As a simple measure of this combination – rich valuations coupled with upward pressures on risk premiums and competing yields – it’s worth considering the current Shiller P/E near 24 (S&P 500 divided by the 10-year average of inflation-adjusted earnings) in the context of rising yields on competing securities. Both the Dow Jones Utility Average and the Dow Jones Corporate Bond Average are down more than 5% from their recent 26-week highs. The last time we saw this combination of weakness in interest-sensitive sectors with the Shiller P/E even above 18 was in September 2008, just before the market collapsed that year. We observed a similar deterioration following overvalued, overbought, overbullish syndromes in January 2000 (though be aware that it took several more months of top formation for the market to decline in earnest) and June-September 1987. This is certainly not the only concern that we have at present, but it illustrates our discomfort with speculative risk-taking here.
As a side note, we’ve seen some arguments disputing the relevance of the Shiller P/E, suggesting that the accounting treatment of writeoffs in recent decades has made this measure obsolete. This might be a more compelling view if other valuation measures such as S&P 500 price/revenue, price/dividend, price/book, and market capitalization to GDP did not all presently indicate exactly the same range of overvaluation as the Shiller P/E does. One of the more intellectually distressing arguments on this front suggested replacing S&P 500 earnings with NIPA (National Income and Product Accounts) profit figures in the calculation of the Shiller P/E. This is an apples-to-oranges calculation, as NIPA figures measure economy-wide profits in dollars and are neither restricted to the S&P 500 nor include the per-share adjustments that index earnings do. Still, one can see why the substitution is superficially attractive: the ratio of NIPA profits to Shiller earnings has surged to the highest level in history in recent years, mirroring similarly elevated profit margins. Unfortunately, even much less breathtaking elevations in NIPA profits / Shiller earnings in the past have been regularly followed by weak subsequent growth in NIPA profits, as profit margins retreat. Moreover, the substitution of NIPA profits into the Shiller calculation substantially weakens, rather than strengthens, its relationship with subsequent S&P 500 total returns.
In any event, we refer to the Shiller P/E mainly because it is an easily accessible shorthand measure that is simple to obtain and calculate. As detailed in Investment, Speculation, Valuation and Tinker Bell, there are numerous other approaches that have a roughly 90% correlation with subsequent 10-year market returns, while many popular approaches (such as the Fed Model) have very little relationship to subsequent returns at all.
On the subject of economy-wide measures, the chart below shows the market value of U.S. nonfinancial equities (from Federal Reserve Z.1 Flow of Funds data) divided by nominal GDP, and the subsequent 10-year S&P 500 annual total return (nominal, right scale, inverted). The present ratio of equity market value to GDP is consistent with an expected 10-year nominal S&P 500 total return of about 3%, which is about the same figure one obtains using the Shiller P/E and other historically reliable measures. Nearly all of this total return can be expected to come from dividend income, suggesting that the S&P 500 index will be little changed, a decade from now, from present levels.
All of this provides a good opportunity to reiterate that profits as a share of GDP remain nearly 70% above their historical norms; that cyclical variations in profits/GDP move closely and inversely with the combined total of government and household savings as a share of GDP; and that elevated profits/GDP are inversely correlated with subsequent growth in profits over the following 4-year period.
A long-term perspective on intermediate-term discipline
A few weeks ago, in The Road to Easy Street I noted that even with the additional exclusions that we’ve introduced in recent years, overvalued, overbought, overbullish conditions sufficient to warrant our strongest defensive outlook have emerged about 5% of the time since 1940. These hostile conditions have historically been associated with average market losses on the order of 40-50% on an annualized basis.
The chart below expands on those comments, showing the cumulative performance of the S&P 500 when our estimates of the market Sharpe ratio were zero or negative (about 37% of history), compared with periods when they were positive (about 63% of history). The Sharpe ratio measures the expected return of the market in excess of Treasury bill yields, per unit of volatility. I’ve plotted the two lines on log-scale to highlight equivalent percentage changes.
This is not a chart of any specific investment strategy - only a partitioning of the data into two classifications to clarify my general approach to the investment process. As I’ve often noted, the worst estimated Sharpe ratios are generally associated with either overvaluation coupled with deteriorating market internals, or the emergence of an overvalued, overbought, overbullish, rising-yield syndrome of conditions. A sequence of overvalued, overbought, overbullish, rising-yield conditions followed by deterioration in market internals is what I’ve termed a broken speculative peak and is what we presently observe. By contrast, the strongest Sharpe ratios tend to be associated with favorable (or at least reasonable) valuation coupled with a firming of market internals.
The advance in the red line at the very right of this chart offers some perspective on the past few years. The fact is that particularly since late-2011, the market has advanced despite conditions that have historically been associated with significant market losses. Frankly, we’re used to similar frustration during late-stage bull market advances – you’ll see a similar creep upward in the red line approaching the 2000 and 2007 market peaks before severe market losses ensued. Still, the recent period has been particularly uncomfortable because of QE-driven speculation. You'll also notice a few similar upward segments in 1969, 1972, 1981 and 1987, indicating that a defensive stance would have resulted in some period of missed returns.
Notably, the recent situation could have been helped by overriding our approach with rules amounting to versions of “Don’t fight the Fed”, but those same rules applied to past cycles and long-term history would have failed dramatically. Remember that the 50% market plunges in 2000-2002 and 2007-2009 both unfolded despite aggressive and continuous monetary easing (see Following the Fed to 50% Flops).
We continue to align our investment outlook with the expected Sharpe ratio that is associated with prevailing conditions at each point in time (see Aligning Market Exposure with the Expected Return/Risk Profile). Strong opportunities to accept market risk have emerged over the course of every market cycle in history (and no stress-testing concerns will interfere with those opportunities as they did in the recent, extraordinary cycle). In my view, the primary concern here is not the risk of missing long-term returns, but of losing intermediate-term discipline in conditions that have historically proved damaging. Even if the S&P 500 Index goes nowhere over the coming decade - as historically reliable measures of valuation suggest - it will probably go nowhere in an interesting and volatile way, providing better value and opportunities that are well-supported by historical evidence. The challenge will be to maintain discipline even when frustration begs investors to abandon it.
No doubt, one can dispense with my concerns about market risk here by observing that the market has advanced despite our increasingly negative return/risk estimates, allowing one to conclude that “this time is different,” and will continue to be so indefinitely. In my view, the more likely possibility is that investor faith in unlimited quantitative easing has supported a severe compression in equity risk premiums; that this faith is based largely on collective belief bordering on superstition – rather than on any meaningful historical relationship between the monetary base and the level of equity valuations; and that the extent of market overvaluation is obscured by profit margins that have been elevated to unprecedented but temporary heights by equally unprecedented but temporary deficits in government and household savings. If that case is true – and I believe it is – then the appropriate conclusion is not that this time is different, but merely that an increasingly weighty anvil has not yet dropped.
As I observed last week:
“Despite individual features that convinced investors in each instance that ‘this time is different,’ my perspective is that the truly breathtaking market losses in history share a single origin: the willingness of investors to forgo the need for a risk premium on securities that have always required one over time. Market crashes are largely synonymous with a spike in risk premiums from previously inadequate levels. Once the risk premium is beaten out of stocks, there is no way out, and nothing that can be done about it. Poor subsequent returns, market losses, and the associated destruction of financial security are already baked in the cake. This should have been the lesson gleaned from the period since 2000, but because it remains unlearned, I am convinced that it will also become the lesson of the coming decade.”
A few observations about economic activity here. One is that while the low level of initial claims for unemployment has been a bright spot, the simple fact is that initial claims are almost always depressed at major market peaks, which contributes to the optimism and euphoria at those highs. In the chart below, I’ve added arrows over the present instance not to imply that the market must be at a peak here, but simply to emphasize that the recent pattern of new claims for unemployment (red line, right scale, inverted) is not at all inconsistent with previous instances of maximum market risk.
More broadly, the best way to characterize the incoming economic data is that we’ve observed some modest improvement in recent months, but still around levels that have historically marked the borderline between economic expansion and recession. Indeed, the composite message from numerous regional Fed and ISM (Institute for Supply Management) surveys is not much different than what we observed just before the last recession. Overall then, we’ve observed a modest bounce over the past few months, but weak and not particularly meaningful from a historical perspective.
Meanwhile, the Federal Reserve’s attempt to support employment growth through quantitative easing has been much like doing a frantic rain dance over thin ice, in hopes of attracting more fish. The first question that should be asked is whether there is any demonstrable mechanism (even one that can be illustrated with a basic scatter plot in historical data) that would link the efforts to the desired outcome; the answer being no. The second question that should be asked is whether the continued attempt creates the risk of unintended consequences; the answer being yes.
So while it’s true that the 3-month average of monthly job creation has declined from 233,000 in February, to 172,000 in May, to 148,000 at present, this is more an indication of the ineffectiveness of quantitative easing than an argument for its continuation. We expect that the Fed will pursue a course of gradually fading out its efforts at quantitative easing, but in a way that aims to ease the potential disruption of the financial markets, where nearly all of the effect – primarily distortionary – has been centered. This really should not be a surprise, given that even the minutes from the January FOMC meeting indicated that “an ongoing evaluation of the efficacy, costs, and risks of asset purchases might well lead the Committee to taper or end its purchases before it judged that a substantial improvement in the outlook for the labor market had occurred.” To quote Dallas Fed President Richard Fisher, commenting last week about the impact of QE on the real economy – “I don’t believe it had any efficacy.”