In defence of the Shiller p/e

By Buttonwood

EVER since the publication of the book Irrational Exuberance, there have been critics of Robert Shiller's cyclically adjusted p/e ratio. The chart here is a reprint from my post two weeks ago but you can find all the data on the Professor's website.

It wasn't actually a new idea on Shiller's part. Ben Graham, the value investor who was Warren Buffett's guru, had suggested a similar measure, involving the averaging of profits over an extended period to smooth out the effects of the economic cycle. Investors were paying little attention in the late 1990s. As the chart shows, the US market in the late 1990s was even more overvalued than it was before the crash of 1929. This was not a welcome message at the time when analysts talked of a "new era" and even speculated that the cycle had been abolished.

But even though the Shiller p/e accurately predicted that the market was overdue for a fall, there are still many critics who today refuse to accept its message. That is because, as you can see, the ratio is close to one of its 20th century peaks in the mid-1960s, which preceded the long bear market of the 1970s.

The ratio presupposes that profits revert to some sort of a mean, at least as a proportion of GDP. Some critics argue that things have changed in that capital is strongly placed relative to labour, thanks to factors such as competition from the Far East and the decline of trade union power. But even if that were the case, the ratio ought still to be mean-reverting. That is because of the nature of capitalism itself. If profit margins are high, then new businesses will be created to try to capture those profits (or existing businesses will expand); the resulting competition will drive margins back down. High profit margins could be maintained in a world of monopolies but globalisation, the same factor cited by profit bulls, erodes monopolies; if Chinese competition is depressing wages, it is also affecting the prices charged by western manufacturers.

Stockmarket bulls prefer to look at the prospective price-earnings ratio, based on next year's profits. But there are many problems with this. First, analysts almost always forecast double-digit profits gains but profits cannot grow that fast in a world of 5-6% nominal GDP growth. The stockbrokers are very bad at forecasting downturns. In addition, bulls often use a sleight of hand that compares the prospective p/e (based on higher earnings) with the average of the historic p/e, based on the previous year's figures. Such comparisons are designed to make the market look cheap.

A more plausible criticism is that profits in the early 20th century were not calculated on a similar basis, so one is comparing apples with pears. But this argument is also unconvincing. After the events of Enron and WorldCom, one can hardly claim that profits are not manipulated today; surely modern executives, motivated as they are by share options, have the incentive to overstate profits whereas those in the early 20th century might have been trying to avoid the taxman and thus understating them.

A related argument is that valuations from the early 20th century are "irrelevant" because of world wars, depressions and the like. One could retort that valuations from the 1980s and 1990s are irrelevant because central banks were inflating a bubble. But the better reply is that one should only deal with the data presented; not exclude chunks of it. We can't predict events. People in 1910 didn't know they were going to have a world war and a depression in the next 25 years; Japanese investors in 1990 might have though they were set for two more decades of prosperity but they were wrong.

For those who doubt the valuation measure, why not look at other long-term approaches. the dividend yield is not 100% reliable; in 1932, at the market low, the stockmarket yields was very low because dividends had been slashed. But by and large, a high dividend yield is a sign of a cheap market and a low yield an expensive one; the yield on the S&P 500, according to ThomsonReuters is 2.3%, at the bottom end of the range. (Careful about arguments that you should add in a bit for share buybacks; you should offset such purchases with issuance of new equity in the form of options.)

Another approach is the q ratio, outlined by Andrew Smithers and Stephen Wright in their book Valuing Wall Street, which came out at the same time as Shiller's opus. This compares the value of the market with the net replacement cost of corporate assets. It too is a controversial measure because of data issues but it has a mean-reverting logic of its own. When market prices are above asset values, it will be cheaper for investors to buy assets directly than it will be for them to invest in equities; arbitrage will eventually drive the two back in line. And the Q ratio, although compiled on a different basis from Shiller's approach, shows a remarkably similar picture of peaks and troughs, including the current overvaluation.

In short, if you don't like what Shiller is telling you, it is because you are a bull who thinks "this time is different".

PS Finally, a couple of links. The first is a piece on the use of statistics and the dangers that face those who try to blog honestly on such a partisan issue as tax. Some of the stuff written on the DSK saga is quite amazing as my colleague Schumpeter points out in typically rollicking style. And there is more justified outrage here and here. As to the argument of the "intellectual" that that New York chambermaids arrive in brigades, not singly, words fail me but not this columnist.