back to Home Page
Client-Only Articles
Email:
Password:
 

 

 Updated February 2021

STOCKS FOR THE LONG RUN?

Abstract:  Long term statistics often conceal volatilities, large declines, and decade or longer periods where equities gave no return and underperformed other asset classes.  Real returns differ dramatically depending upon time of market entry.  From 1929 to 1932 the Dow declined 89% and took until 1954 to break even in price.  Research shows that high equity valuation metrics like those measured by the Shiller CAPE, Buffet indicator, or price/revenue ratio lead to lower future returns but when this occurs cannot be predicted profitabily in advance.  Although the financial industry typically promotes long term investing, the majority of trades on the major exchanges today are done by high frequency trading computers with millisecond to multi-day holding periods.


"Stocks for the Long Run."  This belief is pretty much investment gospel and is repeated incessantly in the financial media.  Simply invest in stocks and let the market bail you out over the long run.  And, in the world of passive and index investing buy and hold is quintessential.  Dr. Jeremy Siegel, Professor of finance at Wharton and author of the very popular book "Stocks for the Long Run" (1998) presents US stock market data showing that stocks returned about 7% above inflation for the past two hundred years and for a twenty year time frame stocks outperformed bonds over 90% of the time.  His equity return estimate is above those provided by Dimson et al for 1900 through 2015 for over 20 countries and Fama Sr./DFA for 1927 to present, both of whom estimate real inflation adjusted returns in equities around 4% to 5% with the lower number likely going forward from 2015.

Very long-term statistics often conceal volatilities, large drawdowns, and decade or longer periods where stocks give no return and underperform other asset classes.  This paper examines historical data and argues for caution in uncritically accepting the hypothesis that stocks are always the best investment for all investors for the long run.

THE LONG RUN MAY TURN OUT TO BE EXTRAORDINARILY LONG, far longer than an investor's investment horizon.  If an investor entered the market last century when the Dow was one-standard deviation above its long-term price trend line, an exuberant bull market top, how long did they have to wait?  Leuthold (InvestmentNews, 5/21/2001) notes that an investor at the peak in 1929 took until August, 1998, almost 69 years later, to reach a nominal return of 10% on their money, including dividends.  This is an after inflation yearly return of about 7%, what Dr. Siegel claims is available from stocks.  Thus, it took 69 years for an investor to reach the long-term average return for stocks.  Had an investor in 1929 relied upon long-term stock return data to calculate their future net worth at retirement or retirement distribution rate they would have been sorely disappointed unless they lived an extraordinarily long time.

In addition, from its peak in 1929 to its low in July, 1932 long-term investors in the Dow Industrials had to endure an 89% decline in the value of their portfolios.  The Dow Industrials held some of America's largest and financially soundest companies in strong industries and cannot be considered an aggressive or speculative part of the stock market.  Many investors in 1929 lost all their money.  Only one stock, GE, was in the DJIA in 1929 and it was removed in 2018.  Indexes are not constants and are continually changing.  Investors choosing this relatively conservative sector of the stock market would have had to have extraordinary nerves and an abundance of decades to see a long-term average return on their investment of 7% after inflation if they saw it.

Long-term U.S. stock market data, beginning in 1802, show that stocks did much better relative to bonds in the twentieth century than in the nineteenth.  Bonds outperformed stocks from 1802 though 1839.  Since there have been five 40 year periods in the last 200 years, this data suggests that the odds that bonds will beat stocks for a 40 year period are roughly one in five or 20%, that's low but not insignificant.  Mark Hulbert reexamined Siegel's data in November 2008 and found that for 19.9% of the last 200 years stocks didn't beat T-bills for a 10 yr. holding period, 5.5% of the time they didn't beat T-bills for a 20 yr. period, and 2.9% of the time they didn't beat T-bills over a 30 year period.  Siegel's data is questionable since price data from before 1890 is very incomplete and of poor quality and real returns in equities may have been worse during the 1800's then is reported in the historical financial literature. This is just another clear reminder that for most investors, equity exposure is not a guarantee of growth over their probable investment time-frames.

"Triumph of the Optimists",  published in Spring 2002 by Dimson, Marsh, and Staunton, is probably the most ambitious study of financial market returns undertaken to date.  It is updated yearly and published online in the "Credit Suisse Annual Investment Returns Yearbook", available free online.  Aggregate global returns suggest that there is about a 17% chance that equities will underperform short-term government bills for a 50 year period and about a 25% chance for a 40 year period.  This is not insignificant.  They conclude, "There is clearly a substantial probability of achieving a negative risk premium, even over long investment horizons...equity investment is not that compelling as a short or intermediate term strategy."  Forty or fifty years is not what most investors are thinking when they are thinking about the "long run".  The Wall Street Journal (2-10-03) offered further insights into long-term risks in Dimson's 2008 pre-financial crisis update on global investment returns.  Out of 16 major national stock markets, investors from only five, only 31%, would have been guaranteed positive annual returns over every 20 year period during the last century.  A November 10th, 2008 Business Week article examined long-term stock returns and concluded that statistics show an advantage for stocks but it all depends on how long you are willing to wait.  They note inflation adjusted equity returns of zero last century for 53 years in France, 55 years in Germany, 51 years in Japan, and 73 years in Italy, not exactly appetizing.

Truman Clarke of DFA (Spring 2003) examined the probabilities that U.S. stocks will beat T-bills over various holding periods.  He employed a statistical procedure called bootstrapping to create 10,000 simulated histories of one through twenty-five year periods based upon data from 76 years of returns on three U.S. equity asset classes.  For a twenty-five year period, the probabilities ranged from 82.4% to 96.2%, certainly not the sure thing that most of Wall Street promotes, certainly not a certainty for 25 year returns.

Barry Ritholtz (The Big Picture website) looked at relative asset class returns for 1970 through 2010.  Ranked in terms of total return for that period, oil, up 3807%, gave the highest return with gold close at 3276%, long Treasury bonds up 1720%, the S&P 500 up 1264%, housing up 905%, and inflation increasing 480%.  In two articles in August 2012 Ritholtz notes that Siegel's "Stocks for the Long Run" is a deeply flawed and erroneous book though it is one of the most widely read books in economic classes and business schools.  He quotes Jason Zweig, "There is just one problem with tracing stock performance all the way back to 1802:  It isn't really valid."  Siegel relied on cherry picked data.  Of over 1000 stocks in existence between 1802 and 1845, Siegel ignored 97% of them, producing enormous survivorship bias and greatly boosting 200 year returns due to erroneously overweighting returns for the earliest period.  Ritholtz also quotes Birinyi Associates on their critique of Siegel.  Siegel's thesis reveals he did little original research and is based primarily on "cobbling" together numerous studies which vary greatly in composition and methodology.  Many researchers question the quality of market data prior to 1871 and Siegel's estimate of a total excess real return of 8.3% for 200+ years in equities is achieved by assuming high dividends and high dividend growth, well in excess of that suggested by other analysts and multiples of today's low dividend yields.

Perhaps the most persuasive case for caution in assuming that equities are always the best choice for long-term portfolios is a relatively recent 30 year period, 1981 through 2011.  Long-term government bonds gained an average of 11.5% per year while the S&P 500 index with dividends reinvested gained 10.8% per year over the same period but with far more risk and volatility.  Stocks had risen more than bonds over every 30-year period from 1861 until 1981.  DFA provides data showing that from 1940 through 1990, 50 years, the inflation adjusted return on long-term government bonds was 0%, breakeven.  Floyd Norris (NYT, 1-7-12) presents charts showing the average annual inflation adjusted total return of the S&P for prior 15 year periods during two much longer periods, December 1943 through December 1980 and December 1980 through December 2011.  The real trailing 15 yr. returns for each year during 1944 through 2011 are plotted and they look nearly identical, rising from 0% to 15% in 1964 then declining back to -2% by 1980, then rising again to 15% in 1999 and declining back to 2.5% by 2011.  Long term returns in any asset class are not a constant and continuously vary depending upon the starting date and termination date chosen.  There is just no way of knowing for certain at what point in asset class returns one might be entering a bull or bear cycle in prices.  There's no way of knowing with certainty which asset class will do best during a 30 year period since any asset class can go out of favor for 30 years or longer.  We at EAM have long found it interesting that when securities prices are at record highs like early 2021 there is an assumption that prices won't decline from record highs and optimism that gains, with brief interruptions, will occur forever.  This has never been the case.

REAL RETURNS DIFFER DRAMATICALLY DEPENDING UPON TIME OF MARKET ENTRY.  For the typical investor with a twenty or twenty-five year time frame this can make very large differences in money available during retirement.  Frequently presented "mountain" charts smooth out and conceal what would have happened to unfortunate investors who happened to enter the market at or near long-term secular bull market peaks.

An investor in the S & P 500 from 1929 through 1949 received an inflation adjusted return of 4.5%.  Yet, an investor beginning in 1932 and holding until 1951 received an inflation adjusted annual return of 10.8%, over 6% greater per year from starting at the bottom of a bear market.  Our 1929 investor received a compound total return of 84.3% for twenty years.  Our 1932 investor received a compound total return of 818.1% for twenty years, ending up with almost ten times as large a portfolio as our unfortunate 1929 investor.  Extend the holding period out to twenty-five years and our 1929 investor does a little better, receiving a compound total return of 319.3% but our lucky investor still receives far more, 2202%.

Let's compare returns from the last big secular market top and bottom.  An investor in 1968, a market top, had to wait until 1983, fourteen years later, to breakeven after inflation.  If they waited until 1987, twenty years later, their annual return was 4.2% after inflation and their compound total return was 489.2%.  If they waited until 1992, twenty-five years later, their annual return was 5.8% and their compounded return was 1132%.  The big bull market ending at an all-time Dow high of 14150 in October 2007 began in 1982 if we consider the 2000 through 2002 sell-off as a correction.  An investor in the market from 1981 through 2000 received a whopping after inflation return of 12.9% annually, and a compound total return of 1741%.  It was the best twenty-five years in U.S. equity market history.  Yet, an investor in Japan at its peak in 1989 still had not recovered to breakeven over 28 years later in 2020 despite one bear market rally of over 100% and several over 50%.  The large cap Japanese Nikkei declined around 50% to 60% from 1989 through 2016.  Japanese residential real estate and other equities were also down for 1989 through 2016.  Time of market entry matters greatly to returns yet there's no evidence markets can be reliably and profitably timed.

In 1997, Peter L. Bernstein, author of several popular and well-respected books on investing, cautioned investors about employing the long run as a benchmark because the long run is not a homogeneous state of the world and a smooth straight line up into the future.  A very large portion of the vaunted equity returns over the risk free return, the equity risk premium, is accounted for by just 32 years out of 200 years since 1800, 1950 through 1981.  Bonds outperformed stocks 43% of the time, and stocks were superior over the very long run, but with a high degree of uncertainty.  He concludes that, "...the long run can tell us perilously little about what kinds of environments lie ahead...we have to accept uncertainty."

Easterling, E. in "Unexpected Returns:  Understanding Secular Stock Market Cycles" examines secular bull and bear markets and how they interrelate with price/earnings ratios (p/e's), dividends, GDP growth and inflation.  Historically, secular bull markets have run as long as 24 years (1942 through 1965) and as short as 4 (1933 through 1936).  On average, they lasted 13.5 years.  Secular bear markets have averaged 11.3 years and have ranged from 4 years (1929 through 1932) to 20 years (1901 through 1920).  Easterling found that for 85 twenty year periods starting in 1919, twenty years being a reasonable investment time horizon for many investors, starting p/e's were strongly correlated with forward returns.  When p/e's were 19 or higher, forward 20 year nominal returns ranged from 1.5%-4.5%.  When p/e's averaged 10, forward returns averaged 11.9-15.0%.  As of Fall 2020 the Shiller CAPE, an averaged p/e ratio with a strong 0.85+ correlation with 10 year forward returns, was around 29:1, twice median Shiller CAPE and at record highs and an extremely poor value.  The long-term average Shiller CAPE is under 15:1.  Interestingly, bear markets exhibited much higher volatility than bull markets.  Easterling's data clearly indicate that high 20 year returns occur when p/e's are low (under 10:1), dividends are high (over 5%), inflation is moving towards stability, and interest rates are moving solidly lower.  As of 2020 none of these conditions were in place.

It's very clear from the above studies that luck or chance in the form of time of market entry plays a major role in the returns most investors can expect to receive from stocks.  Enter at or near a market bottom and good values and you may receive 6% to 8% or more annually than the investor who who enters at a market top.  These are huge differences.  Enter at a market top, and you may have to live through a large decline and receive only a breakeven return for a decade or more.  Unfortunately, there is little evidence that anyone can accurately time big market tops or bottoms or know for certain where we are in long term secular cycles.

HIGH STOCK VALUATIONS LEAD TO LOWER FUTURE RETURNS.  Extensive research has demonstrated that the stock market is unlikely to be profitably timed. (Sherden, The Fortune Sellers, 1998).  See "Tactical Asset Allocation" on our website as well as "Active vs. Passive Investing".  There is, however, persuasive evidence that high stock valuations based upon traditional measures of value eventually lead to low future returns.  Robert Shiller of Yale (Irrational Exuberance, 2000) looked at twenty year returns following market peaks in price/earnings ratios and found that inflation adjusted annual returns averaged -0.2% to +1.9%, far below long term market averages.  Smithers and Wright (Valuing Wall Street, 2000) looked at historical data on the Q-ratio, similar to market-to-book value but based upon corporate replacement costs.  The Q-ratio has an even better record predicting future market returns than p/e's or dividend yields.  They found that bull market peaks in Q have produced later declines in prices from 50% to 80%.  As of 2016 the Q-ratio was 1.00, indicating valuations were very high and comparable to 1929 though a little below early 2000.  Overvalued markets have low expected future returns but overvaluation does not tell us when markets will correct and return to average or below average valuations and why.  J.M. Keynes highlights this by his remark that, "markets can stay irrational far longer than you can stay solvent."

Fama and French (private paper, 2000) and Asness (Bubble Logic, 2000), looked at other valuation measures, including dividend payouts and the so-called Fed model, and reached similar conclusions about future equity premiums and returns.  As Shiller's p/e model predicted, the returns of the S&P 500 for January 2000 through December 2009 were -4.20% with the Dow about breakeven for the decade.  On August 21, 2001, the Wall Street Journal reported that, based on earnings as reported under GAAP, the S & P 500 actually finished the prior week with a p/e ratio of 36.7, higher than any other p/e previously recorded for the index, well above the 1929 peak, and 2.5 times the average historical p/e.

Past corporate acccounting differs substantially from accounting in the 2000's.  Today, most earnings reported by the financial media today only emphasize non-GAAP earnings and don't always mention that they are doing so.  In 2015 non-GAAP earnings were 27% higher than GAAP earnings in the S&P, well above the more accurate GAAP earnings.  As of 2020 most methods for calculating market valuations showed them at or near all time historical highs and optimism about future returns very high.  Even in early 2009 at the end of the 2008 financial crisis and before the huge market rally from 2009 through 2018, p/e's had only declined to average levels by historical standards.  Equities were not a bargain in early 2009 and only at average long-term value.  From the March 2009 lows through Fall 2019 the S&P then more than quadrupled in price and p/e's climbed to record or near record levels far from average levels.

Record or near-record overvaluation and bullish optimism persisted in early 2021.  In September 2016 DFA examined 680 valuation based timing strategies.  The vast majority underperformed well-allocated passive and index portfolios with set allocation targets.  Only 16 of the 680 valulation based timing strategies, 2.4%, outperformed passive portfolios with a positive excess return of more than 0.02% per month, again showing what extensive prior research had shown: markets cannot be timed and timing equity markets based on equity valuation metrics is unlikely to increase returns significantly.  However, historical data clearly show that purchasing equities when they are at or near record overvaluations has always eventually led to large declines in equity prices.  The central problem is that values revert to average or below and when this occurs cannot be predicted and timed in advance.  DFA also notes that "on average" narrow term spreads are associated with lower future fixed premiums (returns) in fixed income.  As of 11-25-19 the yield curve was very narrow with what's considered the safest security in the world, the US Treasury, yielding about 1.5-1.7% for 3 month to 5 year periods while investment grade corporates are only yielding about 1.7% to 2.1% for 2-3 year maturities and 2.4-2.7% for 3-7 year maturities even though corporates are considered below the credit quality of Treasuries, prices fluctuate more and defaults and delinquincies can occur.  The spread between these two categories of fixed income securities usually runs 2% to 4% and sometimes has been far higher than that.  As of early 2021 yields on bond funds are even lower with the mighty VBTLX Vanguard total bond market fund yielding an abnormally low return of 1.1%, over 1% below inflation.

In 2006, for the first time ever more corporate cash flow was devoted to stock buybacks than any other corporate expense, 51%.  This is standard practice as of 2020.  Stock buybacks in 2018 broke records with most of the buybacks funded by bond issues and borrowing.  In 2015 over 100% of corporate net income went into buybacks and dividends while capital expenditures declined.  Several credible analysts estimate that for the entire 2009-2018 period buybacks accounted for all stock price gains.  Capital expenditures are essential for future corporate growth but the Fed's record low interest rates have stimulated financial engineering, i.e. buybacks, that reward top management the most.  Stock buybacks drive up earnings per share numbers as stocks are retired and they also drive up stock prices thus enriching executives, directors and employees who are able to grant themselves large option and stock positions.  Something few investors are aware of is that aggregate p/e ratios for the S&P 500 and other indexes are reported with negative earnings removed.  This distorts the market's pricing mechanism, driving up average "e's" and driving down average p/e's, thus making stocks look more attractive, an important goal of Wall Street.  In addition, close to half of all S&P earnings are kept offshore in order to avoid US taxes and don't benefit the US economy and its workers much if at all.  This strategy, however, greatly benefits the taxes corporations pay on their earnings and compensation to top management.

PAST ASSET CLASS RISK AND RETURN DATA ARE DEPENDENT VARIABLES, NOT INDEPENDENT VARIABLES.  Passive and index allocation strategies focus on quantifying past risks and returns, not the factors that produced them.  Past numbers are the result of countless unpredictable events in the real world.  These include geopolitical conflicts, competitive advantages or disadvantages with other nations, new technologies and industries which arise, creative destruction of older industries, new investment instruments like derivatives and CDS's, changing societal values about consumption and debt, and countless other factors.  These events are wholly unpredictable yet influence economic variables which influence investment risks and returns.  We discuss this on our website in "Financial Superstructures".

REALISTIC LONG-TERM EQUITY RETURNS ARE PROBABLY IN THE 4% TO 5% RANGE, INFLATION ADJUSTED.  Shiller argues that the U.S. was the most economically successful nation in history from 1900 through 1999.  Therefore, recent historical U.S. return data up until the 2000's almost certainly exaggerates returns and the United States in the twentieth century may have been been the exception rather than the rule in terms of expected real returns for stocks.  Jorion and Goetzmann (J. of Finance, 54[3], 1999) studied inflation adjusted stock market appreciation, excluding dividends, for thirty-nine foreign countries for the period of 1926 through 1999 and found that the median real appreciation rate was only 0.8% per year compared to 4.3% per year for the U.S.  Dimson et al. make a similar point.  The favorable social, political, economic, technological, and military conditions that lead to past high U.S. equity returns may not repeat themselves in the future.  Dimson et. al. estimate a long-term inflation adjusted equity premium (return above Treasury bills) of 3.5% to 4.5% annually from globally diversified equity portfolios.  Fama and French of DFA offer similar estimates.

ALTHOUGH WALL STREET EMPHASIZES STOCKS FOR THE LONG RUN FOR INVESTORS, ITS INVESTMENT TIME FRAME TODAY IS MEASURED IN MILLISECONDS.  Historical data show that average professional holding periods for stocks up through the 1960 were in the seven year range while today they are in the three month range.  But, that is an average.  As of 2019 50% to 95% of all daily trades on the major US exchanges was executed by high frequency trading (HFT) and "dark" trading pools where equities. and the FX and commodity markets are also dominated by HFT.  Ultra-fast computers are placed close to exchanges and buy near instant data from exchanges that arrives nanoseconds before retail investors see the price on their screens, just enough to give HFT traders a front runner's advantage over retail.  Their typical strategy is to place place short and long trades thousands of times a second to scalp a penny in front of other HFT computers and retail investors and pension funds, a form of front running, casino gambling, and market manipulation.  Despite the spectacular 900+ Dow "flash crash" of May 6, 2010, now known to be largely the result of HFT computers, regulatory agencies have done little to limit this.  Wall Street likes to take its money off the table as quickly as possible and most certainly doesn't invest for the long run.

CONCLUSION:  Do not buy Siegel's "Stocks for the Long Run" hypothesis uncritically.  It is based upon the presentation of very long-term annualized returns data that is questionable.  It is unlikely to accurately pinpoint what a real investor will experience in a real market over their investment lifetime.  It most definitely underemphasizes the pain and disappointing returns stock and bond investors may suffer if they happen to enter the market at the wrong time.  Wall Street and the money management industry have a vested interest in staying optimistic and moving investors into equities because they get paid more for buying and managing stock portfolios than for bond portfolios or for cash.  Wise investors should look at both sides of the risk/return picture, the investment industry's interests, and adjust their equity/fixed ratios accordingly.

Investors must be also be very careful about assuming past long-term returns will repeat within twenty year periods, particularly if they are dealing with lump-sum or retirement monies and they are over 65 years old.  If that's the case, increasing allocations to bonds is one means of controlling the risk of large losses that can affect income and lifestyle during retirement.  Older investors in markets with high valuations may not live long enough to enjoy the benefits of long term stock investing if they add stocks at high valuations.  They may even outlive their principal if a protracted bear market occurs and they have a high equity allocation.  Younger investors who contribute yearly have an opportunity to buy when markets are down, and much time, which makes it likely, but not certain, that they will still be in the market after the bear has exited if they are not scared off by large market declines.  EAM recommends that investors not place money in equities unless they have a 15 year or longer time frame.  Even at 10 years the odds that equities will outperform fixed income are only about 55% to 45% outperformance for fixed income.

Lump sum and retired investors should also be cautious when using programs which purport to calculate how much money they will accumulate while investing for retirement and how much they can withdraw each year during retirement.  These programs are nearly meaningless when they use linear compounding calculations.  The most frequent and more accurate approach is to use Monte Carlo probability estimates.  But even Monte Carlo simulations depend upon past data and certain assumptions programmed into the calculators.  We cannot know future inflation rates, tax rates, and returns on equities or fixed income.  EAM does assist clients in calculating prudent distribution rates for retirement portfolios.

In sum, a prudent investor should look closely at risk and return data, worst returns data hidden in long-term statistics, and critically evaluate whether the widely accepted belief that stocks are always their best long-term investment fits their particular financial circumstances and temperament.

Site Navigation