One reason to think that the relentless outperformance of the US stock market since the Global Financial Crisis (GFC) will not continue for another decade is its valuation, which has become comparatively stretched over the past year. Admittedly, valuations have not tended to be useful for forecasting the returns from equities over short horizons. But they have had significant predictive power over longer periods, like ten years. This will probably remain the case, notwithstanding a secular drop in real interest rates.
- One reason to think that the relentless outperformance of the US stock market since the Global Financial Crisis (GFC) will not continue for another decade is its valuation, which has become comparatively stretched over the past year. Admittedly, valuations have not tended to be useful for forecasting the returns from equities over short horizons. But they have had significant predictive power over longer periods, like ten years. This will probably remain the case, notwithstanding a secular drop in real interest rates.
- The return from a stock market index is approximately equal to its dividend yield plus the change in index earnings per share (EPS) plus the change in its price/earnings (P/E) ratio. So all else equal, if the P/E ratio is higher than normal, the return will be less than normal, on the proviso that the P/E ratio falls back.
- Valuations are like pieces of elastic – they can stretch a long way before they snap. So a higher-than-normal P/E ratio isn’t bound to fall back to normal any time soon, unless it is at breaking point (as it was in the US in 2000). Nonetheless, a P/E ratio ought to revert eventually to normal, assuming fluctuations in valuations wash out over time. While it is impossible to pinpoint when this will happen, an assumption that the P/E ratio will be back to normal in a decade’s time is arguably as plausible as any other.
- The superior predictive power of valuations over longer time horizons like this can be seen in Chart 1. The Chart includes data points for the period 1881 to 2009. The blue markers are plots of year-end levels of Shiller’s cyclically-adjusted P/E ratio (CAPE) for the S&P Composite/500 against average annual returns from the index in the next two years. The grey markers are plots of the year-end levels of his CAPE against average annual returns from the index in the next ten years. The fit in the second case is far better.
- It is often assumed that the normal P/E ratio is its average over a long period. On this logic, the prognosis for the S&P 500 for the next decade is very poor. After all, Shiller’s CAPE is currently close to 30, which compares to a (harmonic) average since 1881 of only a little above 14. If the CAPE reverted to this average over the next ten years, the annual average return from the index during this period would probably be negative, based on realistic assumptions about earnings growth and the dividend pay-out ratio.
- In our view, though, the normal CAPE is higher now than its long-run average. (See Chart 2.) This is because a decline in real interest rates has cut investors’ required real return from equities. We tentatively put the new-normal CAPE at 25, or an earnings yield of 4%. This compares to an average earnings yield since 1881 – the reciprocal of the average CAPE since then – of about 7%. So we assume that the drag on the return from the S&P 500 in the next decade from a drop in its valuation will actually be fairly small.
Chart 1: CAPE vs. Average Annual Returns (1881-2009)
Chart 2: CAPE: Actual, Average & New Normal?
Sources: Refintiv, Shiller, CE
Sources: Refinitiv, Shiller, CE
- If we are right, the real return from the S&P 500 in the next ten years will probably be positive. However, it probably won’t differ much from our estimate of the new-normal cyclically-adjusted earnings yield, which is only 4%. And it might be less than returns from other stock markets, whose valuations are lower.
- The logic that lies behind the relationship between the valuations of, and future returns from, equities in the US can be applied elsewhere. Admittedly, the evidence isn’t as comprehensive as it is in Chart 1, because the data don’t stretch back as far in time for other countries. But the more limited time series that we do have for them typically tell a similar story.
- Chart 3, for example, shows correlations for the period 1983 to 2009 between month-end levels of the CAPE and average annual returns in the next ten years for a variety of Datastream total market indices of other major developed economies. In all cases – not just the US – the correlations are significantly negative, which means that low valuations at the outset have tended to produce higher returns (and vice versa).
- The level of the CAPE has usually been higher in the US than elsewhere, except for Japan, where it was higher during its bubble in the 1980s and for a while after it burst. This partly reflects the compositions of the indices – the US has a larger-than-average weighting of firms in the information technology sector, for example, whose shares tend to trade on higher-than-average price/earnings multiples.
- Nonetheless, the gaps in CAPEs are currently unusually large in general, suggesting that equities in the US are comparatively expensive. This can be seen in Chart 4, where average (1983 – 2019) and current CAPEs for the countries in Chart 3 are shown by the dark blue and dark grey bars, respectively.
- A key reason why the gaps are so large now is the denominators of the CAPEs. These are average inflation-adjusted EPS over the prior decade. Since the GFC, EPS have grown strongly in the US, but generally not elsewhere. As a result, the current level of EPS is much higher than its decade inflation-adjusted average in the case of the US, but about the same in some other cases, including the UK and Germany.
- This has an important consequence: the wedges between the standard P/E ratios are typically smaller than those between the CAPEs. This can also be seen in Chart 4, where the light blue bars are standard 12m trailing P/E ratios. Even these gaps, though, are larger than usual in most cases.
- We think that if equities in the US were to outperform those elsewhere for a second successive decade, it would require a further increase in their relative valuations. This is because we don’t envisage it being driven by faster EPS growth again, in the way that it was after the GFC until this year.
- Yet valuation gaps are more likely to shrink in our view, given how large they have become. In this case, US equities seem set to underperform, unless we are wrong about the prospects for relative EPS growth.
Chart 3: Correlations Between CAPEs and 10Y Returns
Chart 4: CAPEs & Trailing P/E Ratios
Sources: Refinitiv, CE
Sources: Refinitiv, CE
John Higgins, Chief Markets Economist, +44 20 7811 3912, firstname.lastname@example.org