It will come as no surprise that we continue to anticipate poor 10-year total returns for the S&P 500 over the coming decade. Our present estimate is about 3.3% annually, which includes dividends. That is about 1% less than the 10-year total return that we estimated just a few months ago, but this should make sense: historically, to the extent that the S&P 500 appreciates at an annualized rate of more than about 6% (which is about the long-term growth rate of revenues, nominal GDP and other smooth fundamentals), the expected future total return for the index declines as the market advances. Just like bonds. It is easily understood, but often ignored, that a short-term market advance of 10% - leaving fundamentals relatively unchanged in the interim - cuts about 1% annually off of subsequent 10-year market returns. This can be demonstrated both algebraically and in historical data.

The chart below plots our projections for 10-year annual total returns on the S&P 500 (standard methodology) versus actual subsequent 10-year total returns.

You may notice that unlike the charts we regularly present, which display post-war data, I've included data back to 1928 in the chart above in order to provide as much historical perspective as possible. While the comments below discuss the potential for a further extension of the recent speculative advance, a quick tour of historical market valuations provides some very instructive context.

Looking at 1929, it's clear that although the projected 10-year return (blue) for the S&P 500 actually dropped below zero prior to the 1929 crash, the stock market performed even worse than expected over the following decade, with an actual loss of about -5% annually (which compounds to a 10-year total loss of 40%, even including the fairly high dividend yields that were earned following the initial decline).

As the stock market plunged following the 1929 peak, the expected 10-year total return spiked higher. Once the S&P 500 had lost half of its value, the projected 10-year total return finally moved above 10% annually. Unfortunately, that was not the end of the decline, but just the beginning. By the time the S&P 500 hit its trough in 1932, it was less than one-third the level it was at the point that the projected total return reached 10%, and was more than 80% off of the 1929 peak. The S&P 500 followed the 1932 bottom with annual returns averaging nearly 14% annually over the following decade, but it would actually take until about 1950 and beyond for the market to fully realize the undervaluation that was present at the 1932 low.

By the mid-1960's, stocks were again priced to achieve poor long-term returns. Again, however, the market overshot over the following decade, achieving much poorer returns over the following decade than would have been predicted by valuations alone. Since the losses were excessive on a valuation basis, you'll note that by the mid-1970's, the expected 10-year return for the S&P 500 had increased to more than 16% - a return that was, in fact, achieved despite enormous inflation pressures over the next few years.

Now consider the effect of a market bubble. If you skip forward to 1990, you'll see that the actual total return over the following decade was much stronger than one would have anticipated on the basis of valuations at the time. That overshoot of valuation was not retained by investors, however, because it led to a situation where stocks were priced to predicably achieve negative total returns over the decade from 2000 to 2010, which is again precisely what we observed.

At the 2009 low, the projected 10-year total return for the S&P 500 briefly spiked above 10%. In the context of economic conditions that had no post-war counterpart, the issue was whether that projected level of return was sufficient to accept a materially invested position in stocks. History had not been kind to that assessment following lesser credit crises, and certainly not during the Depression. But in hindsight, that brief foray above 10% projected returns set the 2009 market low.

The subsequent advance in the market has compressed the majority of those prospective 10-year returns into much shorter time frame, leaving little prospect for significant additional returns in the next several years - at least, not additional returns that can be expected to be retained by investors.

For a slightly longer-term view, the chart below presents projected 15-year total returns for the S&P 500 Index, again using our standard methodology. At present, we project the 15-year total return for the S&P 500 at about 5% annually.

We already know that the negative expecation for 10-year total returns in 2000 turned out to be accurate. Above, you can see that the 15-year projection in 2000 implied roughly zero total returns out to 2015. Notice how this works - a 0% total return over 15 years, given a depressed dividend yield near 2%, implies an annual loss of about -2% on the basis of price alone, which compounds to a roughly 26% overall price loss. We're now 11 years into that 15-year horizon. Given that the S&P 500 peaked above 1550 in 2000, the implied projection for the S&P 500 2015 works out to about 1150. From present levels, that would imply a slightly negative annual loss averaging -2.9% annually. Adding dividends, this works out to the expectation of a slightly negative total return for the S&P 500 over the coming 4 years. While total returns over 4-year horizons tend to be less predictable than over 7-15 year horizons, the overall implication strikes us as just about right. It's not that stocks can't advance, nor that they can't plunge. It's just that the recent advance, in our view, has left stocks with little prospect for durable returns for several years to come.

Learning the right lesson

So market returns in excess of the growth of smooth fundamentals are simply displaced from the future. For buy-and-hold investors, this fact matters a great deal for the profile of returns they can expect to achieve over time. Buy-and-hold investors faced predictably negative 10-year total returns in 2000, and the prospect of dismal total returns from the 2007 valuation peak. They should certainly revel in their recent returns as measured from the 2009 low, but should also brace for a profile of future long-term returns that has again turned against them to an extent that has historically been distressing.

For investors who can vary their market exposure in response to changes in expected returns, the variation in expected returns constantly creates new opportunities over time, some having greater risk than others, but forgiving of risk-management provided that steep losses are avoided. We've outperformed the S&P 500 over every complete market cycle (bull and bear market combined), with smaller periodic losses, since the inception of the Strategic Growth Fund, and we remain ahead of the S&P 500 since its 2007 peak. Still, my concerns about the "out of sample" economic strains of 2009 resulted in a missed opportunity to capture the recovery that followed market losses that we largely avoided.

With respect to the present market cycle measured from the 2009 low, it's clear that the some bullish observers would like to call "game over" here while the market is strong. But this isn't how market cycles work. As the song goes, "You don't count your money while you're sittin' at the table." As I've noted before, many speculative factors can drive short-term fluctuations, but they don't change the algebra of long-term returns.

A variety of pure trend-following models have performed reasonably well in the window since March 2009. Still, I am convinced that ignoring valuations, economic factors and other evidence in preference for pure trend-following would be the wrong lesson to learn from this period. As I've noted many times, these alternatives are easy to test - if they performed better historically that the methods we use, they would be the methods we use. Likewise, I am suspicious that the proper lesson from 2009 is that we should have ignored everything but post-war data - even though that approach would have been beneficial, in hindsight.

In response to the "two data sets" challenge we faced in 2009, we've expanded the set of Market Climates we identify, which better use the evidence from multiple data sets and allow more "shades of grey." Taking a weighted-average of post-war and Depression era implications, as we did in 2009, was inadequate. The ensemble methods that we've implemented allow us to distinguish "risk" (the spread of individual outcomes from the average) from "uncertainty" (the possibility that the average itself is unknown) in a more satisfactory way. So there have certainly been lessons to learn. But it would be a very misguided "lesson" to attempt to correct that missed opportunity in 2009 by suddenly abandoning our long-standing and repeatedly validated practice of avoiding risk in overvalued, overbought, overbullish, rising-interest rate environments, which is what we have at present.

That said, I do expect that we'll take on some moderate, periodic exposure to risk before the bull portion of the present cycle is complete - most likely upon clearing some element of the present overvalued, overbought, overbullish, rising-yields syndrome, provided that the market internals remain fairly intact.

"Longer than you can stay solvent"

Two aphorisms regularly become popular during speculative periods in the market. One is the statement by John Maynard Keynes that "the market can stay irrational longer than you can remain solvent." The other is Warren Buffett's remark that "a pack of lemmings looks like a bunch of individualists compared with Wall Street once it gets a concept in its teeth."

It is widely understood that markets can experience extended periods of irrational speculation and valuation bubbles. This becomes problematic when people begin to rely on the irrationality of others as a justification for continuing to speculate, because at that point, they require the assistance of increasingly "greater fools" in order to sustain the advance. Keynes was quite right in the sense that one should never maintain a leveraged position against the market, because leveraged losses certainly do threaten solvency. But investors are not forced to accept risk just because other investors are speculating - there is no great risk in positions that accept no great risk. Moreover, investors cannot safely ignore valuations, because once valuations become rich, the returns from continuing to speculate are not easily retained even if they emerge for a while.

If actual returns did not periodically overshoot the returns that are warranted by fundamentals, we would never observe the extended periods of predictably excellent or dismal market returns that have historically corrected those overshoots. Presently, the probable outcome over the coming years leans moderately toward the "dismal" side, but nowhere as negative what we observed in 1929, 2000 or even 2007.

We have to be realistic that projected returns were actually driven to negative levels at the peaks of 1929 and 2000, so we can't rule out a further price advance that would compress projected long-term market returns to even lower levels than we observe here. Still, both of those valuation extremes were produced by multi-year periods of rapid, low-volatility economic growth, coupled with the introduction of revolutionary new technologies. A Federal Reserve policy that amounts primarily to rhetoric seems to be an awfully thin substitute at present.

Even here, we would expect some opportunity to accept market risk at the point where we clear some component of the present overvalued, overbought, overbullish, rising yields syndrome. So we have the flexibility to deal constructively with yet another bubble if it emerges, but we still expect to hedge against market fluctuations when we observe demonstrably unfavorable sets of market conditions. We face a particularly hostile syndrome now.

Normalized earnings, peak-earnings, and discounted cash flows

As a reminder of how we approach market valuation, we strongly believe that securities are a claim to a stream of future cash flows that can actually be expected to be delivered to investors over time. As a result, we have little sympathy (and history demonstrates little sympathy) for the popular but misguided practice of applying arbitrary valuation multiples to forward analyst estimates of earnings. Generally, these "forward earnings" estimates fail to normalize for fluctuations in profit margins, return on equity, and other factors that have historically driven short-term earnings temporarily above and below levels that that would have a stable, proportional relationship with the present value of subsequent cash flows. Forward operating earnings estimates are more volatile and more influenced by recent short-term behavior than can properly be used as a basis for valuation, and the resulting earnings "misses" can be particularly extreme at turning points.

In the graph below, you'll notice that the prior peak for S&P 500 trailing net earnings has often been a reasonable "rule of thumb" estimate of normalized earnings, but in recent years, temporary spikes in profit margins have periodically driven peak earnings briefly above properly normalized levels. For that reason, as I wrote several years ago, prior peak earnings have become increasingly unreliable. This is particularly true given the actual destruction of book value and revenue in recent years. It's certainly possible to debate the precise level of normalized earnings here, but somewhere in the $70-$75 range, which is where we are at present, is roughly accurate on a trailing net basis. Our estimates also assume continued future long-term growth of slightly more than 6% annually, as reflected by the red channels.

Importantly, since our normalized figure tends to run with earnings peaks rather than earnings troughs, the corresponding multiple applicable to these earnings has historically been less than 14 (and was actually closer to 12 in pre-bubble data, which was typically associated with long-term total returns near 10% annually). Since "forward operating" earnings are typically about 20% higher than trailing net, the resulting historical P/E "norms" should also be adjusted accordingly (which analysts rarely do). None of this is to say that the earnings peak during the current economic cycle has to be limited to the present level of normalized earnings - just that more elevated earnings would not be an appropriate basis on which to compute the long-term value of stocks.

To continue reading, click here.

 

(c) Hussman Funds

www.hussmanfunds.com