This article is part of Morningstar's "Perspectives" series, written by third-party contributors. Here, L&G Equity Strategist Lars Kreckel aims to clear up some of the common misconceptions around equity valuations.
Let’s start with the good news: valuation multiples are useful for investors trying to estimate the long-term return prospects of equities. There is broad historic evidence that multiples such as the popular CAPE (cyclically adjusted PE ratio, calculated as the current price divided by the ten-year average of reported earnings per share, both adjusted for inflation) have a good correlation with subsequent ten-year equity returns. Figure 1 shows that generally the lower the CAPE when buying equities the better the return over the next decade and vice versa.
The bad news is that for all their usefulness in long-term equity return predictions, valuations are nearly useless when it comes to more tactical decisions. In statistical terms, figure 1 shows that the correlation between the CAPE and the subsequent 10-year equity return is a pretty high correlation (R2) of 69%, 100% being the maximum, but when we shorten the investment horizon to 12 months the R2 falls to a paltry 8% (see figure 2). And this is not just a problem of the more long-term oriented CAPE; all other valuation multiples fare similarly poorly when testing their predictive power for shorter term returns.
The lack of correlation between valuations and short-term returns is not exactly a ground-breaking discovery. It is very well covered in academic literature and is the reason we describe our tactical investment approach as value-aware rather than value-driven. But it makes it all the more surprising how much time is devoted to valuations in analyst reports and investor discussions; especially as many investors’ time horizon is likely to be closer to one year than ten, and in many cases may be even shorter.
Complex, And Not Always Useful
Most investors would agree that the value of a financial asset is the present value of its future cash flows. Unfortunately, calculating this is a very difficult exercise. It requires forecasting cash flows, which are difficult to predict to begin with, into an uncertain future and discounting them at an unknown rate.
In other words, it requires making a large number of assumptions about factors we cannot reliably predict. Multiples offer a tempting alternative. In the case of a PE ratio, all we need to estimate are next year’s earnings and we can observe the current share price. The ratio of the two will tell us whether a stock is cheap or expensive. Seems easy, but this shortcut means that the difficult parts of the calculation that are explicit in a discounted cash flow model are made implicit, hidden out of sight, in a valuation multiple. We still need to understand what the implicit assumptions are in order to translate the number into an investment decision.
Another common assumption is that low valuations are a protection against excessive losses. But is this really true? Having shown that there is no correlation between valuations and subsequent short-term returns should make one doubt that assertion and a look at some real life examples confirms our suspicion. Two of the most recent sharp corrections occurred in Turkey during last year’s ‘Taper Tantrum’ and in Russia with the uncertainty around Ukraine earlier this year. In neither case was there any correlation at the individual stock level between the PE before the correction and its peak to trough drop during the correction (figure 3). So the cheapest stocks in Russia and Turkey fell no more but also no less than the most expensive stocks.
One of the best examples was Gazprom (GAZP), which traded on what by most standards is an extremely low PE of 3x before the Crimea invasion, but still lost almost a quarter of its value, roughly the Russian market average, in the weeks that followed. Perhaps this only happens in a sudden correction, like a crash. Could investors behave differently in a more gradual bear market? However, we find the same results in the last two bear markets, where ‘cheaper’ technology stocks fell no less than ‘more expensive’ technology stocks when the TMT bubble imploded and ‘cheap’ banks fell no less than ‘more expensive’ banks during the more recent financial crisis.
Which Multiple to Look At?
There are so many valuation multiples available that at any one time it is possible to find at least one that makes equities look expensive or cheap. Confirmation bias makes it tempting to focus on the multiples that support one’s case.
Is there a more rigorous approach of narrowing down the number of multiples to look at? Are some valuation multiples more useful than others? There are different approaches to answering these questions. The typical approach is to discuss the intellectual merits of specific multiples and perhaps make an adjustment to correct for a perceived inaccuracy of a specific multiple.
We would argue that the most important characteristic a multiple must have is not its intellectual purity but rather that it has a good track record of predicting future equity returns. As long as a multiple is not intellectually flawed our aim is to maximise the predictive power for future returns. This is why we look at valuations in the first place.
An Historical Back-Test
With this in mind, we can run some simple tests on how well different multiples have achieved this goal. Assuming valuations mean-revert in the long-run, say ten years, the total return equities should deliver can be easily calculated as the earnings growth rate, gradual reversion to the mean multiple and the dividend yield. As a test case, we looked at US equities, where the longest historic data series are available.
The first conclusion is that we can discard some valuation multiples. The relative multiples we looked at compare equity valuations with those of bonds, such as equity risk premium and earnings yield ratio, have a very poor track record at predicting long-term equity returns. Arguably this is not surprising as relative multiples aren’t designed to predict absolute returns, but given they are often used as an argument for or against buying equities, it is worth noting that they have had exceptionally poor track records in this task.
For the absolute valuations, the most surprising conclusion is that it doesn’t matter which multiple you use because most have a similarly good track record. We get some of the best results for enterprise value based measures and price to book, but we only have data for these going back to 1980. Of those with a history back to 1940 we get particularly good results for cyclically adjusted PEs such as the Shiller and Bianco PEs with R2 of around 70% with subsequent ten year returns, but other multiples are not far behind.
Summary
What does this mean? It tells us that the next time you hear or read about valuations as an argument to buy or sell equities remember that just because equities are cheap or expensive by historic standards has absolutely no bearing on whether they will go up or down in the next year or two. In addition, be aware that by using multiples you are taking a short-cut that makes many implicit assumptions and that just because a stock may be cheap, it is no realistic protection against painful losses. Be ‘valuation aware’ rather than ‘valuation driven’.
The second main conclusion is that equities may no longer be called cheap, but are equally not yet excessively expensive. Looking at the multiples that have historically been best at predicting future returns suggest we should expect US equities to deliver in the region of 4% and 7% total returns per year over the next decade, while European equities will deliver between 5.5% and 11%. That may sound below par by historic standards, but remains significantly more than one can expect from most alternatives in the fixed income space.
Morningstar Disclaimer
The views contained herein are those of the author(s) and not necessarily those of Morningstar. If you are interested in Morningstar featuring your content on our website, please email submissions to UKEditorial@morningstar.com.
.jpg)
