By John Hussman
November 26, 2012
In the day-to-day focus on the “fiscal cliff,” our own concern about a U.S. recession already in progress, and the inevitable flare-up of European banking and sovereign debt strains, it’s easy to overlook the primary reason that we are defensive here: stocks are overvalued, and market conditions have moved in a two-step sequence from overvalued, overbought, overbullish, rising yield conditions (and an army of other hostile indicator syndromes) to a breakdown in market internals and trend-following measures. Once in place, that sequence has generally produced very negative outcomes, on average. In that context, even impressive surges in advances versus declines (as we saw last week) have not mitigated those outcomes, on average, unless they occur after stocks have declined precipitously from their highs. Our estimates of prospective stock market return/risk, on a blended horizon from 2-weeks to 18-months, remains among the most negative that we’ve observed in a century of market data.
On the valuation front, Wall Street has been lulled into complacency by record profit margins born of extreme fiscal deficits and depressed savings rates. Profits as a share of GDP are presently about 70% of their historical norm, and profit margins have historically been highly sensitive to cyclical fluctuations. So the seemingly benign ratio of “price to forward operating earnings” is benign only because those forward operating earnings are far out of line with what could reasonably expected on a sustained long-term basis.
It’s helpful to examine valuations that are based on “fundamentals” that don’t fluctuate strongly in response to temporary ups and downs of the business cycle. The chart below compares historical price/dividend, price/revenue, price/book and Shiller P/E (S&P 500 divided by the 10-year average of inflation-adjusted earnings) to their respective historical norms prior to the late-1990’s market bubble - a reading of 1.0 means that valuations are at their pre-bubble norm.
Note that outside of the bubble-era, major bull market peaks tended to occur with valuations about 30-50% above the historical norm, while bear markets regularly brought valuations back to the historical norm and often well below that level. “Secular” lows generally occurred at valuations about half the historical norm. The 2000 market peak (which the S&P 500 remains below, more than 12 years later) reached valuation multiples more than three times the historical norm.
Presently, on the basis of smooth fundamentals such as revenues, book values, dividends and cyclically-adjusted earnings, the S&P 500 is somewhere between 40-70% above pre-bubble valuation norms, depending on the measure. That’s about the same point they reached at the beginning of the 1965-1982 secular bear period, as well as the 1987 peak. Stocks are far less overvalued than they were in the late-1990’s, but it is worth noting that nearly 14 years of poor market returns have resulted simply from the retreat from those bubble valuations to the current rich valuations. If presently rich valuations were to retreat again to undervalued levels that have accompanied the start of secular bull markets (see 1982 for example), stocks would produce yet another extended period of dismal returns. That prospect certainly isn’t the reason for our present defensiveness, but it’s worth understanding the dynamic that has produced the pattern of market returns we’ve observed over time.
The defining feature of dividends, revenues, book values and the 10-year average of inflation-adjusted earnings (the denominator of the Shiller P/E) is that they are smooth and insensitive to cyclical fluctuations in profit margins over the business cycle. In contrast, standard price/earnings ratios generally seem very reasonable when profit margins are elevated, and seem extreme when profit margins are depressed. Needless to say, that is no small risk for investors who are enamored with seemingly “reasonable” P/E ratios based on forward operating earnings (which assume that companies will indefinitely earn profit margins about 70% above historical norms).
While we prefer to explicitly model the stream of expected future cash flows in our own valuation work, these multiples can be converted into 10-year total return estimates for the S&P 500 using a fairly “model free” rule of thumb, by associating “fair” value with a 10% prospective return. If we write the normalized price-fundamental ratio as “NPF”, and assume that deviations are gradually corrected over a period of 10 years, we have:
Estimated prospective 10-year S&P 500 total return = 1.10/(NPF^.1) – 1
So for example, an NPF of 1.0 corresponds to a 10% 10-year prospective return. An NPF of 0.5, which we might see at the start of a secular bull market, would correspond to a 10-year prospective return estimate of 1.10/(0.5^.1)-1 = 17.9%.
As a more concrete example, with the S&P 500 price/dividend ratio presently about 43, versus a historical norm of 26, the NPF on dividends is about 43/26 = 1.65. That figure translates into a 10-year prospective return estimate of 1.10/(1.65^.1)-1 = 4.6%.
The chart below compares 10-year total return estimates using this rule-of-thumb with the actual subsequent 10-year total returns (nominal) achieved by the S&P 500. While explicitly modeling cash flows generally produces tighter results in our experience, these “model free” estimates have aligned well with subsequent outcomes. It should be evident that smooth fundamentals such as dividends, revenues, book values, and long-averaged earnings can provide a reasonable basis for long-term return expectations. As always, past relationships between fundamentals and subsequent market returns don't ensure the future reliability of these relationships.
It’s interesting to note both the broad correlation between the estimates and the subsequent returns, as well as the periods where they don’t match. In general, points where the actual 10-year return on the S&P 500 (SPX10YR_TR) shoots well beyond the projected return are points where the terminal valuation at the end of the 10-year period was well out of line with historical norms. Examine 1964 for example – the actual subsequent 10-year return significantly undershot the expected 10-year return. That outcome reflects the fact that the terminal valuation at the end of the 10-year period (1974) was deeply below the norm, so stocks lost more during that period than one would have expected. Similarly, examine 1990. In that case, the actual return substantially overshot the expected return. That outcome reflects the fact that the terminal valuation at the end of the 10-year period (2000) was dramatically above the norm. At present, the return of the S&P 500 over the past decade – though below average – has actually overshot what would have been expected in 2002. This reflects the fact that valuations today are still well above their norms. Unless we assume that valuations will remain rich forever, this doesn’t portend well for returns going forward.
Emphatically, the rule-of-thumb cannot be accurately used with fundamentals like “forward operating earnings” that are sensitive to expansions and recessions. When the denominator of your valuation multiple is affected by cyclical economic fluctuations, the multiple often says more about where you are in the business cycle than where you are in terms of valuation. Likewise, the proper historical “norm” for a valuation multiple is the level that is itself associated with “normal” subsequent returns. We sometimes see analysts using valuation “norms” where nearly half of the data represents bubble valuations since the mid-1990’s. We are now nearly 14-years into a period where the S&P 500 has underperformed Treasury bills. It is lunacy to consider the valuations that produced this outcome as the “norm.”
To illustrate this point, notice that while Shiller P/Es below 12 have historically been associated with subsequent returns averaging about 15% annually over the following decade, Shiller P/Es about 22 or higher have been followed by nominal returns averaging only about 3.6% annually over the next decade, on average (the present multiple is 21.2). With little respite, the Shiller P/E has been above 22 since 1995, and the average Shiller P/E since that time has been over 27. To load that stretch into the calculation of the “normal level” destroys the whole concept of a norm: the valuation norm should be the level that is reasonably associated with “normal” subsequent market returns.
We remain convinced that stocks are richly valued here. A fairly run-of-the-mill normalization of valuations in the course of the present market cycle would imply bear market losses of about one-third of the market’s value, without even establishing significant undervaluation. Then again, there’s no assurance that valuations will normalize, or that stocks will experience a bear market here. Maybe Wall Street is correct that profit margins will remain forever elevated and The New Global Economy™ will never again witness “normal” valuations on these measures at all. There’s no shortage of analysts who effectively embrace that view by focusing only on forward-operating earnings.
Still, it’s worth a moment’s consideration that “secular” lows (which we typically observe every 30-35 years, most recently in 1982, and that serve as launching pads for long-term market advances) have usually been associated with declines in normalized price-fundamental ratios to about half of their historical norms. Such an event even 15 years from today would be associated with an estimated annual total return of just 2.7% for the S&P 500 between now and then. Such an event a decade from now would be associated with a negative expected total return for the S&P 500 in the interim. And while it’s not our expectation, such an event in the present market cycle would make “S&P 500” not just an index, but a price target.
[Geek’s Note: The more general version of this rule of thumb is: Estimated prospective 10-year S&P 500 total return = 1.10/([NPF_current/NPF_future]^(1/T)) – 1. So moving from an NPF of 1.4 to an NPF of 0.5 over a period of 15 years would produce an estimated prospective return of 1.10/([1.4/.5]^(1/15))-1 = 2.7%].
We continue to believe that the U.S. economy joined a global recession during the third quarter of this year. There is a fairly regular cycle of economic “surprises” in U.S. data that tends to run about 44 weeks – a figure that appears related to the tendency of economic forecasters to use standard “lookback” horizons to determine economic trends (see The Data Generating Process). As a result, the recession thesis has had to swim upstream, so to speak, during the positive portion of that cycle, which appeared to run through early November. Our expectation is that economic surprises are likely to be heavily to the downside as we move through the next 4-5 months.
While we don’t have many companions in our recession view, aside from ECRI, the coincident data has quietly become a companion of this view already. Industrial production actually peaked a few months ago, real sales have weakened, personal income has weakened, and despite some divergences among individual reports here and there, the new order, order backlog, and employment components of numerous regional and national Fed and Purchasing Managers surveys have also turned decidedly lower (see Stream of Anecdotes).
The chart below presents a number of “diffusion” indices from the Industrial Production and Chicago Fed National Activity Index (CFNAI) reports. Diffusion indices are volatile, but have the benefit of being very sensitive to turns in activity. Essentially, a diffusion index counts the number of survey respondents that report “better” activity, minus those that report “worse” activity, plus half the number that report “unchanged” activity. The Industrial Production report asks respondents to compare their activity with that of 3-months earlier and 6-months earlier, while the CFNAI survey asks about more general improvement or deterioration. Notice that these diffusion indices have collapsed in tandem over the past few months (the chart presents standardized values: mean zero, unit variance, for comparability). While concerted deterioration toward -1 (one standard deviation below the mean) has not always resulted in recession, the recent weakening is consistent the downturn we’re seeing in broader measures of economic activity (outside of employment, which I suspect is not far behind).
Consistent with this trend, Lance Roberts of StreetTalkLive observes that “This past week the monthly release of the Leading Economic Indicators showed that the leading-to-lagging indicator ratio dropped to 89.5 which matches the lowest level in more than 2 1/2 years. Historically when the leading-to-lagging ratio has fallen below 91 the economy was either in, or about to be in, a recession.”
While our own economic concerns are focused primarily on the present cycle (where our view is that the economy has already entered a recession), some much longer-term concerns about the economy have recently emerged, and are worth discussing. Last week, GMO head and famed value investor Jeremy Grantham published a piece about long-term economic prospects that CNBC’s Josh Brown tweeted “makes Hussman sound like Mary f***ing Poppins”
Grantham argued that “The U.S. GDP growth rate that we have become accustomed to for over a hundred years – in excess of 3% a year – is not just hiding behind temporary setbacks. It is gone forever. Yet most business people (and the Fed) assume that economic growth will recover to its old rates… Going forward, GDP growth (conventionally measured) for the U.S. is likely to be about only 1.4% a year, and adjusted growth about 0.9%...The bottom line for U.S. real growth, according to our forecast, is 0.9% a year through 2030, decreasing to 0.4% from 2030 to 2050. This is all done presuming no unexpected disasters, but also no heroics, just normal “muddling through.” (h/t Business Insider).
These estimates reflect a fairly uncontroversial observation that U.S. population growth is presently much slower than historical rates, and appears likely to average less than 0.5% annually in the next few decades, compared with a rate of 1-1.5% since 1950. Similarly, total hours worked have been gradually declining over time, even ignoring the recent recession. The key additional observation is that annual productivity growth per worker has actually been declining over time, from a peak of about 2.5% in the mid-1900’s, falling to 1.8% by 2000 and on pace to decline further, hindered by a persistent slowdown in net capital formation, which has dramatically worsened in recent years.
Grantham is actually drawing here from a paper by Northwestern economist Robert Gordon, who recently published a working paper titled Is US Economic Growth Over? Faltering Innovation Confronts the Six Headwinds. That paper included the following provocative chart based on actual data from the UK and the US, along with Gordon’s projection for future growth.
We can’t dismiss these concerns outright, particularly since some of the demographic inputs are more or less baked in the cake. On the economic front, the real question is what the prospects for future U.S. productivity growth and capital accumulation are likely to be. To the extent that we continue to follow monetary policies that discourage saving, misallocate capital, promote speculative bubbles, and effectively trade away long-term growth for short-term stability, Gordon and Grantham may very well be correct. But in my view, that outcome is neither necessary nor assured.
The Nobel economist Robert Lucas once calculated that the gain in economic welfare from an increase of just a fraction of a percent in long-term economic growth would exceed the benefit of entirely eliminating business cycle fluctuations. Unfortunately, Ben Bernanke is no Robert Lucas, nor was Alan Greenspan. For the past quarter-century, U.S. monetary policy has elevated bubble creation and short-run economic stability over policies that might have otherwise encouraged the accumulation of productive capital. The promise of Wall Street is that – absent constant monetary distortions – it has the potential to effectively allocate capital toward its most productive uses. Unfortunately, that promise has been derailed by constant Federal Reserve interventions. My impression is that what Gordon and Grantham see as a precipitous decline in per-capita productivity and growth over the past quarter century is actually the product of distorted capital markets and misallocation of capital, thanks to a pernicious duo of monetary interventionists that society would have been better off without.
So, as the appointed Mary Poppins in this debate, I’m willing to take the optimists side – that U.S. economic growth is not over, just injured by absurdly misguided policies. There’s no question that financial markets remain distorted here, and that a mountain of unserviceable but unrestructured debt continues to stand in the way of a sustained recovery. But Robert Lucas was correct – there is a very steep cost to economic welfare from losing even a fraction of a percent in long-term growth. It is worth a great deal of short-term discomfort to restore those long-term growth prospects.
The problem is that the transition from a distorted economy to one that properly allocates capital will not be a smooth one. In order to take the optimists side, we have to be willing to accept the prospect that bad debt, insolvent institutions, distorted valuations, and unworkable monetary arrangements will have to be unwound globally. If the capital markets are allowed to operate without distortion, and economic imbalances are allowed to clear, I have no doubt that the U.S. has enough capacity for saving investment and innovation to produce adequate economic growth in the future. None of that will be painless, but there will be benefits to that adjustment.
From an investment perspective, I think Gordon and Grantham’s arguments are worth taking seriously, but would not dramatically change how we approach the financial markets even if they are correct. Undoubtedly, a shortfall of 1.5% or 2% from long-term growth expectations would mean that even our own estimates of long-term market returns might be optimistic. But note that even over a handful of market cycles, valuations tend to fluctuate enough to shift our estimates of prospective 10-year market returns by 6-10% and sometimes more. Put simply, prospective returns vary so greatly over the course of one or two market cycles that even a moderate change in long-term economic growth would not dramatically alter our strategic investment decisions.
That said, long-term concerns about economic growth like this could very well be reinforced and popularized at the depths of the next bear market and recession, whenever those occur. In that event, the “growth is over” view could easily contribute to the depressed valuations and “death of equities” mood that gives rise to a new secular bull market – though likely from much lower price levels than we observe today.
The foregoing comments represent the general investment analysis and economic views of the Advisor, and are provided solely for the purpose of information, instruction and discourse. Only comments in the Fund Notes section relate specifically to the Hussman Funds and the investment positions of the Funds.
Past performance does not ensure future results, and there is no assurance that the Hussman Funds will achieve their investment objectives. An investor's shares, when redeemed, may be worth more or less than their original cost. Current performance may be higher or lower than the performance quoted above. More current performance data through the most recent month-end is available at www.hussmanfunds.com. Investors should consider the investment objectives, risks, and charges and expenses of the Funds carefully before investing. For this and other information, please obtain a Prospectus and read it carefully.
(c) Hussman Funds