Kevin Spires, CFA, FRM
Another way to calculate expected returns in the stock market is explored by Mark Hulbert at Marketwatch. In his article titled, "Stocks' future returns: just 5.6% annualized," Hulbert explains that a variant of a model he uses "says the long-term return will be a function of just two things: the current dividend yield and real growth in earnings and dividends."
Hulbert then goes on to actually break this down into three things which are the current dividend yield (he never actually states it in the article) the long term growth in earnings of 1.4% and expected inflation (of who knows what). According to Hulbert "These three components today add up to a nominal return of 5.6% annualized, according to Rob Arnott, founder of Research Affiliates, an investment advisory firm — or 3.4% in real terms." In other words, the formula is roughly (because I mistate the formula by leaving out compounding):
Future Returns = Dividend Yield + Dividend Growth Rate + Inflation
Three things to explain 10 year expected returns in the market. This model does have some good points to recommend to the casual investor. The first is the use of a valuation metric - dividend yield - into the calculation of future expected returns. The second is the use of inflation - because we can't eat t-bills - in the calculation of inflation adjusted returns. Unfortunately, there is some bad math involved in their calculations that make the prediction less accurate than it would be if it included some other factors.
I prefer the CAPE10 (popularized in Shiller's Irrational Exuberance) to dividend yields as a valuation metric. Historically, the dividend yield and CAPE10 were closely tied together, but with the advent of modern finance in the '80s, firms started returning capital to shareholders in the form of share buybacks and the ratio of earnings paid out as dividends decreased. The reasons are many for this change, but the main two were the lower tax rates on capital gains and the ability of an investor to defer capital gains taxes indefinitely. These two advantages to share buybacks versus dividends have lowered the dividend payout ratio to the point where I think it is less useful as a predictor than the CAPE10.
But I digress. The dividend yield has been useful as a predictor of future expected returns in the stock market because it has the current market level as a denominator. Higher current valuations result in lower dividend yields and vice-versa. Not only will the dividend yield be smaller if the current market valuation is high, but the future returns will be smaller as the market level returns to a more normal level - resulting in capital losses like we experienced in the period following the tech bubble, or capital gains like we experienced in the 1980's after the extremely low market valuations recorded after the double dip recessions to start that decade. Because share buybacks are now such a normal way that firms return capital to investors, I don't think the dividend yield is very useful going forward - and using the dividend yield in a model like the one Hulbert uses will cause major errors in the future.
The second good feature of this method is the use of inflation to discount the return. Investors care about making more than inflation, so explicately solving for after inflation returns is invaluable when figuring out how much a person will have to spend in the future given their current level of savings.
Finally, the use of an expected growth metric is great as well - although I think using the historical growth rate of dividends is a flawed method for the same reasons I would not use the dividend yield as an input. According to Shiller's work, dividends paid out as a ratio of earnings are less than half what they were just 30 years ago - because firms now return earnings to investors in the form of share buybacks. If current dividends were doubled, the long term growth rate of dividends would be much higher as well. I prefer to relate earnings growth to the nominal gdp growth rate - both for its close historical correlation to profit growth and because it allows one to break out real gdp and inflation as seperate factors that both flow back into nominal profit growth (let's leave that to another post). But to summerize, the back of the napkin future returns formula I use looks like this...
10YR Future returns = Current Earnings Yield (1/CAPE10) + Real Growth in Earnings + Inflation + Mean Reversion to Fair Value Return.
...where the Mean Reversion component assumes market valuations return to fair value over the next 10 years.
In conclusion, Hulbert's article is a good, quick introduction to a simplistic method of calculating stock market returns over the next 10 years. I think it would be fastly improved by the use of the CAPE10 in the same simplistic model versus using the dividend yield, but it is worth reading on its own merits.
-Kevin Spires
Another way to calculate expected returns in the stock market is explored by Mark Hulbert at Marketwatch. In his article titled, "Stocks' future returns: just 5.6% annualized," Hulbert explains that a variant of a model he uses "says the long-term return will be a function of just two things: the current dividend yield and real growth in earnings and dividends."
Hulbert then goes on to actually break this down into three things which are the current dividend yield (he never actually states it in the article) the long term growth in earnings of 1.4% and expected inflation (of who knows what). According to Hulbert "These three components today add up to a nominal return of 5.6% annualized, according to Rob Arnott, founder of Research Affiliates, an investment advisory firm — or 3.4% in real terms." In other words, the formula is roughly (because I mistate the formula by leaving out compounding):
Future Returns = Dividend Yield + Dividend Growth Rate + Inflation
Three things to explain 10 year expected returns in the market. This model does have some good points to recommend to the casual investor. The first is the use of a valuation metric - dividend yield - into the calculation of future expected returns. The second is the use of inflation - because we can't eat t-bills - in the calculation of inflation adjusted returns. Unfortunately, there is some bad math involved in their calculations that make the prediction less accurate than it would be if it included some other factors.
I prefer the CAPE10 (popularized in Shiller's Irrational Exuberance) to dividend yields as a valuation metric. Historically, the dividend yield and CAPE10 were closely tied together, but with the advent of modern finance in the '80s, firms started returning capital to shareholders in the form of share buybacks and the ratio of earnings paid out as dividends decreased. The reasons are many for this change, but the main two were the lower tax rates on capital gains and the ability of an investor to defer capital gains taxes indefinitely. These two advantages to share buybacks versus dividends have lowered the dividend payout ratio to the point where I think it is less useful as a predictor than the CAPE10.
But I digress. The dividend yield has been useful as a predictor of future expected returns in the stock market because it has the current market level as a denominator. Higher current valuations result in lower dividend yields and vice-versa. Not only will the dividend yield be smaller if the current market valuation is high, but the future returns will be smaller as the market level returns to a more normal level - resulting in capital losses like we experienced in the period following the tech bubble, or capital gains like we experienced in the 1980's after the extremely low market valuations recorded after the double dip recessions to start that decade. Because share buybacks are now such a normal way that firms return capital to investors, I don't think the dividend yield is very useful going forward - and using the dividend yield in a model like the one Hulbert uses will cause major errors in the future.
The second good feature of this method is the use of inflation to discount the return. Investors care about making more than inflation, so explicately solving for after inflation returns is invaluable when figuring out how much a person will have to spend in the future given their current level of savings.
Finally, the use of an expected growth metric is great as well - although I think using the historical growth rate of dividends is a flawed method for the same reasons I would not use the dividend yield as an input. According to Shiller's work, dividends paid out as a ratio of earnings are less than half what they were just 30 years ago - because firms now return earnings to investors in the form of share buybacks. If current dividends were doubled, the long term growth rate of dividends would be much higher as well. I prefer to relate earnings growth to the nominal gdp growth rate - both for its close historical correlation to profit growth and because it allows one to break out real gdp and inflation as seperate factors that both flow back into nominal profit growth (let's leave that to another post). But to summerize, the back of the napkin future returns formula I use looks like this...
10YR Future returns = Current Earnings Yield (1/CAPE10) + Real Growth in Earnings + Inflation + Mean Reversion to Fair Value Return.
...where the Mean Reversion component assumes market valuations return to fair value over the next 10 years.
In conclusion, Hulbert's article is a good, quick introduction to a simplistic method of calculating stock market returns over the next 10 years. I think it would be fastly improved by the use of the CAPE10 in the same simplistic model versus using the dividend yield, but it is worth reading on its own merits.
-Kevin Spires
RSS Feed
