Bufetico
In 2021 the ten-year government bond in Japan paid about 0.05%. The American one paid 1.52%.
Written down, that gap looks small. Inside a model it isn’t, because the rate gets applied once for every year you look forward, and the effect compounds. Far enough out, two identical streams of cash come back with values that sit far apart, and the longer the forecast, the wider the gap.
So when a model takes Toyota’s yen and puts the American rate under it, it’s measuring something else and calling it Toyota’s value.
Which rate belongs to which company
That rate stands for what you could earn without taking any risk, in the currency the cash actually arrives in. The last part is the whole point, and it’s the part that gets dropped.
A company that reports in yen produces yen. The alternative to owning it is a Japanese government bond. The American rate describes the choice facing someone holding dollars, which is a different decision about a different currency.
Government rates in Japan, Europe and the United States have been far apart for years. Use one rate for all of them and the low-rate countries come out looking cheap and the high-rate ones come out looking expensive, and that has nothing to do with the businesses. It comes from where they file their accounts.
Any screen that ranks companies across borders on one rate is ranking them, mostly, by their currency.
Reported in one currency, traded in another
There’s another distinction that’s easy to miss and matters as much.
The currency a company trades in and the one it reports in are often not the same. A Chinese company listed in New York trades in dollars and reports in yuan. The accounts are in one currency, the share price in another.
The rate follows the reporting currency, because that’s where the cash exists. Comparing the result with the share price then needs a conversion. Skip it and you’re comparing a value in yuan with a price in dollars, and you get a huge discount that’s all exchange rate.
So I use a government curve for each currency I can get, matched to the currency each company reports in.
Where this breaks
Not every currency has a curve. Where the government bond market is thin, or barely exists, there’s no rate to work with, and I don’t borrow one from somewhere else. That leaves 105 companies with a gray square, and the day I tried to set a rate by hand instead, one of them came back 2,400% undervalued, which is its own story.
A stale rate is worse than a missing one, because it doesn’t look wrong. So only rates published recently get used. A frozen rate from three years ago produces a precise answer that happens to be wrong, and nothing on the screen tells you it’s old.
One question to ask
When you compare companies across countries, ask what rate sat under each valuation. If it was the same rate for all of them, the comparison is measuring something other than what it claims.
The currency is one of the biggest assumptions in the whole model. The tool almost never tells you which one it used.