I've written a few times that historical returns are useful, but most here misunderstand the utility. The average historical return is not a useful predictor of future returns unless current and future conditions are similar to historical conditions.

Realized returns depend on starting yield, growth in fundamentals (e.g., dividends, cash flow, earnings), and valuation. Thus, historical yields, growth rates, and valuations--the components of realized returns--are the useful historical data points (not the realized return).

The starting yield is a given. If it is lower than the average historical yield, then we can expect future returns to be lower than historical returns, holding all else equal. Future returns have to be lower unless future growth is faster and/or future valuations rise higher.

The current valuation is also a given. There is no law of nature that forces it back to the long-term average, but if the current valuation is well above the historical valuation, then a prudent investor ought to expect it to fall. In any event, future returns will be lower than historical returns unless the valuation rises even farther above its long-term average (from its current level).

Research Affiliates wrote about this here: https://www.researchaffiliates.com/en_us/publications/articles/634-ignoring-starting-yields-nabbing-this-usual-suspect-in-poor-investment-outcomes.html?evar36=eml_Advisor_Starting_0917_Sec tion_1_Advisor1_title&_cldee=ZXJ3aW4ucm9zZW5AZ21ha WwuY29t&recipientid=contact-3e23c5770e56e411b063005056bc3cff-6ec388c2e3b74009af67228218cb154d&esid=c6b047ff-e8a2-e711-80d2-005056bc1247.

(Hat tip to Always Passive, a bogleheads.org poster. I just found the link through his/her post.)

Excerpts:

***

***

Realized returns depend on starting yield, growth in fundamentals (e.g., dividends, cash flow, earnings), and valuation. Thus, historical yields, growth rates, and valuations--the components of realized returns--are the useful historical data points (not the realized return).

The starting yield is a given. If it is lower than the average historical yield, then we can expect future returns to be lower than historical returns, holding all else equal. Future returns have to be lower unless future growth is faster and/or future valuations rise higher.

The current valuation is also a given. There is no law of nature that forces it back to the long-term average, but if the current valuation is well above the historical valuation, then a prudent investor ought to expect it to fall. In any event, future returns will be lower than historical returns unless the valuation rises even farther above its long-term average (from its current level).

Research Affiliates wrote about this here: https://www.researchaffiliates.com/en_us/publications/articles/634-ignoring-starting-yields-nabbing-this-usual-suspect-in-poor-investment-outcomes.html?evar36=eml_Advisor_Starting_0917_Sec tion_1_Advisor1_title&_cldee=ZXJ3aW4ucm9zZW5AZ21ha WwuY29t&recipientid=contact-3e23c5770e56e411b063005056bc3cff-6ec388c2e3b74009af67228218cb154d&esid=c6b047ff-e8a2-e711-80d2-005056bc1247.

(Hat tip to Always Passive, a bogleheads.org poster. I just found the link through his/her post.)

Excerpts:

*Not only do we have to deal with a range of possible life expectancies, but a changing average life expectancy over time. A 65-year-old male in 1980 would have been expected to live until 79.1 years, a five-year shorter period of retirement bliss than he would be looking forward to today. The combined positive effects of advanced diagnosis, less smoking, more nutritious diets, better safety in occupational workplaces, vaccines and antibiotics, and seatbelt laws, among other factors, equate to a longer life expectancy. Fairly indisputable. Obviously, it doesn’t make sense to use a mortality table from 1980 to determine a client’s most likely retirement horizon in 2017 (37 years later).**Starting conditions matter in life expectancy, and as we’ll see shortly, also in expected returns.****

*Despite this clear relationship, the retirement calculators at respected investment management firms—those with hundreds of billions of US dollars of client assets—appear anchored on history. For example, Vanguard*^{2}uses a 4.0% real (after inflation) annualized long-term expected return. Dimensional Fund Advisors assumes real annualized long-term expected returns of 1.0% and 5.0% for global bonds and global stocks, respectively, implying a real return of roughly 3.9% for a 60/40 global portfolio.^{3}At first blush, these estimates seem entirely reasonable, and maybe even prudent. Indeed, looking back, a conventional 60/40 US stock/bond blend would have achieved real annualized returns exceeding 4.0% in each of the past 5-, 10-, 20-, 30-, and 40-year periods.*So, why is it unrealistic to expect the same in the future? After all, the past 40 years have witnessed a variety of economic and political regimes and should be a reasonable proxy for the future. To answer this, it is necessary to understand the sources of past returns, going as far back as possible. Let’s focus on equities, the lion’s share of a 60/40 allocation’s return. As Arnott and Bernstein (2002) explain, the majority of the real return on stocks over the past two centuries came from three sources: 1) dividends paid, 2) real growth in dividends paid, and 3) rising valuation levels. Valuation matters. Over the last 40 years ending June 30, 2017, rising valuation levels’ contribution to the annualized real return of the US stock market approaches one-third (2.1% of 7.4%).*^{4}*Today, yields are dramatically lower, and equity valuations are dramatically higher. The average bond yield over the last 40 years was 6.7% (Barclays Capital US Aggregate), and the average cyclically adjusted price-to-earnings (CAPE) ratio was 21.1 times. As of August 31, 2017, these measures are 2.4% and 30.3 times, respectively.*^{5}In other words, today we are experiencing some of the lowest yields and highest equity valuations in modern history. Expecting bond yields to fall even further and CAPE ratios to continue soaring doesn’t seem sensible—but that’s precisely what’s happening when investors and their advisors rely on history to gauge the future.***

*Are our models overly pessimistic? Are we simply being “permabears” and raising false alarms? We don’t think so.*^{9}Using the same equity decomposition framework we just explained, Vanguard’s founder, Jack Bogle, recently gave a fairly low assessment of future US equity returns:*Looking forward, Bogle said, in the next 10 years investors should expect only 2.0% from dividends and 4.0% from earnings growth. With P/E ratios at 26.3 (now 30), Bogle said investors could expect to lose 2.0% from P/E contraction, for a total return of 4.0% (Huebscher, 2017).**For an apples-to-apples comparison with our previous real-return estimates, we can net out our US inflation assumption for the next 10 years, 2.1%, to get a real return for US equities of 1.9%, based on Bogle’s estimate of 4.0%. If equities produce a real return of 1.9%, then bonds would have to deliver a return —net of inflation—of nearly 7.2% to get to a 60/40 real expected return of 4.0%. From a starting yield of 2.4%? Not. Gonna. Happen.*
## Comment