Is There Risk to the Risk Premium?
It is obvious now to almost everyone that the stock market returns of the late 1990s were drastically above average. Investors are starting to lower their expected returns from the stratospheric levels of a few years ago, but according to a recent study by John Hancock Financial Services investors are still expecting annual returns of 16% over the next 20 years. Not only is this unrealistic, but dangerous as investors plan for retirement. Investors using these expectations for planning purposes will not be able to build an adequate retirement nest egg. These high expected rates of return also help justify the historically low savings rate. Investors assume that since their portfolio will be growing at such a high rate, they do not have to add additional funds. Once investors realize equities are producing returns substantially lower then expected, the savings rate will have to rise. This in turn will curtail the spending spree that U.S. consumers have enjoyed with obvious repercussions throughout the economy.
So 16% equity returns are unrealistic, but what should investors expect as a long-term rate of return? According to Ibbotson Associates, publisher of the annual yearbook, Stocks, Bonds, Bills and Inflation, since 1926 equity investors have earned an 8% real rate of return. Ibbotson Associates also calculate that bonds have delivered real return of 3%. This means equities have a risk premium of 5%. The equity risk premium is simply the return over bonds that equity investors require for assuming more risk.
The recent issue of Financial Analyst Journal included an article, What Risk Premium is Normal, by Robert Arnott, managing partner at First Quadrant, L.P, and Peter Bernstein, president of Peter L. Bernstein, Inc., and author of Against the Gods and The Power of Gold. The authors ask if investors in 1926 expected real returns of 8%. No, is the short answer and the long answer covers almost 20 pages. Arnott and Bernstein create a model to calculate what investors would have expected the real rate of return on stocks to be over time. They contend the real rate of return should be the dividend yield, plus (or minus) dividend growth, plus (or minus) changes in valuation multiples. Of course investors cannot realistically expect valuation multiples to change when determining future expected returns.
During the early part of the 1900's investors did not perceive the stock market as a long-run economic growth investment. Investors simply tried to buy low and sell high and never considered the notion of buy-and-hold. Additionally, investors assumed that company managers would siphon off any economic gains by issuing more shares to themselves (that sounds familiar), investors did not assume growth in dividends either. (I highly recommend reading Edward Chancellor's Devil Take the Hindmost for a complete history of financial markets which discusses this "robber baron" capitalism.) Investors could not have assumed any expansion of valuation multiples either. And with the U.S. on the gold standard there was virtually no trend in inflation. This leaves the 1926 investor expecting to earn the dividend yield, which was 5.1% at the time. With government bonds yielding 3.7%, investor only required an equity risk premium of 1.4%
In hindsight we know investors earned a 5% risk premium as the real return on stocks was 8% and government bonds returned about 3% after inflation. These results have become regarded as the historical average returns investors should expect. Arnott and Bernstein found that the "historical" 5% risk premium for equities was due to several "accidents."
After World War II, "expected inflation became a normal part of bond valuation." This development resulted in a "one-time shock to bonds that decoupled nominal yields from real yields." As bond investors factored in inflation, nominal yields increased costing bond investors 0.4% a year over 75 years. This accounted for almost one-tenth of the excess return equity holders earned relative to bondholders.
The second accident was the rising valuation multiples that stocks experienced since 1926. Arnott and Bernstein calculate that from 1926 to 2001, multiples rose from 18 times dividends to 70 times dividends currently. This multiple expansion accounts for fully one-third of the excess return over the past 75 years, even though the entire increase in valuation multiples happened since 1984.
Regulatory reform accounted for about one-fourth of the excess return. Prior to 1926, stocks did not pass though much economic growth to shareholders as company insiders issued themselves more shares, which left outside investors unable to fully participate as the company grew. Now shareholders receive a substantial amount of the economic growth generated by companies. This led to a 1.4% increase in real dividend payments and in real earnings.
Lastly, investors have benefited from survivorship bias. Arnott and Bernstein contend that the US has benefited from having no wars fought on its own soil, or having experienced a revolution. Also, during the past century, four of the fifteen largest stock markets at the beginning of the century experienced a total loss of capital (China, Russia, Argentina, and Egypt). Two others came close - Germany (twice) and Japan. While impossible to quantify, the authors say, "U.S. investors in early 1926 would not have considered this likelihood to be zero, nor should today's true long-term investor."
Getting back to the model to forecast what rates of returns investors should expect. As previously stated, Arnott and Bernstein contend that equity returns are derived from the dividend yield, growth in dividends, and changes in valuation multiples.
Investors commonly assume that in the long-run earnings growth will equal GDP growth. The authors show that growth in stock prices is more closely aligned with per capita GDP growth. The reasoning makes sense as well. Investors should only benefit from the growth in the companies they are invested in. Since a growing economy will spur new companies, these new companies will crowd out a portion of the economic growth experienced by existing companies.
During the past 200 years, 85% of equity returns has come from three sources: inflation, dividends, and the rising valuation levels experienced since 1982.
Over the past several years dividends have played a lesser role in total returns. Recently, investors preferred companies that reinvested earnings back into the company in order to achieve faster growth. Tax laws have also influenced investors' preference away from receiving regular dividends. The authors found that it is not historically beneficial for companies to reinvest dividends back into the company. Since 1871 (there is no reliable source of earnings information prior), the average earnings yield has been 7.6% and the average dividend yield has been 4.7%. This results in a "retained earnings yield" of about 3%. In other words, companies should grow earnings and dividends at 3% otherwise investors would rather receive the money and invest it themselves.
Interestingly, the article states that companies experienced an internal growth rate of 1.2% to 1.5% over the past 200 years. Since the economy as a whole has grown at a 3.6% rate, the majority of the growth came from new enterprises and not from reinvesting in existing companies. Oddly enough, in an earlier article, Bernstein found that earnings and dividend growth is actually higher when dividend payout ratios are higher. Bernstein proposes that when managers have less money to invest, they are more selective about their investment projects. The article also points out that growing companies tend to issue more shares, which imposes another drag on per share dividend and earnings growth.
Putting all this together, the article supposes real GDP growth of 3% over the next 40 years with an annual 1% population growth rate. This would yield a ceiling of 2% growth for dividends. Subtracting out the average historical dilution of 1%, results in future dividend growth of only 1% with the economy growing at a healthy 3% rate.
The article determines that the expected real stock returns over the past 192 years averaged 6.1% derived from three components; an earnings yield of 5.0%, per capita GDP growth of 1.7%, less 0.6% shrinkage of dividends relative to real per capital GDP growth. The actual return was 6.8% over the 192 year period. The major factor in the 70 basis point excess return was due to the rise in valuation multiples that started in 1982 with the other "accidents" making up the remainder. The 6.8% real return is also noticeably lower than the 8% return experienced since 1926.
Before determining the risk premium for equities, the expected bond return needs to be determined. The biggest determinate of bond yields is what expectations are for inflation. The authors estimate expected future inflation based on the prior three year inflation rate. Until after World War II inflation was largely non-existent during peace time. By this model, the authors found a real bond rate of 3.7%, comprised of a nominal rate of 4.9% and an inflation rate of 1.2%. The actual inflation rate experienced was 1.4%, so the model is with in the ballpark. Arnott and Bernstein also mention that during the 1950s and 1960s bond investors kept expecting inflation to go away and were mis-pricing bonds for an extended period of time.
By subtracting the real expected bond return (3.7%) from the real expected stock return (6.1%) the authors derived the expected risk premium of 2.4% during the past 196 years. This is well below the 5% commonly assumed today. Additionally, Arnott and Bernstein calculate that the "current risk premium is approximately zero, and a sensible expectation for the future real return for both stocks and bonds is 2-4 percent, far lower than the actuarial assumptions on which most investors are basing their planning and spending."
Stock market history is often reduced to the period since 1926. This time period has been extremely good to equity investors. But it appears that the excess returns stocks offered were due to several one-time adjustments and not a sustainable trend. The article did include one bright spot. Since investors should only expect a 2.5% risk premium, the market does not have to correct as much as it would if investors required a 5% premium. Unfortunately, this "still requires a near halving of stock valuations or a 2 percent drop in real bond yields (or a combination of the two)."