Is the Tesla Stock Split Paradox Really Paradoxical?

David Hirshleifer tweeted :

Evidence of stock market inefficiency: A year ago Tesla’s stock price rose 18% in the two days after its 5:1 split announcement—$50b rise in market value. Hard to explain in terms of rational signaling or increase in liquidity.

Professor Hirshleifer cited the Cornell Capital Group’s nice essay, “The Tesla Stock Split Experiment“. That article is very nicely written, but it has no theory and it focusses on how no new information became public at the time of the stock split, and wonders how the stock split could be a credible signal of anything. My answer below is basically, “Because why would Tesla want to fool anybody?”

Suppose the value of Tesla is 10 or 190 with equal probabilities, and boss Musk knows the truth but the public does not, andwill only find out the truth next year, so the stock price is 100. Quite aside from any signalling, Musk would like to do a stock split now if the value is 190, but not if it is 10. If the value is 10, a stock split costs him personally 5 in utility; if it is 190, the split benefits him 8 (why does he care? aesthetics maybe).

With just what we have so far, if there is a stock split, the price will immediately rise from 100 to 190. So, no puzzle yet.

What isn’t in the model so far, though, is the common sense idea that Musk would like to see the stock price go up. But this needs some thought. Remember, the price will only go up for one year if Musk tries to fool the public. It will rise from 100 to 190, and then fall to 10, and he will have lost his 5 for nothing. So what “common sense” says to add to the model is that Musk would be willing to pay 5 for a temporary rise in the stock price.

Would he? Why would Musk like there to be a temporary rise in the stock price? If he could sell his own stock while the price was temporarily high, he’d like that, but it’s realistic to think that if he does that, he will thereby reveal the truth and the price will fall to 10 immediately instead. He can’t sell. So it has to be something else.

How about pride? Could it be he would like having a price of 190 a year because it would make him look good? Maybe, but he’d look extra bad if he pulled this trick, and the pain of going from 190 to 10 would be worse than of going from 100 to 10.

How about pay? If Musk were paid with stock options, he could pump up the stock price, get a big bonus, and then not care if it dropped after that. That’s a big problem with most corporations, but not for Tesla, because Musk is an owner-operator, not a president trying to fake out the board of directors.

How about operations? If the stock price is 190, it will be easier to get bank loans and sell bonds, and very profitable to sell new stock issues. This is more plausible. But the banks, bond-buyers, and stock-buyers all know this, so it’s like with Musk selling his own stock– if he does a stock split and immediately tries to get extra capital, it won’t work.

So the key to showing that the market is inefficient here is to show why Musk would like to see the price rise temporarily, knowing it will fall back later.

David Hirshleifer and I both joined UCLA’s business school in 1984 fresh out of grad school, so when he refers me to “The valuation effects of stock splits and stock dividends,“ Journal of Financial Economics, 13, Issue 4, December 1984, Pages 461-490,, it’s quite nostalgic. Grinblatt, Masulis, and Titman were the “senior juniors” then, and were very kind to us, something especially appreciated by me, who was not even in the finance department. They write as one of the possible explanations for stock splits what I brought up above:

It might be that managers do not deliberately attempt to convey information when they
announce a split or stock dividend. For instance, conventional wisdom suggests that managers
split to keep the price of their shares within a customary trading range. Given the costs associated
with splits and reversals, managers with unfavorable inside information might decide not to split,
even if their firm’s stock price is high, because they expect that future events will force the price
of the split shares to fall below the customary trading range. Investors, observing the correlation
between splits and subsequent stock performance, could then use the split announcement to draw
inferences about this information.

One drawback to this ‘trading range hypothesis’ is that the managers of some overvalued firms
might have little concern about the trading range of their firm’s stock and split simply to obtain a
temporary increase in its price (e.g. when the firm plans to raise capital or when the manager
plans to reduce his stock or stock option holdings in the firm). Thus, the above scenario
implicitly rules out such incentives or ascribes to investors the ability to discriminate between
these two types of managers.

The 1984 paper is empirical, though. It doesn’t go deeper into the theory, and it uses regression analysis to look at average behavior, not at individual companies. That’s standard, and it’s good for analyzing systematic anomalies, but it’s bad for looking at individual anomalies– a bit like IQ or education being good for explaining group differences in income (high t-statistics) but bad for explaining individual differences (low R squared). That’s because it could be that the Trading Range Hypothesis explains some cases perfectly well– Tesla perhaps– even if it is a flop for explaining most stock splits. It is a good theory for Tesla because it seems, at first glance, that Musk has no incentive to want to convey falsehood and the market can see that, but it is a bad theory for most stock splits because most corporations are run by CEO’s who do have strong and visible incentives to temporarily inflate their stock price.