Sunday 06/16/02 : Hotline Update
The Market Climate remains on a Warning condition, freshly deteriorated. Last week, the market lost support from breadth (advances versus declines) and leadership (new highs versus new lows), as internal action surrendered a great deal of the resilience that has marked recent months. That doesn't necessarily mean that conditions will worsen further. As usual, I have no forecast of market direction. But we need no forecast to know that the present, identifiable Climate is unfavorable. That might change next week, next month, or next year, but until it does, we will maintain a defensive position.
This insistence on maintaining a position in line with the prevailing Market Climate, while requiring no meaningful forecast of future market action, is what distinguishes us both from buy-and-hold investors and from market timers. We focus strictly on the present. The present is the only point in time where reality can be observed, and where action can be taken.
The prevailing Market Climate is the result of two observables: valuations, and market action. Currently both are negative. We know that historically, this combination has been associated with a very poor return/risk profile for the market, taking all such periods together on average. But we measure the Market Climate weekly. On a weekly basis, the average loss works out to a very small number - much smaller than typical weekly volatility. So beyond a small, unreliable, and statistically insignificant tendency for the market to decline over the coming week, we have and need no meaningful forecast of market direction. And yet, we have great comfort holding a fully hedged position. It takes a while to fully understand this approach. At its core, our strategy is focused on observable reality rather than hope, forecasts, or blind expectations of reward from market risk.
Of course, reality is always interpreted by its observer, and the same fact can be interpreted in different ways by two observers, depending on the clarity or baggage that they bring along. Probably the best example is last week's statement by Abby Joseph Cohen on CNBC that the S&P 500 is 20% undervalued. In order to understand why we observe reality and see stocks as overvalued, while she observes reality and sees them as undervalued, you have to understand that while reality is reality, the concept of value is subjective. When anybody talks about value (including us), they are implicitly assuming some particular long-term rate of return that they believe is reasonable.
Consider a bond that promises to pay $100 a decade from now. If an investor believes that bonds should be priced to deliver a 4% long-term return, the "fair value" for that bond is about $68. In contrast, if an investor believes that bonds should be priced to deliver a 12% long-term return, the "fair value" for that bond is about $32. Which view of "value" is correct? It depends on what you believe about other investors and their willingness to accept 4% versus 12% as a sustainable rate of return.
The same holds true in stocks. In stocks, the price of a stock is also the discounted value of the future cash flows it will throw off to investors, only there is an additional issue since the cash flows are growing and uncertain. Still, the problem is much simpler than widely believed, and the range of plausible outcomes is also much narrower than widely believed. As Warren Buffett noted last year, among the 100 leading stocks in the market, you can wager a tidy sum that only a tiny handful will achieve 15% earnings growth over a 20-year period.
Earnings for the S&P 500 as a whole are even better behaved, since earnings growth simply does not differ much from nominal GDP growth over the long-term. From year-to-year, of course, earnings vary a great deal, so trough-to-peak growth can be extremely high, and peak-to-trough growth can be horrifyingly negative. But regardless of whether you look over the past 10, 20, 50 or 100 years, peak-to-peak earnings growth for the S&P 500 has been strikingly well behaved, growing at a nominal rate of 6% annually with great consistency.
Given that fact, the question of "fair value" comes down to the model you use. Take the "dividend discount model" for example. This model says simply that the long-term rate of return on stocks is equal to the long-term growth rate of dividends, plus the dividend yield. (As long as you take dividends as per-share values, this model is valid even in the presence of stock repurchases). The real problem is that some people make frighteningly implausible assumptions about the "fair" long-term return that stocks should be priced to deliver.
For example, suppose you believe that stocks should be priced to deliver 2% more than long-term bonds over the long term, and long-term bonds are yielding 5%. Well, you're now looking for the level of stock prices at which stocks will be priced to deliver 7% long-term returns. If dividend growth is 6% annually, the dividend discount model will then tell you that the "fair" dividend yield is 1% (i.e. 7% total return = 6% growth + 1% yield). With dividend yields currently at 1.5% then you will conclude that the S&P 500 is undervalued here, and that "fair value" is 50% above current levels.
But let's test the robustness of this model. Suppose that bond yields rise by 1%. In that case, we need to get an 8% long-term return, with only 6% contributed by dividend growth. Suddenly, we need the dividend yield to be 2% instead of 1%, implying a 50% decline in "fair value" in response to a little 1% rise in interest rates. This result should bring the careless use of the dividend discount model into question when very low risk premiums are being assumed. Essentially, the Glassman and Hassett "Dow 35000" garbage is essentially an irresponsible contortion of dividend discount.
Now consider the so-called Fed Model, which holds that the prospective earnings yield on the S&P should be equal to the 10-year Treasury yield. As long as you're willing to forecast a wildly optimistic level of operating earnings for the coming year (Abby's specialty), you'll also be willing to call stocks undervalued.
Of course, operating earnings include not only claims of shareholders, but also interest owed to debtholders and taxes owed to the government. And even including good signals from two extreme readings (positive in '74, negative in '87), deviations from the Fed model have absolutely zero statistical relationship with subsequent returns. I fully believe that the Fed model lacks any usefulness as an investment management tool. In fact, the raw, unadjusted earnings yield on the S&P 500 is statistically more reliable. But go ahead and use it anyway if you enjoy confusion and disappointment.
One of the questions I've received lately is whether my view of stock valuations has changed due to the recent decline in the 10-year bond yield. The answer is yes, but so negligibly as to be irrelevant. When you form an assumption about the "fair" long-term return on stocks, you have to consider not only current bond yields, but also their historical tendency to change. A 1% increase in yield to maturity produces a decline of only about 7.8% in a 10-year bond, but a 1% increase in dividend yield would currently produce a decline of about 40% in stocks. Clearly, a change in bond yields does not, and should not, translate percent-for-percent to a change in long-term stock yields.
In my view, stocks should currently be priced to deliver long-term returns in the 9% area - a rate of return that reflects a relatively low level of interest rates, but a greater risk premium than typically assumed by perennial bulls. On the assumption of 6% long-term growth in fundamentals, a 9% long-term total return implies a 3% dividend yield. Currently the yield on stocks is half that level. No plausible decline in interest rates would justify a market dividend yield of 1.5% (compared to a historical average of 4%), a market price/book ratio of 5 (compared to a historical average of about 1.5) and a price/peak-earnings ratio of 20 (compared to a historical norm of 14 and average bear-market lows below 9).
Essentially, the total return on stocks can be partitioned into three pieces: 1) capital gains from long-term growth in fundamentals, 2) capital gains due to changes in valuation, plus 3) dividend income. Let's first assume that valuations are actually fair at current levels. What is the long-term return implied by current prices? Well, since the peak-to-peak growth in nominal S&P 500 earnings has been very well behaved at about 6% over the past 10, 20, 50 and 100 years, a constant price/peak-earnings multiple for the S&P 500 would also produce long-term capital gains of about 6% annually. Add in a 1.5% dividend yield, and stocks are priced to deliver about 7.5% over the long term, if earnings move back to peak levels and grow at 6% annually, and the price/peak-earnings ratio remains at its current level of about 20 into the indefinite future.
Of course, a price/peak-earnings multiple of 20 is the level seen at the 1929, 1972 and 1987 peaks. So we also have to be sensitive to the possibility that valuations will decline in the future. In that event, the contribution to returns from changes in valuation becomes negative. So a long-term return of 7.5% from the S&P 500 should be taken as somewhat optimistic. By comparison, the run from 1982 to 2000 took the price/peak earnings rato from 7 to 34, which provided a massive source of capital gains in addition to earnings growth itself. And while many people rave about a new era of productivity growth, I'm not quite convinced that productivity growth born of constant output and falling employment is a sustainable trend. Even if it were, it should be understood that among economists, an increase of even 0.5% in long-term productivity would be considered a fantastic improvement. So maybe we're off by half of a percent. In any event, I believe that stocks are currently priced to deliver a long-term rate of return much lower than investors have historically earned, and more importantly, much lower than they are likely to eventually demand.
If you believe that over the long-run, investors will demand a long-term return in the area of about 9%, then stocks are about 50% overvalued. But that is not a forecast in any meaningful sense. Overvaluation means only that stocks are currently priced to deliver an unsatisfactory long-term rate of return. It implies little about short-term returns.
The bottom line is simple. The term "fair value" always incorporates the assumptions of the analyst. Currently, stocks are priced to deliver a long-term return of something less than 7.5%. If you are a long-term investor, and think that 7.5% is a great return, and that future investors will agree that 7.5% is a great return far into the indefinite future, then go ahead and buy stocks. If 7.5% seems implausibly low as a sustainable long-term rate of return, the conclusion is that market risk does not have investment merit, and the only rationale for buying stocks on an unhedged basis is speculative merit. On that front, trend uniformity is unfavorable, so market risk currently lacks that merit as well.
With our approach indicating a lack of both investment merit and speculative merit to taking market risk, we remain fully (and comfortably) hedged. That position will change when the prevailing Market Climate changes.
www.hussman.com |
