Many shall be restored that now are fallen and many shall fall that now are in honor.
– Epigraph by Horace in Security Analysis (1934) by Graham & Dodd
In our May contribution we discussed the danger related to valuation risk, i.e. the danger of overpaying for the hope of growth (Montier, 2009). We documented that the expectations concerning long-term earnings growth rates of glamour stocks are overly optimistic and how this overoptimism results in inferior returns as well as (relatively) strong price drops following the publication of disappointing results. We concluded that (a) in their quest for successful growth stories investors should be careful before relying too heavily on long-term earnings growth forecasts made by financial analysts and (b) investors should avoid stocks with significantly above average valuation multiples. In this contribution we will add a new, third warning to the list: we warn to take care when using past earnings growth rates as a reliable precursor for future growth. It so happens that firms with high past earnings growth rates cannot be counted on to repeat their strong relative performance in the future; at the same time companies that over the past years have dangled round the growth group are not necessarily doomed to generate low growth. The findings warn against extrapolating past success into the future.
By custom we start with the vision of Benjamin Graham (1949) regarding the added value of past earnings trends in assessing future performance. In a second step we discuss the results of two research studies (Chan et al., 2003; Haugen, 2010) and elaborate on the important implications for growth and value investors respectively. We end with some conclusions.
-Old-School Value Investing
In chapter Group Studies of Earnings and Price Developments Benjamin Graham (1949) formulates an answer to the question “How permanent are trends?”:
Wall Street’s judgment has been influenced by past trends more than by any other single factor related to security values. The avowed object of people in the market is to anticipate future developments, and the past is held to have no significance except as it aids in such anticipation. Yet in practice it is almost the universal habit to base forecasts of future happenings on a projection of past trends. This is notoriously true of both the professional’s and the public’s view of market prospects. Nearly everyone is optimistic (or “bullish”) because the market has been enjoying a spirited advance and pessimistic (or “bearish”) after a decline. In the same way an industry or a company which has grown in the past is almost always expected to keep on progressing; those which have been on the downgrade are expected to get worse and worse.
The last attitude is expressed in categorical fashion on page 458 of Mead & Grodinsky’s book The Ebb and Flow of Investment Value, as follows: “Declining industries, therefore usually continue to decline until they reach the point where they pay nothing to the investor.”
Our own thinking during the past thirty years has been out of sympathy with this viewpoint. It is true that every established trend has a certain momentum, so that it is more likely to continue for at least a while longer than it is to reverse itself at the moment of observation. But this is far from saying that any trend may be relied upon to continue long enough to create a profit for those who get aboard. Rather extensive studies which we have made of the subject lead us to conclude that reversals of trend in every part of the financial picture occur so frequently as to make reliance on a trend a particularly dangerous matter. There must be strong independent reasons for investing money on the expectation of a continuance of past tendencies, and the investor must beware lest his weighing of future probabilities be unduly influenced by the trend line of the past. (Emphasis added)
The message is clear: past trends are not in the least a reliable guide to the future. The underlying reason has already been mentioned in our previous contribution. High growth rates and high profit margins attract rivals which results in increasing competitive pressures and downward pressure on growth and margins. On the other hand corporate bankruptcies and reorganisations will eliminate overcapacity and contribute to higher profit margins and growth rates.
In Haugen (2010) the author refers to a study by Rayner and Little (1966). Rayner and Little look at British companies over the 1951-1961 period. Firms are ranked based on growth in earnings per share over the 1951-1956 period; the same procedure is repeated for the second half of the 1950s. The ranks in the two subperiods are subsequently plotted on a graph. The rank of a company in the first subperiod is plotted on the x-axis; the rank in the second subperiod is shown on the y-axis. If past growth rates serve as a reliable guide to the future we should find a pattern similar to the one in GRAPH I. Firms with the highest relative growth rates over the past years continue to bring this performance in the following years; firms that have underperformed in the past will continue to do so in the future.
GRAPH I: THE PAST AS A RELIABLE GUIDE TO THE FUTURE
If growth rates in the two subperiods are unrelated a pattern similar to the one in GRAPH II will emerge.
GRAPH II: PAST GROWTH NOT A RELIABLE PRECURSOR FOR FUTURE GROWTH
Rayner and Little find a pattern similar to GRAPH II. Using past earnings growth rates as a guide to the future is about as reliable as the fifty-fifty odds in flipping a coin. The study by Rayner and Little dates back from 1966. The evidence has been updated in “The Level and Persistence of Growth Rates” by Chan et al. (2003) for US companies. The study concerns the 1951-1998 period. The researchers verify whether firms that have demonstrated consistently high or low past growth rates have continued this pattern in the future. They proceed in the following way. At the end of each year Chan et al. select the firms that have realised superior past growth over the past three years. They subsequently check the number and percentage of companies that continue to deliver this strong relative performance in the following years. The results are shown in GRAPH III.
For year one in GRAPH III there are 259 firms that time and again realised an above-median growth in earnings over the past three years and that still exist after one year. 125 out of the 259 firms (which is 48.3 percent) succeed in realising an above-median growth in earnings for the fourth year in a row. If the possibilities of outperformance were simply a matter of chance, we would expect to find 130 companies (259*0.5) with above-median earnings growth for four consecutive years. Concerning the second year, 240 companies have realised an above-median growth in earnings over the past three years in a row and survive the upcoming two years. 57 firms out of the 240 – which is 23.7 percent – realise an above-median earnings growth during five consecutive years. By the laws of probability we would expect 60 firms (240*0.5^2) to be successful. As a consequence the number of companies achieving sustained high growth rates in earnings is not much different from what is expected by chance. Concerning year three to five in GRAPH III we also find that past and future performances are unrelated. Companies that realised the strongest past relative growth rates in earnings cannot be relied on to repeat this outperformance in the future.
GRAPH III: PERCENTAGE OF FIRMS WITH ABOVE-MEDIAN GROWTH EACH YEAR FOR THE PAST THREE YEARS AND ABOVE-MEDIAN GROWTH EACH YEAR IN THE FOLLOWING YEARS
Chan et al. also look at “the dogs”, i.e. companies with inferior past earnings growth rates. In GRAPH IV they select companies with below-median growth in each of the three past years and evaluate their relative performance in the five subsequent years. When we look at the third year of the following graph we find that 28 out of the 184 companies with below-median growth in earnings over the past three years – which is 15.3 percent – realise a growth in earnings above (!) the median during each of the three post-formation years. If the possibilities of outperformance were simply a matter of chance, we would expect to find 23 companies (184*0.5^3) with above-median earnings growth in each of the three post-formation years. Otherwise said the dogs from the past are not doomed to bring permanently low growth.
GRAPH IV: PERCENTAGE OF FIRMS WITH BELOW-MEDIAN GROWTH EACH YEAR FOR THE PAST THREE YEARS AND ABOVE-MEDIAN GROWTH EACH YEAR IN THE FOLLOWING YEARS
The findings by Chan et al. (2003) confirm the early evidence documented by Rayner and Little (1966). The reliability of past relative performance when predicting future relative performance is as reliable as the fifty-fifty odds in flipping a coin.
What are the consequences of these findings for growth and value/contrarian investors respectively? For growth investors two warnings can be stated. First high past relative earnings growth rates should not be considered to be representative for the future. Secondly one should try to avoid that the substantial growth from the past results in willingness to paying significantly above average valuation multiples. The following simplified example illustrates that the extrapolation of past earnings growth rates, which results in overoptimistic future earnings forecasts, in combination with the willingness to paying an overly liberal multiplier, results in an extremely overoptimistic assessment of intrinsic value and consequently in a significant risk of substantial long-term losses.
In the study by Chan et al. 10 percent of the companies realise an average annual earnings growth rate higher than 21.3 percent over a five-year period. Based on the evidence presented above we assume that half of these companies will generate above-median earnings growth rates in the next five years, the other half will generate below-median earnings growth rates. Let’s assume that these companies, on average, succeed in realising an average annual growth rate in earnings of 9.8 percent over the next five years (9.8 percent is the median annual growth rate in earnings for all companies over the 1951 to 1998 period). At the moment the 90th percentile of the price-to-earnings ratio for US stocks is 50. We assume that this valuation multiple is representative for the 10 percent companies with the highest average annual growth rates in earnings over the past five years. Chan et al. provide evidence that investors actually are willing to pay higher valuations for stocks having achieved several consecutive years of strong growth, believing these companies will continue their stellar performance.
We now decide to invest $100 in the group of stocks with the 10 percent highest earnings growth rates over the past five years. At a price-to-earnings ratio of 50, we obtain average earnings per share of $2. If growth investors naïvely assume that these companies will succeed in repeating the past growth rate of 21.3 percent over the next five years, they will project earnings after five years of $5.25. Applying the same liberal multiplier of 50 the growth investor obtains an estimated intrinsic value of his stock portfolio of $262. Given the evidence presented above there’s not much of a chance that the firms once again will succeed in realising an average annual growth rate of 21.3 percent; when considering a more conservative and realistic average annual earnings growth rate of 9.8 percent, the estimated earnings within five years amount to $3.19. At a multiplier of 50 this would produce an estimated intrinsic value of $160. The growth investor is however misled to believe that investors would be prepared to pay a multiplier of 50 for a group of average companies. The median price-to-earnings ratio for US companies at the time of writing is 16.42. Applying this more conservative multiplier to the estimated earnings of $3.19 after five years we obtain an intrinsic value of only $52.3, representing a loss of approximately 48% compared to the initially invested capital!
This simplified example demonstrates the dangers related to extrapolating past growth rates to the future and at the same time rewarding high growth companies with high valuation multiples, a dangerous confluence of two factors that Benjamin Graham (1949) has warned us against:
What seems to happen, rather, is that the price remains high until the earnings actually show a definite falling off – which invariably seems to take the followers of the issue by surprise. Then we have the market decline usually associated with a disappointing development – a decline perhaps intensified by the fact that the price level of the growth stock had been dangerously high.
Overoptimism in future earnings growth rates in combination with a significantly above average multiplier turns out to be the perfect receipt for guaranteeing a permanent loss of capital over the long term.
It is understood that the aforementioned example can be reversed and implemented from the perspective of a value or contrarian investor. Let’s put the focus on the companies with the 10 percent lowest average growth rates over the past five years. In the study by Chan et al. companies at the 10th percentile realised an annualised average earnings growth rate over the past five years of -6.4 percent. If the market handles these stocks as if their future would be a reflection of their past lacklustre growth, their stock prices could be reprimanded too heavily, offering lucrative opportunities for contrarian investors. In The Intelligent Investor (1949) Benjamin Graham remarks:
Our final category of opportunities for enterprising investment lies in the field of undervalued or bargain issues. These are the direct antitheses of growth stocks. If the latter may often sell to high because they are too popular, many not non-growth stocks often sell too low because they are too unpopular? We believe the answer to this question is definitely yes. A sound analogy may be drawn between a depressed general market and a stock that is individually unpopular. Just as declines of the whole market tend to go too far because public sentiment is generally pessimistic, so the price of many single issues may fall unduly low because their future is considered to be relatively unpromising. In both cases unfavorable circumstances are present – as actual developments, or perhaps only as prospects – and in both cases the response may be excessive and thus create a genuine opportunity for the intelligent and courageous investor. In the depths of a depressed or bear market the average person can see no ray of light ahead and can think only in terms of worse to come. So too, when an individual company or industry begins to lose ground in the economy, Wall Street is quick to assume that its future is entirely hopeless and it should be avoided at any price. The two types of reasoning are similar, and equally fallacious. (Emphasis added)
At the moment the 10th percentile of the price-to-earnings ratio for US stocks is 7.7. Applying the same methodology as before, an investment of $100, at an average annual growth rate in earnings of only 7.6 percent (< 9.8 percent) and a multiplier of 11.52 (25th percentile at the time of writing), would turn $100 into $216 over a five-year period.
Based on the aforementioned statements readers might have noticed the substantial dispersion of the price-to-earnings ratio for US stocks. The 10th percentile is 7.7; the 90th percentile is 50. The numbers closely correspond with those of Chan et al. from 1999 (7.4 for the 10th percentile, 53.9 for the 90th percentile). This large gap implies that the market expects companies in the top ten percent will realise a significantly higher growth over the upcoming years. Based on the evidence presented chances are small that these stocks will come up to the expectations. As long as this “relative pricing structure” stays in place, contrarian and patient value investors will be rewarded nicely for picking up the dogs of the past years at low valuations (guaranteeing of course the presence of sufficient financial strength, we refer to our April contribution).
Once again the findings from Haugen (2010) and Chan et al. (2003) confirm some important insights from Benjamin Graham (1949) concerning the chances of successful implementation of, and the risks attached to a growth stock program, and they are comprehensively summarised in the following notice:
All in all, our evidence on the limited predictability of earnings growth suggests that investors should be wary of stocks that trade at very high multiples. Very few firms are able to live up to the high hopes for consistent growth that are built into such valuations. – Chan et al., 2003
At the same time the findings offer support for the courageous value investor. All too often investors once and for all write off the dogs of yesterday. Stocks that have been penalised for their low relative performance over the past years and that dispose of a strong financial position (guaranteeing that the companies survive a shake-out in the industry or a major reorganisation) can be picked up at low valuations providing patient investors with both a significant margin of safety and attractive long-term investment returns.
In the world of securities, courage becomes the supreme virtue after adequate knowledge and a tested judgment are at hand. – Graham, 1949
Chan et al. provide evidence that investors actually are willing to pay higher valuations for stocks having achieved several consecutive years of strong growth, believing these companies will continue their stellar performance. Companies are sorted annually based on the average annual growth in earnings over a period of ten years and divided in deciles. Chan et al. subsequently calculate the median book-to-market (BM) at the start and end of this ten-year period for the ten decile portfolios. The results are shown in Table I. The median annual growth rates for the portfolios are shown in the first row. Looking at book-to-market, stocks with weak financial performance over the past ten years (decile one) were punished with significantly lower valuations; the book-to-market ratio increases from 0.653 to 1.048. The opposite evolution can be found with companies having the strongest growth rates over the past ten years (decile ten); these stocks are reasonably rewarded with the median book-to-market decrease from 0.817 to 0.622.
In July we will provide the reader with an illustration of the value added of descriptive historical evidence. Taking a look at many finance programs let students and professionals believe that any knowledge of financial history is completely irrelevant. Nevertheless time and again the finance industry finds itself astonished when reoccurring financial crises are brought on by EXACTLY the same causes. ALL of them boil down to two causes: COMPLEXITY and LEVERAGE. In August we will discuss the concepts of Super Stocks versus Stupid Stocks as introduced by former professor Robert Haugen.
Chan, L.K.C., J. Karceski, and J. Lakonishok. (2003). “The Level and Persistence of Growth Rates.” The Journal of Finance, Vol. LVIII, No. 2, 643-684.
Graham, B. (1949). The Intelligent Investor. HarperCollins Publishers.
Haugen, R.A. (2010). The New Finance – Overreaction, Complexity, and Their Consequences (Fourth Edition). Pearson Education.
Montier, J. (2009). Value Investing. Wiley.
Rayner, A.C. and I.M.D. Little. (1966). Higgledy Riggledy Growth Again. Basil Blackwell.