From The Economist print edition
Averages need to be treated with caution
CONSIDER these two statements: the American stockmarket was overvalued in 2000; company profits are high relative to economic output. Many people would agree with both propositions, but implicit in each is that there is a correct, or normal, value for shares or profits. The corollary is that when share price or profits are too high or too low, they will eventually revert to the mean. But what do people mean by mean?
Efficient-market theory, which states that prices already reflect all available information, presents an immediate problem for the idea of a reversion to the mean. If share prices were obviously too high, investors would sell their holdings until prices fell to the correct level. There would be no extremes to revert from.
However, in the light of the dotcom bubble, it seems inherently unsatisfying to argue that markets are always fairly valued. Such a position also gives a green light to those who argue that “it's different this time”, and who use any old valuation measure (price per eyeball for internet stocks) to justify share prices.
Pinning down the right mean, however, is difficult. American profits look high by the standards of the past 20 to 30 years. But they do not look unreasonable relative to 1950s and 1960s levels. Perhaps the 1970s and 1980s (which saw high inflation and double-digit interest rates) were the aberration and the immediate post-war era was the norm.
Things get even more complicated when investors start to talk about stockmarket valuations. Most analysts tend to use the prospective price-earnings ratio as their chosen measure, rather than the historic figure. They justify this by saying that investors look forward, not back. But banks are in the business of selling shares. Since analysts nearly always forecast rising profits, the prospective p/e ratio is usually lower than the historic one; that makes the market sound cheaper.
But it also creates a problem when making historic comparisons. It is hard to get figures for earnings forecasts before the 1980s. So when analysts say the market looks cheap on a historic basis, they are really saying it looks attractive relative to a period that saw one of the greatest bull markets ever known.
You can avoid this by using a cyclically adjusted p/e ratio, which averages profits over the past ten years. On that basis, Wall Street looks pretty expensive, but the ratio assumes that profits are set to mean revert.
Switching to dividend yields or asset prices as a valuation measure does not eliminate the problem. The average dividend yield on the American market between 1926 and 1950 was 5.5%, according to Barclays Capital. With Wall Street currently yielding less than 2%, that makes shares look very expensive. But bulls argue that conditions in the first half of the 20th century, which suffered two world wars and a depression, and had a stockmarket dominated by individual investors, are irrelevant to today's markets.
To critics such as Andrew Smithers, of the consultants Smithers & Co, such an argument counts as “data mining”, the exclusion of statistics that do not support the bulls' case. He says the stockmarket does revert to the mean, in that periods of high returns tend to be followed by periods of low returns. On his preferred measure, the q ratio (the relationship of share prices to the replacement cost of net assets), stocks are now clearly overvalued.
However, the Smithers thesis presents investors with two obvious problems. The first is that shares have looked overvalued, on his measure, since the mid-1990s. Full reversion has not yet occurred; this has not been a very useful trading signal. The second is that there have been relatively few peaks for q. Have there been enough cycles to make a convincing case?
A subtle variant of the mean reversion case is put forward by Jeremy Grantham of the fund management group GMO. He says that share valuations do revert to trend, but the trend is rising over time (thanks to better economic stability and lower transaction costs). While this may sound plausible, it is a slippery concept to use in practice. If valuations rise, how do you know whether the market is overvalued or that the trend rate has accelerated?
Investors need to distinguish between short- and long-term means. The first may bolster the case of bullish analysts, but depend on “it's different this time” arguments. Longer-term ratios are more intellectually robust, but investors can go broke waiting for them to revert. Sometimes, means can be very mean.
No comments:
Post a Comment