SYNOPSIS: Krugman explains why Dow 36K is crankonomics
Some of my correspondents think I gave Glassman and Hassett too easy a ride in my Sunday column. Maybe I did, partly because the column was more about the lockup of Republican economists than about what was, after all, only a book. But here, anyway, is my understanding of where the issues are.
1. What is the Glassman-Hassett argument?
Imagine the whole U.S. corporate sector as if it were a single company. And imagine that this company - and the economy - will grow steadily forever, say at 5 percent nominal (3 percent real plus 2 percent inflation). Suppose also that the interest rate is 6 percent. What is this company worth?
The answer should be that it is worth 100 times dividends. A dividend that grows at 5 percent per year, discounted at 6 percent, has a present value of 100 times this year's level.
The G-H argument is that historically people have discounted dividends at a much greater rate than warranted, because they perceived a lot of risk, which wasn't actually there; so in reality stocks should be valued much more highly than anyone now thinks - and the claim is that the right number is something like 36,000.
2. What is the glaring error?
In the original WSJ articles, that number - Dow 36,000 - was calculated as 100 times earnings. Now earnings are not the same as dividends, by a long shot; and what a stock is worth is the present discounted value of the dividends on that stock - period, end of story.
Glassman and Hassett have repeatedly - and vituperatively, in Glassman's case - insisted that they are not confused about this point. But it is hard to get away from the fact that their number corresponded at the time of writing to a P/E of 100, which makes it hard to believe that they did not think that earnings were the right thing to discount.
Now their book does offer an alternative calculation, in which they discount dividends plus repurchases of stock (which do return cash to investors, and increase the value of the shares remaining by allowing dividends per share to grow faster than total dividends). This leads them to an estimate of Dow 18,000. So why didn't they retitle the book?
Aside from what one suspects to be the likely explanation - that it would have been too embarassing - they claim that the free cash flow of companies is larger than the cash they actually return to investors. I have to admit that I don't understand this. If it is really free, that is, is not needed for the expansion of the business, where is it going? And anyway, in the end a stock is worth the discounted value of the cash the investor actually receives - not what he hypothetically might have received. So the supposed free cash flow does, somehow, someday, have to show up in actual cash flow to investors somewhere.
The authors offer what they claim is a counterexample: companies like Microsoft, that do not pay dividends. But this is, I think, silly: what makes Microsoft worth so much is the expectation that stockholders will at some future date either receive large dividends or have their stock bought back by the company at high prices. Or to put it differently, in the case of Microsoft the expectation is that dividends will grow more rapidly than profits in the future (easy when they start from zero). There are other stocks that pay large dividends but do not have high prices (utilities), because those dividends are not expected to grow rapidly.
To claim that the right valuation, even accepting the rest of the G-H argument, is 100 times something more than dividends plus repurchases, you must claim that the cash flow to investors from the corporate sector as a whole will grow faster than profits. Maybe - but that is an argument they do not make explicitly, and it also undermines the steady-state assumptions that are the basis of the calculation.
I have to admit that despite numerous belligerent explanations from Glassman, and my own conversation with Hassett, what I still think is that they simply made a mistake in their original argument, and have since tried to throw up a smoke screen to cover up that mistake.
Why haven't more commentators picked up on this mistake? My other uncharitable suspicion is that it never occurred to them that the authors might be confused about such a simple point; and in the book the authors claim to be taking it into account - before suddenly jumping back to a valuation of 100 times earnings, after all.
3. What are the other problems?
There are at least three other reasons not to believe in the book's view about stocks.
First, the assertion that there should be a near-zero risk premium is based on the historical performance of stocks. But the authors claim that the historical valuation of stocks was all wrong, and that a new paradigm is needed. Um, wouldn't this change not only the prices of stocks but their riskiness? In fact, anyone who tries to do something like the G-H calculation quickly realizes that it is incredibly sensitive to the assumptions; lower the growth rate or raise the interest rate a bit, and the "right" value of stocks changes dramatically. Doesn't this suggest that a decline in the risk premium will itself increase the actual riskiness of stocks, and hence that the assumption that the right risk premium is near zero is wrong?
Second, a related issue: Dow 18,000, or whatever the right value if you avoid double-counting turns out to be, is a calculation based on the current interest rate. But a further huge rise in the stock market would tend to increase both investment and consumption demand - and therefore raise the real interest rate. So you really need to do an equilibrium analysis, which would surely give you a lower number.
Third, the deepest issue: today's stocks are not a claim on the earnings of the U.S. corporate sector into the indefinite future. They are a claim on the earnings of today's corporations into the indefinite future. That's a big difference, if you look far enough ahead: unless something terrible happens, U.S. companies will be earning a lot of money 70 years from now; but much of that money will be earned by companies that do not now exist, or at any rate are not in the Dow or even in the S&P 500.
Now you might say, who cares about the world 70 years from now? The answer is, Glassman and Hassett care - because their formula places a huge weight on the very distant future. If you take the present discounted value of an earnings stream that supposedly grows 5 percent a year forever, and is discounted at 6 percent, it is equivalent to taking the value of a constant stream discounted at 1 percent - which is to say, half of the value comes from cash flow more than 70 years in the future, 1/4 from cash flow more than 140 years in the future. If this seems silly to you, you are right.
When all is said and done, Dow 36,000 is a very silly book; and the attempts of Glassman, at least, to deny the patent silliness are borderline dishonest. But I am prepared to view Hassett's role as a youthful indiscretion.
Originally published in The Official Paul Krugman Site, 2.00