At the heart of modern portfolio management is the concept of “strategic asset allocation.” Basic to this concept is the phenomenon that, by combining investments in two or more asset classes that react differently to outside market forces, one can achieve returns representing the average returns of the asset classes, but whose expected risk (volatility) is less than the average risk of those asset classes. The more prices of the various asset classes behave differently, or are “non-correlated,” to use investment jargon, the greater the amount of risk avoided.
Imagine two assets, A and B, whose prices always move in opposite directions by the same amount. If the expected return from A is 10 percent per year and that of B is 5 percent, a portfolio that’s a 50-50 mix of A and B will have an expected return of 7.5 percent and a volatility of zero, since the price movements of A and B will exactly cancel each other out. That’s the dream of every portfolio manager.
Modern portfolio managers focus much attention on the non-correlation of asset classes. This attention has led to the widespread implementation of the “endowment model” for investing, pioneered by the Yale University Endowment.
Unfortunately, managers have paid insufficient attention to the expected returns of many assets classes they’ve sought for their portfolios. The most common sin they’ve committed is equating expected return with historical return.
Consider gold. You’re contemplating adding some gold bullion costing $100,000 to your $100,000 bond portfolio, to hold for 10 years. Before investigating whether bond and gold prices are uncorrelated, you need to determine what returns you can expect from the bonds and the gold. For this exercise, let’s assume the bonds will be worth exactly what you paid for them because the bonds come with a promise to return your principal to you in cash 10 years hence. Between now and then, the bonds will pay you 5 percent each year in interest. We can safely assume an expected return of 5 percent annually, based on the contractual promises to pay interest and to repay principal.
How much will the gold bullion be worth after sitting in your safe for 10 years? We know the gold will pay you no cash interest or dividends during the period you hold it. So your expected return can only come from price appreciation or depreciation. Here’s where the problem of mixing historical returns with expected returns arises. Reviewing the price of gold over many decades, we find wild swings in prices and significant interference of governments in the market. Indeed , it was illegal for most U.S. citizens to own gold from 1933, when President Roosevelt signed Executive Order 6102, until the restrictions were lifted at the end of 1974. So whether the historical returns from gold are positive or negative over any 10-year span depends entirely on what period you choose to investigate.
For purposes of analyzing likely investment results, it’s appropriate to use an expected return of zero for gold or, indeed, for any commodity. But my purpose isn’t to pick a fight with the “gold bugs.” I’ll let them advance their own arguments that people should hold gold as a hedge against government folly and economic collapse. Rather, I want to explore the appropriate way to determine the expected return from financial asset classes, specifically stocks.
In 1962, Myron Gordon, professor emeritus of finance at the Rotman School of Management, University of Toronto, published a paper exploring expected returns from stocks. As is typical of academic investment papers, Professor Gordon’s was filled with abstract concepts and complex formulas. But the idea it generated was very simple and is readily understandable by anyone who passed basic math: The expected return from stocks is equal to the sum of their dividend yield plus the expected rate of divided growth.
For the past five decades, Professor Gordon’s model has been met with near-deafening silence. Other academics have advanced numerous models for determining expected returns from stocks, many of which are designed to cope with the fact that many stocks don’t pay dividends. Some companies explicitly eschew dividends, which are returns on capital made at the discretion of a company’s board of directors. Under the Gordon model, the expected return of stocks that don’t pay dividends must be zero, since, by definition, the rate of dividend increase for a company that pays no dividends is zero. So, zero yield plus zero rate of increase equals zero return.
Surely Gordon must be wrong; millions of investors gladly pay handsome prices for stocks of companies that don’t pay dividends (think Google). Does the evidence bear out the Gordon model?
To find out, I turned to research done by Ned Davis Research, located in Venice, Florida. The organization has kept data series since 1972 that track the returns of stock portfolios broken into categories by dividend policy and practice—companies that pay no dividends, those that cut their dividends, those that maintain their dividends and those that grow their dividends.
From inception in January 1972 through February 28, 2009, the average annual return from stocks that maintain their dividends has been 5.52 percent and, for those that increase their dividends, 8.19 percent. How about the non-dividend payers? Their average return during the past 37 years has actually been -0.10 percent. Let’s not quibble and just call it zero.
At no time during the period covered by the Ned Davis data has the average historical return for non-dividend paying stocks exceeded 5 percent, and it’s mostly been far below that of less risky bonds. For me, the Ned Davis data emphatically confirms Gordon’s model. Because zero plus zero equals zero, I avoid investing in stocks that don’t pay dividends.
David Raub has 19 years’ experience as a registered investment adviser. He is co-owner of Raub Brock Capital Management in Larkspur. You can reach him at draub@northbaybiz.com.
