The S&P 500 set a marginal new high on Friday, in the context of a broad rollover in momentum thus far this year that we view as likely – though of course not certain – to represent a broad cyclical peak of the sort that we observed in 2000 and 2007, as distinct from spike-peaks like 1987. Valuation measures remain extreme, with the market capitalization of nonfinancial stocks pushing 130% of GDP (relative to a pre-bubble norm of about 55%), the S&P 500 price/revenue ratio at 1.7, versus a pre-bubble norm of 0.8, and the Shiller P/E near 26 – which while lower than the 2000 extreme, exceeds every pre-bubble observation except for a few months approaching the 1929 peak. We presently estimate 10-year nominal total returns for the S&P 500 Index averaging just 2.3% annually, with zero or negative total returns on every horizon shorter than about 7 years.
A side note about valuations and profit margins – my concern about record profit margins here is emphatically not centered on what profit margins may do over the next few quarters or years. The relationship between cyclical movements in earnings and stock prices is simply not very strong. Rather, as I noted in The Coming Retreat in Corporate Earnings, “my present concern is much more secular in nature. It can be expressed very simply: investors are taking current earnings at face value, as if they are representative of long-term flows, at a time when current earnings are more unrepresentative of those flows than at any time in history. The problem is not simply that earnings are likely to retreat deeply over the next few years. Rather, the problem is that investors have embedded the assumption of permanently elevated profit margins into stock prices, leaving the market about 80-100% above levels that would provide investors with historically adequate long-term returns.”
In other words, we should not be concerned about extremely elevated profit margins because earnings are likely to weaken and stock prices might follow over the next couple of years (although that may very well occur). We should be concerned because investors are pricing stocks as a multiple of current earnings. They are implicitly using current earnings as if they are representative and proportional to theentire stream of future earnings going out over the next five decades or more. That’s what it means to use a valuation multiple. It means that you take some fundamental as a sufficient statistic for the stream of cash flows that the security will deliver into the hands of investors for decades to come. If you think you know what wage rates, competitive pressures, interest rates and tax policy will be 10, 20, 30, 40 and 50 years from now, and that the present situation is representative and permanent, good luck with that. Otherwise, investors should recognize that because of the variability of profit margins over the long-term, valuation measures that adjust for variations in profit margins have been dramatically more reliable than unadjusted measures over time (see Margins, Multiples, and the Iron Law of Valuation)
Low volatility and suppressed short-term interest rates are a breeding ground for yield-seeking speculation. This reach for yield has now driven junk bond yields to about 5%, which is interesting given that yields are now near or below typical historical default rates. Meanwhile, the majority of new debt issuance today is taking the form of leveraged loans to already highly-indebted borrowers, with “covenant lite” features that provide little recourse in the event of default. This is the sort of behavior that should wake investors up like a triple Espresso. It doesn’t because they have been conditioned to focus on yield alone, without considering the minimal amount of capital loss that would wipe that yield out. This is not a new dynamic, and precisely because it is not a new dynamic, one can always find solace from the same Broadway kick-line of dancing clowns that reassured investors that credit was sound, subprime was contained, and stocks were still cheap in 2000 and 2007.
Meanwhile, overbought measures also remain extreme, as investors have become conditioned to a perpetually diagonal advance like those that brought stocks to similarly overvalued, overbought, overbullish pinnacles throughout history (see The Journeys of Sisyphus). As I’ve often noted, there are countless ways to operationalize the concept of “overvalued, overbought, overbullish” depending on the severity of the condition one wishes to capture. Conditions that capture less severe market peaks also tend to include various false signals, while more restrictive conditions limit the set to the worst pre-crash peaks in history, but will tend to miss cyclical peaks that were less extreme). The following set of criteria falls into the more restrictive category, and limits the set to the 1929, 1972, 1987 peaks, three tech-bubble instances (the peak before the 1998 Asian crisis, a pre-correction peak in 1999, and the 2000 top), the 2007 peak, and of course, a solid band of warnings today. From our perspective, this time is different only in the extended period over which these and similar conditions have been sustained in recent quarters without consequence.