Semi-strong Efficiency and Some More Anomalies

Semi-strong Efficiency and Some More Anomalies

Academic and financial analysis in the semi-strong school of market efficiency believe that all public information about a company is always reflected in the stock's price. Thus, they are skeptical about the ability of “fundamental” security analysts to pore over data concerning a company's earnings and dividends in an effort to find “undervalued” stocks, which represent particularly good “value” for investors. Two of the anomalies given considerable attention in the 1980s are quantification of two value techniques of security analysts: Look for security that sell (1) at low multiples compared with their earnings and (2) with high dividends compared with their market prices. Has this new scientific evidence proved that the Wall Street security analysts were right all the time?

There is some evidence that stocks with low price-earnings multiples outperform those with high multiples: We have come to another potential anomaly with which I have considerable intellectual sympathy. One of my cardinal rules of stock selection is to look for companies with reasonable growth prospects that have yet to be discovered by the stock market and thus are selling at relatively low earnings multiples. I have also warned investors repeatedly about the dangers of very high-multiple stocks that may be the current favorites of the investment community. Particularly since earning growth is so hard to forecast, it's far better to be in low-multiple stocks; if growths does materialize, both the earnings and the earnings multiple will likely increase, giving the investor a double benefit. Buying a high-multiple stock whose earnings grows fails to materialize subjects investors to a double whammy. Both the earnings and the multiple can fall.

Keep your eyes on the stars, and your feet on the ground (Theodore Roosevelt). Photo by Elena

There is some evidence that a portfolio of stocks with relatively low earnings multiples has often produced above-average rates of return even after adjusting for risk. To some extent, this phenomenon is related to the “small firm effect”. But again, as with the small firm effect, the “P/E effect” appears to vary over time – it is certainly not dependable over every specific investment period. Moreover, even if it can be shown to persist on average over a long period of time, one can never be sure if the excess returns are due to increased risk or to market abnormalities. The studies which have documented abnormal returns have used relative volatility or beta to measure risk. To the extent that one has reason to believe that beta is not a perfect, or even in many instances a useful, risk measure – one should treat the low P/E anomaly with some suspicion.

And don't forget that low P/Es are often justified. Very often companies on the edge of some financial disaster will sell at very low multiples of reported earnings. For example, just prior to declaring bankruptcy in 1983, Continental Illinois Bank sold at an unusually low earnings multiple. The financial community in this case was entirely justified in disbelieving reported earnings. The low multiples reflected not value but a profound concern about the viability of the bank. It turned out that Continental's reported earnings bore little relation to the actual economic earnings of the bank.

Another illustration will show how difficult is to implement a low P/E strategy. Suppose two identical banks have $10 per share in earnings for the year, half of which represents “pay-in-kind” interest from financially weak less-developed countries (LDCs). The LCDs can't pay their interest but instead just write a new IOU for the unpaid interest. Bank One reports the whole $10 in earnings while Bank Two reports only $5 as earnings, preferring to set up the more questionable extra $5 "pay-in-kind” interest as a reserve against future potential defaults. Which bank will show the higher P/E multiple? Most likely it will be the more conservative Bank Two, which reported the lower earnings. If both banks sold at $50 per share (and by assumption they are identical except for their accounting policies), then the conservative Bank Two would have a P/E multiple of 10 while Bank One, which set up no reserves and just called everything “earnings,” would have a multiple of only 5. It's easy to see how a low P/E criterion could in some instances give a poor measure of true value.

Higher initial dividends have meant higher subsequent returns: Another apparently predictable relationship concerns the returns realized over several quarters of years from stocks and the initial dividend yields at which they were purchased. For example, 25 percent of the variability of a two- to four-year holding period on returns can, in certain periods, be predicted on the basis of the stocks' initial dividend-price ratios. Such a finding is perfectly consistent, however, with an efficient-market view of security price determination. Those prices are low relative to dividends (that is yields too high) when general market interest rates and thus required returns are high. Such a result is also consistent with the findings of mean reversion (return reversals) described above. An economic shock that raises general market interest rates will be associated with a decline in stock prices, which will lower realized returns. But the price decline raises both the dividend yield and the future rate of return. Assuming that the cumulative price effects from fluctuations in market interest rates are roughly zero, the time variation of expected returns can give rise to mean-reverting components of market prices.

The point, then, is that at a time when bod interest rates are high, dividend yields on stocks are also likely to be high and those higher dividend yields will presage higher subsequent stock returns. This is a logical outcome and is entirely consistent with efficient markets. And there is no doubt that, other things being the same, stocks with higher dividend yields represent better value. Unfortunately, a strategy often does do well, and a portfolio yielding high dividends may be especially appropriate for certain individuals in low tax brackets or with high income needs for living expenses. But there is no evidence that the market systematically and consistently fails to adjust properly to current and prospective dividend returns.

In sum while it is true that departures exist from the weak and semi-strong forms of the efficient-market hypothesis, departures from randomness are generally small and are not consistent over time. An investor who pays transactions costs cannot generally formulate an investment strategy that is profitable on the basis of these anomalies. Moreover, the more dependable relationships, such as those associated with general movements in interest rates, may be perfectly consistent with market efficiency. Although the random-walk hypothesis is not strictly upheld, the documented departures from randomness do not appear to leave significant unexploited investment opportunities that are inconsistent with the efficient-market hypothesis.

Burton G. Malkiel. A Random Walk Down Wall Street, including a life-cycle guide to personal investing. First edition, 1973, by W.W. Norton and company, Inc.