Bufetico
A company on my list once came out with a gap of 2,427% between what my model said it was worth and what it cost. It was trading at 44.20 dollars a share, and the model put it at a small fraction of that.
That’s a broken measurement, and the interesting part is how I knew before I looked at the company at all.
The number that gave it away
The growth rate the model used landed exactly on my ceiling.
I cap growth, because no business compounds at any rate forever, and when a computed input sits right on its own limit, the limit did the work. The number stopped describing the company somewhere along the way.
The company reports in Argentine pesos. Its cash flows grow at something close to the rate of inflation, which is enormous, so the model tried to project growth far above the ceiling and I clipped it. Then it discounted those clipped flows at the peso rate, which is roughly what the peso costs to borrow, and which I don’t clip at all.
Amputated growth against an uncapped rate. Compound that gap year after year and the present value collapses to nearly nothing, and the gap on the screen runs to four digits.
The result described my ceiling. The bank had nothing to do with it.
Why there was a rate to argue about at all
To turn future cash into a value today you have to discount it, and the rate I use is the one the government behind the company’s reporting currency pays to borrow. A Japanese company reporting in yen gets the Japanese rate. A Spanish company reporting in euros gets the euro rate. I went through why a few days ago with Toyota (/blog/sovereign-curve-discount-rate). The short version is that the gap between what Japan and the US paid on their bonds moves a valuation more than most of what the company does.
For most currencies that rate is published and I can read it. For seventeen of them there’s no usable curve to be had.
Seventeen currencies, one hundred and five companies
Argentine peso: 35 companies. Russian ruble: 34. Peruvian sol: 5. Egyptian pound, Colombian peso, Kazakh tenge: 4 each. Georgian lari, Turkish lira: 3 each. Nine more currencies with one or two companies apiece.
One hundred and five companies in total. That’s 0.35% of the list.
Every one of them shows no fair value, no gap, and no gray line on its chart. The square stays gray.
The obvious fix, and why it fails both ways
The obvious fix is to type the rate in by hand. I already do that for five other currencies, taking the last published figure. It’s a pretty reasonable method and it works.
I tried it. Putting the peso in at its own rate is what produced the 2,427%.
So I tried the other direction. Use the rate that country pays to borrow in dollars instead. That fails worse: the capped growth comes out higher than the rate, and every one of those 35 companies lands at the top of the screen, where they’d be the first thing anyone sees.
One setting buries them and the other promotes them, and the business is the same in both. Whatever number I pick for the rate, I’m still comparing a growth figure I’ve capped against a rate I haven’t, and in a country with 200% inflation that mismatch is bigger than anything the company does.
So they stay gray
I decided not to enter a rate for any of the seventeen. The 105 companies get no valuation, and the chart now says so in plain words instead of drawing an empty box: no fair value for this company, so there’s no gap to chart.
It’s the same call I make when a company owes more than it owns and the standard debt measure stops describing anything. When the arithmetic leaves the range where it means something, I leave the square gray. A gray square annoys people. A number they’d use, and then I’d have those companies at the top of the screen again.
Where this breaks
Leaving them gray has a cost.
If there’s a genuinely undervalued company among those 105, my method will never surface it, and someone screening only on colors will walk past it forever. 0.35% is a small share of the list, and every one of them is a company that stays invisible on my screen.
The mismatch is fixable in principle. Either both figures should be in real terms, stripped of inflation, or both in nominal terms with it in. The defect is in the mixing. The missing rate is where it happened to show up first. It isn’t fixed yet, and until it is, entering a rate by hand will produce the same 2,427% no matter how carefully I pick the number.
Update, September 15: there’s another problem with this company that has nothing to do with the peso. It’s a bank, and cash flow measures something else in a bank. I got to that one in its own article: /blog/why-cash-flow-cannot-value-a-bank.
The gray square on those companies means I don’t know, which is a lot harder to sell than a number.