For years, John Y. Campbell and Robert J. Shiller have been calculating long-term P/E ratios. When they were invited to a make a presentation to Alan Greenspan in 1996, they used the statistic to argue that stocks were badly overvalued. A few days later, Mr. Greenspan touched off a brief worldwide sell-off by wondering aloud whether “irrational exuberance” was infecting the markets...
Today, the Graham-Dodd approach produces a very different picture from the one that Wall Street has been offering. Based on average profits over the last 10 years, the P/E ratio has been hovering around 27 recently. That’s higher than it has been at any other point over the last 130 years, save the great bubbles of the 1920s and the 1990s. The stock run-up of the 1990s was so big, in other words, that the market may still not have fully worked it off.
As we all remember, the "worldwide sell-off" in the wake of Greenspan's "irrational exuberance" comments was followed by the biggest stock-market bubble of all time. But have a look at Leonhardt's chart – it turns out that even after the bubble burst, long-term p/e ratios still remained at or above the "irrationally exuberant" levels of 1996. Greenspan's ideas of what was overpriced seem, in retrospect, to have been something of a floor in terms of how far the market could drop.
And I simply don't understand what Leonhardt is talking about when he refers to the market "working off" its bubble. Bubbles aren't "worked off". They burst. Dramatically. And the market, according to all economic received wisdom, tends to overshoot, not undershoot, in such a scenario – in other words, far from falling too little, it tends to fall too much.
Maybe there's some kind of meta-bubble explanation. Stocks in general got bubbly in the early 90s, and then a tech bubble grew out of the more mainstream bubble. The tech bubble burst, but the mainstream bubble didn't. But I don't buy it myself.
Or maybe, as Leonhardt himself proposes, there really has been a secular change:
Over the last few years, corporate profits have soared. Economies around the world have been growing, new technologies have made companies more efficient and for a variety of reasons — globalization and automation chief among them — workers have not been able to demand big pay increases. In just three years, from 2003 to 2006, inflation-adjusted corporate profits jumped more than 30 percent, according to the Commerce Department. This profit boom has allowed standard, one-year P/E ratios to remain fairly low.
Going forward, one possibility is that the boom will continue. In this case, the Graham-Dodd P/E ratio doesn’t really matter. It is capturing a reality that no longer exists, and stocks could do well over the next few years.
Leonhardt felt the need to say that maybe stocks aren't over-valued if profits keep growing rapidly (sounds like Alan Greenspan in the 90s). Well don't hold your breath on that one. Profits peaked in the 3rd quarter of 2006 and were down sharply in the 4th quarter of 2006 and the first quarter of 2007. It's always possible that they will bounce back, just like it's possible that President Bush will sign the Kyoto agreement, but I don't know anyone who will bet on either event.
I'm more sanguine than Baker on this one. Leonhardt isn't talking about quarter-to-quarter fluctuations in corporate profitability, and he's certainly not saying that the business cycle has been repealed. He's just saying that maybe there's been a permanent move whereby capital gets a larger share of the total economic pie, relative to labor, than it has done historically. If that were the case, then it would indeed be silly to compare stock prices today to those companies' earnings ten years ago, before that change really kicked in.
Remember that Leonhardt is looking at the price of stocks today, divided by average earnings over the past ten years. Given that recent years have seen much higher profits than those recorded a decade ago, the denominator of that ratio is going to increase steadily even if quarterly earnings do decline over the next year or two.