So here’s kind of a silly way to conceive of yesterday's Berkshire Hathaway buyback, complete with Google Docs spreadsheet. Because everybody likes Google Docs spreadsheets.
Berkshire is a company that has assets, which have a book value that tracks their actual value more or less closely. More closely: the book value of cash or mark-to-market equity securities is, at least at quarter end, equal to their actual value. (Really it’s always equal to the actual value but book is only disclosed at quarter-end.) Less closely: the book value of property, plant and equipment is their cost minus some arbitrary depreciation. Berkshire’s net book value is about $163 billion. With 1.65 million Class A share equivalents, that gets you a book value of around $98,715. Buffett is willing to buy at or above 10% over that, or $108,587. That number is a rough proxy based on last quarter’s 10Q, and equities are down since then, but whatever.
You can divide Berkshire’s book value into two components: cash and not cash. Of course some of its non-cash stuff is just securities lying around that are effectively cash equivalents, and some of its cash is needed to pay insurance claims and put coal in the railroads and whatever. So let’s use two slightly different names: “free cash lying around” and “everything else.” You can quibble over what the numbers are; Whitney Tilson thinks that there’s $77 billion of cash lying around, counting all fixed income securities in that bucket. That seems aggressive to me – insurance business! – so I’m going to say there’s $40 billion of free cash, which is just the “cash and equivalents” line for the insurance business minus a $3 billion cushion. But it doesn’t matter, pick a number. It's way larger than $20 billion (since Buffett said he'd always keep that much as a cushion) and probably no larger than $77 billion.
Okay. Now pretend that Berkshire stock trades at a multiple of book value – which is not a crazy thing to pretend, since after all Buffett sort of does. Then, by arithmetic, you can decompose that multiple into (1) a multiple of the free cash and (2) a multiple of the “everything else.” In a sensible normal company you might think “cash is worth no more in Reed Hastings’s hands than in mine, and probably less” so you slap a 1.0x multiple on the free cash. Then you can solve for the everything-else multiple. In a normal company that multiple is probably (well) above 1.0x, because you’d hope a normal company is making more productive use of its assets than their depreciated purchase price. But that's not a necessary relationship; banks, for instance, whose balance sheets tend to be closer to mark-to-market and who sometimes enjoy destroying value, can trade below book.
But it’s just arithmetic; BRK/A, the day before the buyback announcement, was trading at around $100k and so you can math out the “everything else” multiple to around 1.02x.
Now what is your theory of the motivation for the buyback? Well if you just think that those multiples come from reality then they should stay the same – just reducing amounts shouldn’t matter. In that case, buying stock at $108k when the “right” price was $100k destroys value (column E of the spreadsheet, if you need proof). So that’s dumb. And I guess Buffett doesn’t do dumb.
So you can have two alternative theories (or a mix of them): the multiple on everything else should move, or the multiple on cash should move. The first theory implies that Buffett knows – as the market doesn’t – that the market is undervaluing the insurance/railroad/whatever businesses lurking in Berkshire Hathaway. Then Buffett is just spending cash to buy that business, because he thinks it’s the cheapest business out there. That makes sense and is sort of a normal justification for buybacks. By my math (column F) his buyback implies a 1.13x book multiple for the insurance/etc. business, which is reeeeeally cheap to historical levels.
So, entirely plausible. Though "cheap by historical standards" is not, of course, the same as "cheap by absolute standards" - and you could quibble about why Buffett should know so much better than the market that his operating businesses are undervalued. But it's a respectable position.
My theory, if you care – which you probably shouldn’t – is slightly different. In my mind it’s insane to value “cash in Warren Buffett’s hands” at its nominal value. If I have $100 I can spend it buying Rick Perry futures on Intrade. (I actually did this.) If I give it to Warren Buffett he can put his guarantee on some bank’s box at well-off-market terms. He gets opportunities that I can only get by putting my cash with him.
But that’s only true if he can invest it. A popular take on this buyback is that Buffett doesn't see much opportunity for his cash right now. Even more important - as Tilson accurately points out - he really never sees that much opportunity to spend $40 or $60 or whatever billion dollars. His wheelhouse is the $5-to-$15 billion zone where he can add value. He wants to get back to around $20 billion in cash so that he can add his traditional value to that cash. $40 or $60 billion to Warren Buffett isn't worth much more than it's worth to me (a lot, if anyone's offering); $20 billion in Warren's hands, however, will find itself in the way of some attractive opportunities.
Take that to the extreme and assume that the multiple on railroads/whatever remains constant at 1.02-ish, a little more, a little less, whatever. Assume that Buffett buys back at 10% above book and that he spends $20 billion out of his $40 billion in free cash, getting back to his preferred cash-for-opportunities of $20 billion. And assume that the stock stays where Buffett buys it (it's above the supposed target now, and not clear he's actually bought any stock). Then how much is a dollar of the remaining cash in Buffett's hands worth to the market? About $1.69.
That's an extreme view (assuming that BRK's operating businesses aren't undervalued at all), but if you believe that Buffett is regularly offered opportunities to buy stuff worth $150 for $100, it's not all that far off.