stock graph down

Dow 36,000 – Fact or Fiction?

stock graph down

In a sequence of op-eds in the Wall Street Journal, an article in the Atlantic Monthly, and a book, James Glassman and Kevin Hassett of the American Enterprise Institute argue that the stock market is undervalued, possibly by a factor of three or more. Their case is predicated on the assumption that the expected return of equities should equal that of treasury bonds, or equivalently, that equity risk premium is undeserved.

As evidence in support of this view, they borrow from Jeremy Siegel's superb study of the history of U.S. capital markets, which shows that stocks have outperformed bonds over every single 20 year period from 1802 to the present time. Unfortunately, the logic that underlies their argument contains a number of flaws, and it behooves us to explore these flaws in some detail.

  • Messrs. Glassman and Hasset assert that stocks do not deserve a risk premium, as they have outperformed bonds over every single 20 year period.[/:Author:]
    Surely, stocks have outperformed bonds precisely because of their higher expected return – if the expected returns of stocks and bonds were identical, we would expect bonds to outperform stocks in half of all such periods. Furthermore, if stocks deserve no risk premium, we must conclude that investors have acted irrationally in demanding a risk premium for two centuries. Given our faith in the efficiency of capital markets, this appears to be an implausibly long lived inefficiency.

    In addition, the fact that stocks have outperformed bonds in the U.S. over every 20 year period in the past does not necessarily guarantee that they will outperform bonds over every 20 year period in the future. In Australia and Japan, for example, stocks have underperformed bonds for the past decade. Was this a random occurrence, or a failure on the part of investors to comprehend the long haul advantage of equities? Are stocks in these countries guaranteed to outperform bonds if we wait a full twenty years instead of stopping the timer at ten?

    The following analogy illustrates the Glassman/Hasset error in a more familiar setting. Consider a race between a Ferrari (stocks) and a Yugo (bonds). The Ferrari consistently wins, thanks to its more powerful engine (or higher expected return). Now replace the engine of the Ferrari with one from a Yugo (equalize their expected return) and rerun the race. Is the Ferrari still sure to win? I think not! This is the primary flaw in the Glassman/Hasset logic. They assume, incorrectly, that because stocks have consistently outperformed bonds when their expected return was higher, they will continue to do so even if the two expected returns are equalized
    The Glassman/Hasset logic implies that the expected return of corporate bonds ought to be lower than that of treasury bonds.[/:Author:]
    Look at the stocks vs. bonds question from the viewpoint of a corporate treasurer. If the expected return of equities ought to be same as that of treasury bonds, what ought the expected return of corporate bonds to be? Corporate bonds are riskier than treasury bonds – unlike the government, corporations cannot print money to pay off bondholders. Rational investors will demand some compensation in the form of higher expected return for the added risk they assume.

    It follows that the expected return of corporate bonds should be higher than that of treasury bonds, and a quick look at bond yields will show this to be true. If, as Messrs. Glassman and Hasset suggest, the appropriate expected return for stocks is the same as that of treasury bonds, the expected return of corporate bonds must be higher than that of stocks. But this logic is flawed. Equities are junior claimants to corporate earnings, and consequently, stocks are riskier than corporate bonds. It follows that the expected return of stocks must be higher than the expected return of corporate bonds, which in turn must exceed that of government bonds.
    • A fair value for the U.S. equity market is between 3 and 4 times its current level.[/:Author:]
      Spectacular if true, but unlikely to be so. In the long run, corporate earnings cannot grow faster than revenues, which in turn cannot grow faster than the economy. Earnings in excess of that needed to finance growth will be returned to shareholders via dividends or stock buybacks.

    From these elementary principles, with the further assumptions that profitability is both time invariant and high enough to finance growthwithout requiring infusions of new capital, it can be shown that the long run expected return of the stock market is Nominal GDP Growth + (ROE – Nominal GDP Growth)xB/P, where B/P is its book to price ratio (the reciprocal of its price to book ratio), and ROE is its prospective long run return on equity (earnings divided by book value).

    Currently, the book to price ratio of the S&P 500 is 0.2. Assume that the ROE of the S&P 500 averages 14% in the future, in between its long term average of 11% and its current value of 18%. Nominal GDP can be expected to grow at about 5% per annum – 3% real growth with 2% inflation. Substituting these figures into the equation gives an expected return of 6.8%, and a risk premium of 0.5%, a far cry from the 3% risk premium that Messrs. Glassman and Hassett posit. If, as they suggest, the risk premium collapses to 0, stock prices will rise at most by 50%. Furthermore, the fair value of the market is exquisitely sensitive to changes in interest rates. A 50 basis point increase in the risk premium can reduce fair value by 25%.

    For a different perspective, view stock prices through the eyes of American CEO's. If they thought their equity was enormously undervalued, would they pay for acquisitions with stock? Would entrepreneurs allow investment bankers to take their companies public for a third of their true worth? I believe that the answer to both questions is a resounding no.
    • A P/E of 100 equalizes the cash flows of stocks and bonds[/:Author:]
      I assume that this computation is based on the classic Gordon growth model in which dividends grow in perpetuity at a fixed rate. This model has an important limitation that one must be aware of, especially when the expected return is only slightly larger than the growth rate of dividends. A substantial portion of the net present value is derived from dividends paid is the very distant (i.e. many centuries) future. For example, if we discount future dividends at 5.5%, as do Messrs. Hasset and Glassman, only 37% of the value of the stock market can be attributed to dividends paid from now till the year 2100.

      I would be wary of making so rosy a prediction for so long a period. The old saying – the father wealthy, the son noble, the grandson a pauper – applies both to families and to economies! In a competitive economy and capital market, it is far more reasonable to assume that the expected return of equities will be a little (about 2%) less than the ROE of corporations, and that valuation levels will adapt accordingly. Both the historical return and valuation level of equities can be derived from first principles under this assumption . The answers accord well with history.

      In conclusion, given our faith in the general efficiency of markets, a very strong case must be made to support the assertion that markets are underpricing stocks by a factor of 3 or more. I do not believe that Messrs. Glassman and Hassett's present such a case. However, I hesitate to come to the opposite conclusion – that the market is substantially overvalued. To determine whether or not the market is overvalued requires knowledge of two quantities – the expected return embedded in the current price of equities, and the rate of return that investors require from their investments.

      If the second quantity is greater than the first, the market is overvalued. If the second quantity is less than the first, the market is undervalued. Unfortunately, it is not possible to precisely determine investors' expectations independently of market prices, and as a result it is not possible to determine if equities are overvalued or undervalued.

      However, one can determine the expected return of both equities and bonds and compare them to each other and to one's own required rate of return. Currently, the expected return of equities is about 7%, while that of bonds is about 6.5%. In real (i.e. inflation adjusted) terms, the expected return of stocks is 5%, a not insubstantial figure. Investors must decide for themselves if this real return and this expected return differential are sufficient to induce them to hold equities.

      1Siegel, Jeremy, ?The Shrinking Equity Premium?, The Journal of Portfolio Management, Winter 1999, pp. 10-17.

      2Philips, Thomas K., ?Why Do Valuation Ratios Forecast Long-Run Equity Returns?, The Journal of Portfolio Management, Spring 1999, pp. 39-44.