Is the market overvalued?

In times of wild markets, market commentary and investor behaviour tends to become more short term. As volatility rises, we hear of the "flight to safety" as investors sell stocks and buy government bonds. In times like these I definitely find that it steadies the nerves to take a look at the bigger, longer term picture.

Is the stock market fundamentally overvalued? The answer to this question is probably yes, but not to such an extent that one should not be in the stock market at all. The most common measure of "value", at least that which is used by brokers and financial journalists, is the infamous p/e ratio. The idea behind the p/e ratio is that it gives you some idea of how much you are being asked to pay for the earnings that the companies in the stock market are generating. Earnings reflect value because they (in theory) belong to the stock holders who should see the benefits either in the form of dividends or from growth (if the earnings are reinvested). Currently the p/e ratio of the S&P500 is roughly in line with its long term average, a point which is often cited by stock market bulls.

Unfortunately, statistical analysis shows quite clearly that the p/e ratio in this form does not tell you anything about value. My approach to assessing value metrics is to ask two questions: 1) Is there evidence that the metric is "mean reverting" over time? 2) Does the metric provide a useful indicator of future returns? In the case of the static p/e ratio, analysis shows quite clearly that the answer to question 1) is yes, but for question 2) it is a no.

Taking question 1) first, the point here is that in order for the ratio to have any use, you need to be sure that there is some force within the economy that brings it back to its average. Testing the data statistically yields a highly significant result for mean reversion - i.e. when the ratio is high, there is a high probability that it will come down in the future. Intuitively, this result can be explained in a number of ways. The most simple is to flip the p/e ratio upside down and think of it in terms of the e/p ratio instead, also known as the earnings yield, which is like the dividend yield except it includes all the earnings not paid out to stock holders. Now if the e/p ratio were, say, 2%, which would equate to a p/e ratio of 50, then you would have to be pretty (over)confident about the future earnings growth prospects of the stock market to buy it. Otherwise you would not be getting very much "e" for your "p", while at the same time also taking a substantial risk of capital loss. Usually, wild misvaluations such as these tend to correct themselves sooner or later, restoring some semblance of a risk premium for the ownership of equities.

The main problem with the p/e ratio is that has two parts to it - the "p" and the "e" - and the adjustment, the reversion to the mean, can happen by either of these values adjusting. For the ratio to be a good indicator of value, and hence future long-term returns, the adjustment needs to occur via the price. Unfortunately, history has frequently produced situations where the p/e ratio has been high because the "e" has been low, not because the "p" has been high. The two classic examples are 1933 and 2002-3, both of which saw high p/e ratios, but were nevertheless two very good years to invest in stocks (the former the best year of the whole 20th century in fact). The reason is that they were both followed by a cyclical rise in earnings, pushing down the ratio even while the "p" rose. It is this fact - the cyclicality of earnings - that renders the static p/e ratio useless as an indicator of value.

Thankfully, it is possible to get around the problem by taking a "cyclically-adjusted p/e ratio". In this calculation, the price is divided by an average of the past 10 years of earnings. The point is to eliminate the cyclical problem cited above, the assumption being that 10 years is enough to represent an economic cycle in corporate earnings. Unfortunately, the cyclically-adjusted p/e ratio currently tells a very different story about stock market valuation to the static version favoured by stock market brokers. Because corporate profit margins have been expanding since 2003 and are currently at multi-decade highs, the cyclically-adjusted p/e ratio is much higher than the static p/e ratio. Indeed, the ratio has been around 27 for the past couple of years, which is considerably above its long term average of 17, and clearly implies that the market is overvalued.

There is no question that, for all the reasons cited above, the cyclically-adjusted p/e ratio is statistically and intuitively a far superior measure of value than the static p/e ratio. However the fact that it is higher than its long-term average does not mean that owning stocks is a bad idea. Rather, it simply means that the probability of realising the long-term real return of 6.75% on equities is quite low, because at some point in the next 20-30 years the ratio is likely to mean revert. But even if the actual real return is less than 6.75%, it is more than likely to be higher than the alternatives of cash or bonds.

Interestingly, the key objection to this argument, and the one that I consider most valid, concerns the validity of using the past as a guide for the future. While the past has shown a cyclicality in earnings, as profit margins expand in booms, and contract in recessions, it could be argued that the current high profit margins are a result of structural rather than cyclical factors. For example globalisation and the emergence of a huge low cost labour force have unquestionably shifted the distributional balance in favour of capital, at the expense of labour, a development which could lead to permanently higher profit margins. Whether this is truly a case of "it is different this time" remains to be seen...