Dividends Do Matter We have written many times before on the importance of dividends to investors. And we continue to stand by this view. A dividend represents the actual cash returned to an investor in relation to their ownership of the company. Our previous work has been an analysis of the US and UK long run data. We have recently come across data for a far wider sample to explore. For those who are unfamiliar with our approach, here is a brief summary. The total real return an investor receives can be decomposed into three components:
Of course, during the long bull market, dividends became a dirty word. When earnings growth was seemingly invincible, only the truly dull would concern themselves with the miniscule return embedded in the dividend yield. It is easy to see how investors ended up in this situation. Looking at the period 1995-2000, over 80% of the return achieved was generated through multiple expansion - an unprecedented contribution from this element! Such an occurrence has never before happened. [By multiple expansion Montier means the increased value that an investor puts on the value of earnings. That means a rising P/E ratio and not the actual growth of earnings was responsible for 80% of the increase in the value of the stock market from 1995-2000.) However, we know that people tend to extrapolate the more recent past into the future. So it is easy to see why investors tend to think that price appreciation is the best way of accessing returns. However, a longer time horizon exposes the fallacy of this view. Taking a broader view of history shows that dividends are far more important than the experience of the late 1990s would suggest. For instance, decomposing the returns from the US market over the long run since 1900 reveals that some 66% of the total return was generated by the dividend yield. Investors sometimes question the quality of the early data, so examining the post War data from 1950 helps bypass this concern. Ironically, the case for the importance of dividends is strengthened rather than weakened by the focus on this sub-sample. Between 1950 and 2000 some 72% of the total return an investor achieved was delivered via the dividend yield! Just in case you are wondering, stock repurchases don't alter this picture to any meaningful degree. Repurchases added 73 basis points (0.73%) to the dividend yield on average between 1995-2000. Besides which, the above analysis deals in per share data, hence any repurchases should have increased the future rate of growth, so the whole repurchase issue is something of a red herring in the current context. Whenever we have presented this work in the past we get asked if we have the data for any other countries. Until very recently we have been unable to perform similar analysis on any other country except the UK. However, we recently came across data that allows us to explore this issue in a far wider context. The data are drawn from the masterful study of returns, "Triumph of the Optimists" by Dimson, Marsh and Staunton. The results from our analysis of the data show that the US isn't an isolated case. Across nine major global markets we find that an average 65% of total return has been generated by the dividend yield since 1950. A quick glance at the data reveals a remarkable consistency to the importance of dividends. For instance, in Europe (excluding in this case the United Kingdom) the average contribution of dividend yield to total real returns has been 62%. So, nearly two thirds of investors' total real returns in the European (ex UK) arena are provided by dividends. Of the countries considered only Germany stands out as having a remarkably low contribution to total returns stemming from the dividend yield (just 30%). Effectively Germany has been one of those rarest of creatures - a growth story that actually worked! However, the tangible growth of Germany is played out still via the actual growth in dividend yield of 65% (thus showing that the only true growth is still perhaps tied to dividends). All of this is interesting, but can it tell us anything about the future? We believe the answer is yes. We view the framework set out in the beginning of this piece (that total return from stocks can be characterized as the dividend yield, any growth in real dividends, and any change in valuations over the holding period) as an example of the "decomposition" approach to forecasting. In the book Principles of Forecasting (2001), Donald MacGregor defines decomposition as: A method for dealing with problems by breaking down the estimation task into a set of components that can be more readily estimated and then combining the component estimates to produce a target estimate. There is a large literature from psychologists on the benefits of decomposition forecasting. But the basic idea is very simple. Using a building blocks approach to (in our case) estimating returns is likely to be less subject to behavioral biases than simply trying to come up with a return forecast in totality. The equation we would use is: the real return = Dy [dividend yield] + real g [growth in dividend yield] + delta P/D [change in the valuations or price of the shares in terms of dividends]. What parts of the equation will change? Given that most markets are either fair value or [as in the case of the US] expensive then it would seem unrealistic to forecast further multiple expansion in the long term. Hence we are left with the dividend yield plus the real growth rate of dividends. The dividend yield part of the equation is obviously easily available. However, the growth rate causes investors much angst. Believers in market efficiency argue that low dividend yields must be offset by higher future growth rates (the Modigliani and Miller theorem (M&M)). We have shown elsewhere [and convincingly - John] that this is not a good description of the US. In fact, if anything there is a 'perverse' relationship between dividend yields and long term dividend growth in the US: higher yields tend to go with higher dividend growth! However, this appears to be the result of a few outliers [data which is either very high or very low from the median and which distorts the averages]. However, even removing these, you would be hard pushed to find any relationship at all. Hence, still a poke in the eye to M&M. [Montier goes into a long discussion of the above principles in the European markets and demonstrates roughly the same conclusion- John.] Given reasonable skepticism over M&M, some choose to plug GDP growth rates in as a substitute for long-term dividend growth. However, this ignores the dilution effect, and the non-listed firm effect. As Arnott and Bernstein (2003) show, dividends have consistently grown below the rate of growth of the economy in the US. And we found similar results in an international extension of their work. [Readers may remember I did an extensive review of Arnott and Bernstein's work a few months ago.] The Ice Age Given these difficulties we have chosen to use the historic real rate of growth in dividends as our proxy for g [growth in dividends]. Plugging all the numbers (no growth in valuation, current dividend yield and historic growth in dividends gives us an expected real return of 2.5% for the US markets. Real returns for the United Kingdom is forecast for 6%, Europe (ex UK) is 4.3% and Japan is 1.3%. This is in line with our long held argument that in a world of low returns (i.e., what Montier calls The Ice Age, a slow period of real returns) dividends will be an increasingly important source of return generation for investors. Some investors see this low return forecast as part of the normal consensus, and hence seem to give it scant regard. Yet these self same investors are happy to play the greater fool/sucker rally that we are currently witnessing. We suspect that they are performing the investment equivalent of St. Augustine of Hippo's plea: "Lord, make me chaste, but not yet" - a near perfect example of cognitive dissonance. Perhaps investors would be better advised to heed Confucius' sagely words "If a man gives no thought about what is distant, he will find sorrow near at hand." |