Excerpt from the Hussman Funds' Weekly Market Comment (1/27/14):
With the S&P 500 just 3% below its all-time high, there’s very little change our views here. Last week’s mild retreat only looks something other than mild when viewed in the context of a late-stage parabolic advance that has not seen a single 2.5% weekly decline since June of 2012. At the typical historical frequency, we would have had eight of them.
The ratio of nonfinancial equity market capitalization to nominal GDP is presently about 120%, compared with a historical average prior to the late-1990’s bubble of just 55%. The comparison - about double the historical norm - is about the same if one uses the Wilshire 5000, which includes financials, and for Tobin’s Q (price to replacement cost of assets). The price/revenue multiple of the S&P 500 is presently 1.6, versus a pre-bubble norm of just 0.8. All of these measures have a correlation of about 90% with subsequent 10-year S&P 500 returns, even including recent bubbles and subsequent busts.
The reason we generally don’t include late-90’s bubble data in the calculation of historical norms (though one should always be explicit about it), is that the S&P 500 has achieved total returns of hardly more than 3% annually for almost 14 years since the 2000 peak as the result of those valuations, and yet the historical extremes remain only partially uncorrected. We currently estimate nominal total returns for the S&P 500 averaging just 2.7% annually over the coming decade - no more than the present yield on 10-year Treasury bonds (though stocks are likely to experience far greater volatility and interim losses). These estimates incorporate a broad variety of fundamentals, including properly normalized forward operating earnings (see Valuing the S&P 500 Using Forward Operating Earnings).
Disagreements between valuation methods can be quickly resolved by examining the data. In every case, measures suggesting stocks as fairly valued or undervalued are those that take current profit margins at face value, and assume that current earnings are a sufficient statistic for corporate profitability over the next five decades (which is roughly the duration over which discounting must occur – see Superstition Ain’t the Way).
It’s tempting to look at the 2000 valuation peak as if it’s some sort of goal to be achieved again, rather than a point that has already resulted in 14 years of 3% total returns with the likelihood of another 10 years of similar returns. One hastens to respond (and hastens for reasons below) that the success of the S&P 500 in reaching that pinnacle of valuation was followed by a decline that wiped out the entire total return of the S&P 500, in excess of Treasury bills, all the way back to May 1996, and yet another decline a few years later that wiped out the entire excess return all the way back to June 1995. Taking all of the instances for which data is available, even including the late-1990’s bubble, S&P 500 nominal total returns have averaged less than 1% annually in the 5-year period following Shiller multiples similar to the present. Taking a broader set of historically reliable measures into account, our actual estimates of S&P 500 total returns are negative at all horizons shorter than 7 years.
Of course, valuation isn’t a timing tool, and elevated valuations can persist for some time. Increasing our concern is the emergence in recent months of an extreme syndrome of overvalued, overbought, overbullish, rising-yield conditions exclusively seen at the 1929, 1972, 1987, 2000 and 2007 market peaks. We’ve seen even extreme variants of this syndrome emerge in February and May of last year without consequence, but we doubt that moving further along the edge of the cliff has diminished the likelihood of a fall.
Increasing our concern is also a very well-defined log-periodic bubble running from 2010 to the beginning of this year, which we estimate is now past its “finite-time singularity” (see A Textbook Pre-Crash Bubble). Increasing our concern is the shift to increased short-interval volatility just at the point of that singularity (which we estimate as January 13 – when 2-10 minute fluctuations began looking like the p-wave on a seismogram). Increasing our concern is the highest level of option “skewness” in history as option prices reflect extreme "tail risk" as they did prior to the 1987 crash (the 30-day average of skewness reached a record high on Friday - see Estimating the Risk of a Market Crash).
My overall impression of the global economy here couples disruptions in developing economies with year-over-year growth in U.S. real GDP, real final sales, payroll employment and household employment all close to the border that has historically delineated expansions from recessions. The Wall Street Journal reported last week that “The China Beige Book, a quarterly survey of Chinese businesses and banks, concluded that the country’s ‘credit transmission is broken.’” The same appears to be true in the U.S., where there are $2.4 trillion of excess reserves already in the U.S. banking system. Additional quantitative easing does little but ease a constraint that is not at all binding in the first place. The Fed is correct to conclude that it has done enough.
We view economic weakness as the dominant risk, but given that the correlation between leading measures (purchasing managers surveys, regional Fed surveys) and subsequent economic activity has largely collapsed in the past two years (see When Economic Data is Worse than Useless), it’s quite difficult to get a good near-term read on economic direction here. That said, we don’t expect upward pressure on short-term yields in the foreseeable future, and given the relatively wide spread between Treasury bond yields and Treasury bill rates, we continue to view Treasury debt constructively overall. In effect, deflationary risks, yield spreads, and the prospect of a flight-to-safety dominate the risk of positive economic surprises the effect of Fed tapering, in our view.
In contrast, corporate credit spreads appear dangerously narrow, and we view corporate debt with concern. Likewise, despite the popular belief that equity yields and bond yields should move closely together over time, this doesn’t hold well at all in the data. In fact, equity yields and bond yields were weakly or negatively correlated in historical data prior to 1970, and have been weakly or negatively correlated since the late-1990’s. The strongest positive correlation between the two was when 10-year yields were above 7%, reflecting the inflation-disinflation cycle from the 1970’s to the late-1990’s. Moreover, stocks are 50-year duration instruments from a present-value standpoint, while the benchmark 10-year Treasury has a duration of about 9 years. In short, there is little evidence that contained Treasury yields are a compelling argument for equities here.