The mid-1980s are popularly remembered as a time of excess, especially in the United States where the avarice of financiers was immortalised by Gordon Gecko’s “greed is good” line from the 1987 film Wall Street. Hollywood may script exaggerated villains and anti-heroes, but the drama that unfurled on Black Monday in October 1987, when the S&P 500 suffered its worst ever one-day fall, would have been difficult to overstate.
The 1987 crash came as a result of market participants sleep-walking into a situation of false security with semi-passive investing methodologies that came unstuck. Computerised stop-losses designed to get institutional investors out of losing positions became like the uncontrollable tools of The Sorcerer’s Apprentice, creating negative selling momentum that reached crisis point. There are some worrying parallels with today, given the amount of money in passive investment products, which would be required to sell out of positions in a downturn. Add to this the difficulties so-called smart-beta products could have, thanks to some quite illiquid holdings and complex collateralisation structures, and it is not difficult to imagine a panic scenario.
How much risk comes fromcurrent valuations?
Stretched valuations are often given as one of the factors that led to Black Monday, although by the standards of the present day, or the expensiveness of markets ahead of other notable crashes in history, the S&P 500 wasn’t that dear. Using the cyclically adjusted price/earnings ratio (CAPE) computed by Yale Professor Robert Shiller, the US market wasn’t horrendously overpriced in September 1987, despite a five-year bull market. The CAPE ratio was 17.7, only slightly above the 30-year rolling average at the time, which was 15. Contrast this with the dot-com boom, when CAPE hit 44 (against a long-run average of 16), and in the run-up to the global financial crisis of 2008, when CAPE once more breached 27, above a rolling 30-year average that, inflated by the previous boom of the late 1990s, stood at over 20.
It has been said market timers exist to make astrologers look better, and trying to use the absolute or relative levels of CAPE as an entry or exit signal would only have meant missing out on some of the best upside of the past 30 years. What we can say is that, on the whole, annualised returns have been higher subsequent to CAPE being low, and conversely returns have been lower following periods of high CAPE. This still isn’t helpful in deciding whether to be in or out of a market – as the benchmark for what is expensive is constantly changing. So instead of thinking about the market in terms of potential returns, a better starting point is in considering the risks.
Research by US risk company Covisum, which estimates probability-based market tail risk (a more robust technique than traditional Gaussian statistical analysis, which is based on the erroneous assumption of normally distributed asset returns), suggests that in the worst 0.5 per cent of months the S&P 500 will lose, on average, 16 to 17 per cent of its value. This is a one-in-200 probability but it doesn’t mean such a loss could only occur once every 200 months. Bad months tend to happen in clusters and, extrapolating an annual figure from their algorithms, Covisum suggests that in the worst years the average loss investors should expect from the S&P 500 is about 44 per cent.
Is this magnitude of loss more likely when the market is expensive? Well, looking at some of the most famous stress episodes of the past 30 years, we can see that over subsequent 12-month periods, prices have taken longer to recover when the market was at more elevated levels.
The S&P 500 may have seen a fifth of its value wiped out on Black Monday, but a year later on 19 October 1988 it was only 2.4 per cent lower than it had been on the Friday close before Black Monday. By contrast, a year after the peak of the tech bubble on 31 December 1999, the S&P 500 was more than 10 per cent lower. Working backwards, from the nadir of that savage bear market on 11 March 2003, the 12-month loss had been over 31 per cent (the peak-to-trough drawdown of the whole three year bear market was 45.5 per cent). The worst one-year fall for the S&P 500 cam at the trough of the last major bear market. From 9 March 2008 to 9 March 2009 the US index dropped 52 per cent.
So, anecdotally we can say the past points towards the worst sustained periods of loss coming from very high valuation multiples. This chimes with common sense – things have further to fall when they’re expensive – but it doesn’t follow that high prices are predictive indicators of an imminent downturn. We started looking at data from the beginning of 1987 and in more than four-fifths of months, CAPE has been above its rolling 30-year average. This shows a powerful trend for equities becoming more expensive. Historically it has always been an incredibly bad strategy to be out of the market for long periods or to bet against momentum, so any system that attempted to use valuations to time entries and exits would have been folly.
If we accept that expensive multiples are not a reason to be out of markets entirely, then a logical question to ask is whether risk can be quantified differently at higher valuations, so that portfolios can be adjusted to tweak levels of exposure. Looking at S&P 500 month-end data since 1987, Covisum found that in months when CAPE was above the rolling 30-year average, it was not conclusive whether we could model for an increased risk of the worst subsequent 12-month periods. So whether markets are cheap or expensive, we should just stick with the overall risk estimate. In any case with so many end-of-month observations showing the S&P 500 above its rolling average CAPE, more often than not the market has been relatively expensive by historic standards at any given point in the past 30 years.
