The Dividend Discount

The classic way to value a company is the dividend discount model, or DDM. The first serious model was invented by John Burr Williams in 1938 in

The Theory of Investment Value

. He started from the premise that serious investing was done in the bond market. The value of a bond was determined by the present value of all the coupon interest that would be received over the life of the bond, plus the present value of the principal repayment at maturity. Doing the arithmetic in 1938 was somewhat laborious, but tables had been compiled, making it easy to look up the answer.

Williams argued that the value of a common stock was the present value of all the dividends that an investor would receive in the future. A stock does not have a maturity date, so one doesn’t have to worry about principal repayment. One does have to consider the difference in payout stream, because bonds pay a constant coupon for a known number of years, and common stocks pay a dividend stream that can go up or down in the next year. Furthermore, the length of that dividend stream is unlimited.

Winter Zombies From today's perspective, it is generally believed that company earnings, and therefore dividends, will be in a continuing uptrend reflecting the growth in the economy and most businesses. That would not have been obvious to an analyst in 1938. The economy of the United States and of the rest of the world evolved rapidly in the 40 years that Williams would have considered, but there was no great evidence of steady growth.

An analyst would have remembered the catastrophes of the Great War and the Depression of 1921, then the euphoria of the late 1920s, followed by the Depression, when the ghosts of bankruptcy and Bolshevism caused the same fear of the future as when the White Walker zombie army crashed through the Wall into a helpless Westeros in Game of Thrones. The Dow Jones Industrial Average was 100 in 1905 and 100 in 1942. So, the idea of a buy-and-hold strategy would have seemed preposterous to the handful of investors who had not already been turned to zombies in the dreaded winter of 1932.

There are several practical problems with the Williams dividend discount model. First, the present value of an income stream extended far into the future is so hard to estimate correctly that the present value calculation is nearly meaningless. The math tends to be dominated by events more than 10 years in the future about which you and I know nothing. If you think otherwise, find a collection of the analyst reports that you and your colleagues wrote 10 years ago.

Another blow to the DDM is the disappearance of dividends, leaving nothing to discount or model. In the early 1990s, corporate managements made the agreeable discovery that the success of their company was primarily due to the managerial genius of themselves. Therefore, because the increase in stock prices was created by these managers, they were clearly undercompensated for the riches they were bestowing upon their indolent shareholders.

The invention of stock options righted this wrong. It allowed the management group to seize 4% of the company every year, under the threat that less compensation would cause them to transfer their wealth-making power to a competitor. In this model, dividends were a nuisance because they were not factored into the option calculation. Dividends were cut or eliminated, and the money that would have gone to shareholders was put into stock-repurchase programs. A large stock-repurchase program would offset the stock dilution caused by the stock-option program. That way, at the end of the year, the company would have the same number of shares outstanding but with the managers owning more and the shareholders owning less. This transfer of wealth from owners to managers has been remarkably easy. Shareholders never knew what hit them.

Sell Signals Despite the numerous shortcomings of the dividend discount model, it has been useful in the past and should continue to have value in the future, probably by broadening the definition of dividends to include stock repurchases.

I first used it as a theme of my senior thesis at MIT in 1955. I took the Williams model of 1938 and fiddled with it to account for senior securities on the balance sheet and added a way to taper future growth to adjust for the inevitable mean reversion of a highly profitable company. There was a new DDM in academia at that point, the Gordon-Shapiro model, which led to the equation p=d/(r-g).1 I was able to show that this was formally the same as the Williams DDM.

When I started running a mutual fund in 1970, I thought that a DDM would be useful in comparing stocks in different industries, so I adopted a model invented by U.S. Trust. At that time, U.S. Trust was a leading institutional manager using the same “Nifty 50” strategy as every other bank in New York. By 1972, the model was throwing off signals that stocks were wildly overpriced. The terrible bear market of 1973–74 proved that the DDM had given a valid signal. Everybody saw their portfolios drop 50%.

In 1974, I called one of my buddies at U.S. Trust to congratulate him on the success of the firm’s model forecast. My friend was not in a good mood. He told me that the bank, seeing that the model was generating sell signals, did the sensible thing and scrapped the model. My buddy had been fired and was just in his office to pack up the kids’ pictures and the golf trophy. (U.S. Trust was acquired by Bank of America BAC in 2007.)

Future Value The world has changed a lot since the 1960s. In those days, most businesses were capitalintensive; to grow, they needed a high return on equity. They could then pay out half the profits in dividends and plow the other half into expanding their factory. The clearest example of this was the electric utility industry. The number of kilowatt-hours used in the country grew at 7% per year from 1912 to 1972. Utility companies, therefore, had to continuously build new generating capacity. State regulators allowed the utilities a high rate of return on the new plant, so that shareholders received a constantly increasing dividend stream. Because the technology of generating electricity improved steadily, utilities were able to reduce the price of electricity to consumers, delighting the public and the regulatory agencies. 'Twas a veritable paradise on earth.

The DDM was ideally suited to value electric utility stocks, and I remember having earnest discussions as to whether the long-term growth of Commonwealth Edison would be 6% or 6.5%. Then in 1973, OPEC found that it had the power to quadruple the price of oil, and the price of coal and electricity soared as well. With higher prices, the growth of utility usage went flat. Beginning in 1990, electricity growth dropped to zero. Utility companies went to their regulators to have prices adjusted upward and were surprised to find that the same regulators who had been so cooperative approving price decreases were no longer friendly at all. The DDM was no longer useful in predicting utility stock prices.

So, the dividend discount model is not that useful, because the future is too uncertain. But we already know that. The DDM is still of value, solving for g, the growth rate implied by the stock price. If the implied growth rate is suspiciously high, you might decide that you are assuming that a high return on equity may be true now, but mean reversion will spoil your longer-term euphoria.

1 Gordon, M.J. 1959. “Dividends, Earnings, and Stock Prices,” The Review of Economics and Statistics, Vol. 41, No. 2, pp. 99–105. I did my work from Gordon’s work at MIT in 1955.

This article originally appeared in the December/January 2018 issue of Morningstar magazine. To learn more about Morningstar magazine, please visit our corporate website.