In any endeavor, history can serve as a useful guide to what might happen in the future. The good news for investors is that studies of historic investment returns are far more detailed and accessible than they used to be. Triumph of the Optimists by Dimson, Marsh and Staunton is one of the most useful and should be a core part of any serious investment curriculum, but there are others.
The bad news for investors is that even when good information can be attained, its translation into investment advice and portfolio strategy can vary substantially. Much like background noise and poor connection quality can make it hard to understand a person on the other end of a phone call, so too can "noise" interfere with the quality of the signal investors receive in the form of advice. This phenomenon is readily apparent in regards to establishing appropriate guidelines for expected investment returns.
For starters, the quality of underlying data regarding returns is fairly good - which is often not the case with investment research. It encompasses long periods of time and multiple geographic markets. The Dimson Marsh and Staunton (DMS) study (see [here] for our book review) encompasses returns between 1900 and 2000 for 16 different countries. Jeremy Siegel also conducted a study of stock returns focusing on just the US but dating back to 1802 which he popularized in his book, Stocks for the Long Run. The studies are similar for the depth of their research and for the fact that both found US stocks providing a real return of 6.7% over their study periods.
The path of these research efforts diverges when it comes to interpreting the results for the purpose of establishing expectations, however. DMS focuses on analyzing the patterns they see in the historical returns and normalizing them as the basis for making a sensible forecast. One of the key points they highlight is that valuations have changed considerably over their study period and this provided a one-time, unsustainable boost to returns.
They report, "Since 1900, there has also been a dramatic change in the valuation basis for equity markets. The price/dividend ratio (the reciprocal of the dividend yield) is much higher now than it was in 1900. After adjusting for the difference, they conclude that the ex ante risk premium for US stocks is 1.7% lower than the historical premium. "Our assertion in this book ... is that the equity premium is markedly lower than many people suggest." Indeed, this outlook is very consistent with Dimson's recent assessment in the Economist [here] that "the likely future long-term real return on a balanced portfolio of equities and bonds will be 2-2.5%."
A second finding from DMS is that the unusually strong returns in the second half of the twentieth century appear to be statistical flukes and unlikely to be repeated. They note, "This was a period [the latter half of the twentieth century] when most things turned out better than expected. There was no third world war, the Cuban Missile Crisis was defused, the Berlin Wall fell, and the Cold War ended. There was unprecedented growth in productivity and efficiency, improvements in management and corporate guidance, and extensive technological change. Corporate cash flows grew faster than expected, and in all likelihood the equity risk premium fell, further boosting stock prices. In short, it was the triumph of the optimists."
In other words, the phrase for their book title, Triumph of the Optimists, is intended to be a mild warning in regards to expectations. They conclude their study by highlighting, "Statistical logic tells us that future expectations must lie below today's optimists' dreams. We can hope for, but we cannot expect, the optimists to triumph in the future. Future returns from equities are likely to be lower than those achieved in recent decades ... experience should teach us realism, not optimism."
Siegel, by contrast, take a very different approach when establishing expectations for future returns by highlighting the constancy of stock return through history. As he often does, he started and ended his November presentation at the CFA Institute's Equity Research and Valuation Conference [here] with a graph showing the returns to stocks, bonds, bills, gold, and the dollar. The chart shows stocks on a nearly linear upward trajectory with the returns for all of the other assets on considerably less attractive paths. Although he stops short of proclaiming 6.7% as his expected return for stocks, he clearly relishes in the moniker "Siegel's constant" being applied to his findings. By leaving the graph of historical stock returns on the screen at the end of the presentation, he leaves a strong visual impression, and implied message, that past is prologue.
Siegel also takes a very different approach to the subject of valuation. For one, he prefers using price/earnings (PE) as an indicator, despite the fact that just like with returns, one year's worth of earnings can be hugely unrepresentative. To his credit, he does discuss Shiller's cyclically adjusted price to earnings ratio (CAPE) which actually does a very good job of indicating future returns. However, after noting that the conventional CAPE methodology forecasts only 2% real returns for stocks, he moves on to describing how he believes the CAPE metric should be adjusted. His conclusion is that with certain adjustments, current valuation metrics point to expected returns to stocks very much in line with the long term average of 6.7%
So we have two very different takes on essentially the same data set of stock returns. Siegel is bullish in finding stocks right on track to continue their long run record of 6.7% real returns - which is well above the returns of other asset classes. DMS, while also recognizing the historical superiority of stocks, are considerably more cautious in their expectations for future returns. Both perspectives are well informed views by respected academics. Unfortunately, this conflict creates even more of a challenge for conscientious investors trying to establish an appropriate portfolio strategy. How should investors cut through the noise?
In an important sense, we enjoy having multiple sides to debates like this because it forces us to understand the positions very clearly and to disentangle what can be very subtle issues and assumptions. The case of return expectations is an excellent example because both views seem quite plausible. We begin our investigation, as we often do, by searching for inconsistencies and differences in underlying assumptions.
One key assumption Siegel makes is that although stocks can deviate materially over the short term, those deviations become progressively smaller over longer periods. This is an important tenet in his thesis "stocks for the long run" but one that is not uncontroversial.
Zvi Bodie, another noted academic, argues that Siegel's view understates the long run risk of stocks. He describes in his paper "The long run risk of stock market investing: Is equity investing hazardous to your client's wealth?" in the Financial Analysts Journal [here] that, "Economic uncertainty, especially, is magnified with time. What is the worst thing that can happen over the next 5 years compared with over the next 10,15,20,30, or 100 years? In 100 years' time, a myriad of catastrophic things could happen." This is an issue we highlighted in the blog post "Spring Cleaning" [here] where we noted that this observation is common in fields outside of economics and is a key factor in engineering (long term) infrastructure projects.
Two other academics, Lubos Pastor and Robert Stambaugh, also addressed this issue in a paper entitled "Are stock really less volatile in the long run?" [here]. They acknowledge that "Conventional wisdom views stock returns as less volatile over longer investment horizons." However, they also report that "stocks are actually more volatile over long horizons from an investor’s perspective."
They go on to explain: "Investors condition on available information but realize their knowledge is limited in two key respects. First, even after observing 206 years of data (1802–2007), investors do not know the values of the parameters of the return-generating process, especially the parameters related to the conditional expected return. Second, investors recognize that observable “predictors” used to forecast returns deliver only an imperfect proxy for the conditional expected return, whether or not the parameter values are known. When viewed from this perspective, the return variance per year at a 50-year horizon is at least 1.3 times higher than the variance at a 1-year horizon."
In other words, the future is uncertain and hard to predict. Indeed, an important element of their findings is that they explicitly call out the difference between assessing variance after the fact, or ex post, and assessing variance in the future, or ex ante.
In contrast to Siegel, the notion that the future is inherently less certain permeates the language of DMS. This is evidenced when they say, "downside risk is always present," and "because of the power of compound interest rates, the very worst that could happen to an equity investor worsens as the investment horizon is lengthened." When DMS "examine the range of risk premia that can be anticipated over various future time horizons," they find that "There is clearly a substantial probability of achieving a negative risk premium, even over long investment horizons."
Another subject that Siegel treats very differently than DMS is valuation. It is interesting to note that while Siegel sees fit to examine 200 years of stock returns, he uses only the current year's price/earnings as his primary valuation metric. Using a single year's worth of earnings makes his analysis vulnerable to being incredibly unrepresentative of the longer term and in doing so, seemingly antithetical to his effort to capture the big picture revealed by an extensive history.
He does also consider a more robust valuation metric, the Shiller CAPE, which has one of the best records among valuation metrics for correlating with future returns (higher CAPE suggests lower future returns). However, when he finds that the current CAPE suggests future returns to stocks on the order of 2%, he deems it appropriate to adjust the earnings input to CAPE. In doing so he arrives at a CAPE ratio that suggests "very slight overvaluation" and an expected return to stocks very much in line with the historical average of 6.7%.
There are at least a couple of things interesting about Siegel's approach to valuation. For one, he does not appear to make an effort to calibrate for the fact that market valuations today are higher than they were at the beginning of the study periods. DMS explicitly address this as an issue that likely overstated historical returns relative to what can be expected in the future. Siegel makes no such valuation adjustment which means that in order to enjoy the same equity returns in the future as the past, valuations will have to continue to rise at the same rate, all else being equal.
Another interesting aspect of Siegel's approach to valuation regards the adjustment he makes to earnings for CAPE. Rather than comparing the price of the S&P 500 to the earnings of S&P 500 companies, he compares it to the profits from the entire economy. Effectively, he compares apples to oranges. John Hussman provided an excellent analysis of the "adjustment" [here] and James Montier at GMO has also chimed in with well- reasoned, and critical analysis of Siegel's position [here] and [here].
In summary, we find flaws in several key aspects of Siegel's thesis that serious challenge the credibility of his return expectations. For one, reference to any set of asset returns as a "constant" is absurd and defies underlying economic reality. In addition, the failure to clearly highlight the difference between realized historical variance and the variance of uncertain future events unnecessarily biases and complicates the assessment of return expectations. Further, Siegel's valuation work suffers from clear inconsistencies in what Montier calls "a strange way of honestly adjusting a valuation measure."
It is also striking that Siegel does not call out the unusually strong returns in recent years and the negative impact those results may have on future returns. Specifically, the S&P 500 has returned 14.40% per year over the five years through November 2015. This is even greater than the 13.6% annual return achieved between 1982 and 1999 in what Siegel himself calls "the greatest bull market in history", a period which he acknowledges as having generated returns more than double the longer term average.
As a result, one key takeaway from this analysis is that we place more weight on the DMS work in regards to return expectations than that of Siegel. While we believe that, in general, stocks are worthy long term investments, we also believe they entail real risk, especially over horizons of less than ten or twenty years. Currently, based on the conventional CAPE ratio, we believe stock returns for the next ten to twelve years will be in the very low single digits, nearing zero. This is relevant for anyone who depends on achieving much higher returns, is retired or may be retiring shortly, or for whatever reason may need access to their investment funds in less than 30 or 40 years.
We also believe that "Siegel's constant" of 6.7% is an interesting historical occurrence, but that it says very little about the future and creates an "anchor" that can inhibit more productive intellectual inquiry. In order to calibrate that realized return of 6.7% to potential future results, we must consider how things may differ in the future. We know that the US experienced remarkable growth since 1802 and that is unlikely to repeat, at least not to the same degree. We know that productivity has recently crashed and that if it remains at current levels, it will be extremely difficult to achieve historical return levels. Demographic trends point to an aging society which is typically more averse to risk and has less demand for stocks. And debt and entitlement burdens are at record highs. Any one of these issues could depress future returns and if all of them exert pressure, future returns could be materially lower. After all, equity is only what is left after all other liabilities.
Finally, this exercise also reveals one of the great challenges of investing and is symbolic of one of the industry's major shortcomings: the almost constant need to cut through the noise. While we respect Dr. Siegel's work, at the same time we believe that much of it used to fuel a bullish narrative at the expense of a clear discussion of issues relevant for investors. This doesn't happen because he isn't aware of the issues. Unfortunately, it makes things harder for investors when smart, authoritative figures produce overly ebullient outlooks that inflame the already troublesome tendency many have to extrapolate past results into the future. History can inform the future, but past is not prologue. We believe the more useful approach is that of DMS which appropriately tempers that enthusiasm with the lesson that "experience should teach us realism, not optimism".