Barclays has launched three indices based on Robert Shiller’s cyclically-adjusted price-earnings ratio for sectors. Martin Steward met the Yale academic to discuss what they bring to the growing world of ‘smart beta’
The cyclically-adjusted price-earnings ratio, or CAPE, was 24 years old this year. First defined by John Campbell and Robert Shiller in a July 1988 Journal of Finance paper, it is usually calculated as the ratio of a company’s or market’s price to its 10-year earnings, adjusted for inflation. It was important as the first real attempt systematically to apply long-term earnings results to a stock valuation metric – an idea that went back at least to 1934 and Benjamin Graham and David Dodd’s groundbreaking study, Security Analysis, which recommended working with earnings data stretching back “not less than five years, and preferably seven to ten years”.
The ratio is very well known, not least because Shiller, Arthur Okun professor of economics and professor of finance at Yale University, used it to describe the tech-market bubble – and predict its bursting – in his 2000 book Irrational Exuberance. He repeated the trick with the US housing market in the 2005 edition of the same title.
Given its long pedigree and its apparent efficacy in calling over-valued markets, it is surprising that Barclays is only now the first to take the plunge and apply it in the investable index context. Its three products, called Enhanced Value, Focused Value and Market Neutral Value, all divide the US equity market into 10 sectors and weight them according to a specially modified version of their CAPEs.
The first is the one that looks most like a conventional index: the Shiller Barclays CAPE US Sector Tilted index represents all 10 sectors, overweighting the four most favoured and underweighting the six least favoured. The Focused Value Shiller Barclays CAPE US Sector index is long the four most favoured sectors and completely excludes the six least favoured. Finally, the Market Neutral Shiller Barclays CAPE US Sector Market Hedged index pairs the Focused Value four-sector portfolio with a short position in the S&P500, sized according to an analysis of the long-term beta of each sector, with a view to isolating the market-neutral CAPE value premium.
“We wanted to take this long-established and well-respected financial concept and find a way to make it work in the indexing context,” says Laurence Black, a director in equities and funds structured markets at Barclays. “We plan to launch various products linked to the indices, and we have also closed some trades on the Market Neutral Value index with pension funds that are attracted to the low-volatility exposure and the long-term value premium we identify.”
That all sounds simple – but in reality it was anything but. Testing the robustness of CAPE is tricky because good earnings data is difficult to source back beyond about 15 years. Even after deciding to focus on US industry sectors, Barclays’ research teams had to find earnings data on each individual company before aggregating all of that information up to sector level: they managed to create what Black claims is “a unique set of sector data going back 40 years”.
For a select group of sectors – industrials, utilities and railroads – the data stretches back to 1881, and plotting CAPEs back that far illustrates some startling bubbles and buying opportunities quite apart from the tech boom.
“The charts all look very different,” Shiller observes. “That makes the case for comparing different sectors as these indices do. In 1929, the highest CAPE ratio is to be found in utilities. These weren’t the boring, safe stocks that we know today: they were the glamorous new electricity companies. It was also high back in the 1890s – when gaslights were being fitted into homes and were the latest modern thing. By contrast look at how low the CAPE ratio went for railroads between the 1930s and the 1950s: it was seen as an endangered industry not only because of the arrival of aeroplanes and automobiles but because the labour unions had gotten very strong.”
At the closer, 40-year time horizon, where the US economy is diverse enough to identify 10 distinct sectors, the data shows a strong negative correlation, ranging from –0.25 in financials to –0.68 in utilities, between a sector’s CAPE and its subsequent annualised two-year real return. When the CAPE was very low, future returns were very high, and vice versa.
“At the moment you can see that the financials CAPE ratio is 12.03 times,” says Black. “Plot that on the scatter chart and you can see that that level of CAPE has been followed, on average, by a 10–12% annualised return. That’s attractive. Telecoms has a CAPE ratio of 34.89 times, which has led to an average two-year annualised return of –5%. That suggests that it’s not necessarily the best sector to be in.”
So that implies that the CAPE ratio tells us something useful about sector valuations. But it takes another step to turn that information to weighting those sectors within an index. The CAPE ratio of cyclical sectors like industrials or materials are not only more volatile than those of defensive sectors like utilities, they also fluctuate around very different levels: the industrials CAPE is pretty much always higher than the utilities CAPE. An index that weighted according to the pure CAPE would therefore always overweight utilities, and for that reason Barclays developed its ‘apples-to-apples’ Relative CAPE indicator, based on the ratio of the current CAPE ratio for a sector to that sector’s 20-year average CAPE ratio.
The added value of CAPE versus standard metrics like the 12-month price-to-earnings ratio has always been recognised as revealing divergence of valuations from their averages through time. Given that index construction is focused on those relative valuations at any one point in time, does the CAPE still justify itself against the standard valuation metrics? Does it perform differently, as part of an index-weighting process, than MSCI’s Value Weighted indices, for example, which divide price by book value and three-year data sets for sales and earnings?
Shiller argues that it does. It isn’t simply that 12-month P/E ratios for individual companies are exceptionally noisy – it is that they are exceptionally noisy in very different ways relative to one another, making the comparisons necessary for index weighting difficult. Black notes that the Barclays indices rotate through their sectors quite slowly, whereas weighting by 12-month P/E would increase turnover significantly.
“The 12-month earnings throw in noise not just from the business cycle but from earnings fluctuations year-to-year, and those fluctuations do cut across sectors,” adds Shiller. “You could have one sector whose earnings have fallen dramatically this year because it responds very sensitively to recession. Its 12-month P/E would look very high. But another sector whose sensitivity to the business cycle is different, but whose earnings are on the same ultimate trajectory, would show a 12-month P/E that remains relatively low. This is why Graham and Dodd told us that one year’s earnings was not enough to make any kind of judgement. The real puzzle is why the whole market doesn’t focus on these longer-term ratios.”
There are other objections that could be raised against the new CAPE indices. Like all mechanical value strategies, a pure CAPE-related one could expose investors to ‘value traps’ – the stocks that are ‘cheap’ for a very good reason or the sectors that are going to take a longer time to revert to their mean valuations.
Barclays has tackled this problem by introducing a momentum factor into the process. This is why the long or overweight portfolio is made up of four sectors rather than five: after identifying the five most undervalued sectors, the one that exhibits the greatest momentum towards a lower Relative CAPE is removed or underweighted.
Then there is the decision to focus at sector level. This means that stocks with high CAPEs that happen to be in a sector with a low CAPE will be included or overweighted, and vice versa. Shiller’s argument here is precisely that focusing on sectors makes sense because individual companies can exhibit valuations out-of-kilter with their industry groups – if they innovate in such a way to take market share from competitors, for example.
“A very high CAPE can be perfectly justifiable if earnings are growing fast and lagged earnings begin to look small by comparison,” he observes. “For sectors the ratio is much more stable.”
Not that Shiller suggests investors rule out alternative value indices, or value indices that weight individual stocks one against another. He sees the Barclays products as deploying one process for exploiting one particular market anomaly, which should ideally be implemented alongside a range of other systematic strategies targeting other such anomalies.
For this reason, he is puzzled at the way some alternatively-weighted products – and particularly the most popular of them all, fundamental indexation – have been presented as better solutions for core portfolios than traditional market cap-weighting strategies. Once we have recognised that the supposedly optimal cap-weighted portfolio is particularly subject to bubbles, the correct response is not to put forward an alternative as the one-and-only optimal portfolio.
“The real insight to be gained from recognising that the supposedly optimal market cap-weighted portfolio has anomalies – like bubbles – is that those anomalies can be exploited,” Shiller explains. “Once you’ve recognised that, why would you stop at fundamental weighting? There is a premium for value, there is a premium for momentum, there appears to be a premium for low-beta stocks – and there are others, too. What we are doing is focusing in on one specific anomaly – the value premium – and presenting a proven way to identify fundamental mispricing. Our ‘Focused Value’ index is pretty aggressive in that respect: we wouldn’t anticipate anyone putting their entire equity portfolio in it, but it would fit nicely into a broader equity strategy.”