-1 plan, and you hold back on bad news until just after you've sold. This is surely worse than just monkeying with the 10b5-1 plan: It's not just insider trading, it's also keeping all shareholders in the dark about material news.

Jesse Eisinger thinks he's caught the chief executive officer of Questcor Pharmaceuticals doing it:

Good things keep happening to Questcor in the middle of the month. Here’s what’s notable: The middle of the month just happens to be the time that the company’s chief executive, Don M. Bailey, sells stock through his regular selling plan.

He lists six examples over the last ten months of Questcor releasing good news just before Bailey's sales, which from Bailey's Form 4s seem to be scheduled for the 13th of each month (or the next business day, if it's a weekend). Some of these announcements sort of had to be made when they were made (e.g. filing presentations made at banks' conferences, which aren't scheduled around Bailey's trading plan), but the timing of others does look pretty discretionary (e.g. disclosing a new "strategic advisory committee" on December 11). But as evidence that list of anecdotes seems sort of meh? So I went into casual debunking mode and did this:

Whoops! That's more of a rebunking.4 The stock really does tend to go up a lot more in the lead-up to Bailey's stock sales -- an average of 1.5 percent a day over those four days a month -- than it does on other days (an average of 0.25 percent a day).

A more granular breakdown is ... more granular, but if you squint there does seem to be a sort of a pattern, with the days around when Bailey traded (say days 10-15) being mostly very positive and the days surrounding those days -- the days into which you'd "push" disclosure, if you were being sneaky -- being more mixed-to-negative:

So I guess I'll say, endorsed, this looks fishy.

That said, that doesn't mean that it is fishy. A just-about-plausible model of stock price moves is that they consist of:

  • a random number generator, plus
  • manipulation.5

It would be nice to think that the random number generator will generate random-looking numbers, and that any fishy-looking patterns -- say, the stock price tending to go up right before the CEO sells stock -- reflect the manipulation. But of course there's no guarantee that a random number generator will generate random-looking numbers. If you run a random number generator enough times, some of those times will produce things that look like patterns. With thousands of listed stocks in the U.S., all bopping around randomly, odds are that some of them outperformed in the middle of the month in 2013. And the odds that one of those will have a CEO whose 10b5-1 plan calls for trading in the middle of the month are, I mean, I don't know, probably not that low.

The point is that finance creates an amazingly rich set of data. All these stocks, trading every day, and someone writing down all their prices. So many numbers! Surely they must mean something! The temptation to mine that data -- to see patterns where there is only randomness -- can be overwhelming.

When investors do this, we call it "technical analysis."6

But sometimes regulators and journalists and other investigative types can be seduced as well. Don Bailey's timing looks really unusually lucky, but in a market with thousands of CEOs, someone's going to have unusual luck. Is he that unusually lucky guy? Or is his company in fact fiddling with when it announces news in order to help him sell stock at high prices? I don't know. And we can't tell just from noticing his luck.

To contact the writer of this article: Matt Levine at mlevine51@bloomberg.net.

To contact the editor responsible for this article: Tobin Harshaw at tharshaw@bloomberg.net

1 "These people have expensive lifestyles."

2 Really you can have a grid -- "sell 1,000 shares if it's above $55, or 2,000 if it's above $60, or ..." etc. Or you can do lots of other more complicated things. So it's not necessarily obvious just from filings (which disclose how much stock was sold) what the 10b5-1 plan said, whether it was amended, etc.

3 Here is a post where I explain how to play this game, and here is a footnote to that post where I explain why you shouldn't and why it probably doesn't work.

4 Here's daily outperformance vs. the S&P 500 index, not much different:

5 Oh also fundamentals, sure. I'm only aiming for just-about-plausible with that model.

6 OH NO I'M KIDDING, COME ON, DON'T EMAIL ME ABOUT THIS.