When seeking value in the stock market, it helps to start with industries rather than companies. Cheapness is often found clustered in certain parts of the market, perhaps because an industry is going through hardship or because it has fallen out of favour with investors for the moment.
In recent years, the cyclically-adjusted price-earnings ratio or 'CAPE' has become a widespread tool for spotting value in stock market indices as a whole. Rather than valuing today's price in terms of one year's earnings - whether trailing or forecast, CAPE involves looking at today's price in terms of earnings averaged over several years, adjusted for inflation.
The logic of using several years' earnings is to smooth out the distortions of the economic cycle. In a boom year, earnings may be high, which can make stocks seem flatteringly cheap, especially if bad times are just round the corner. Likewise, in a recession year, earnings tend to get depressed, which can make stocks look misleadingly dear.
I recently showed how CAPE could be used for picking the cheapest national stock markets around the world: http://bit.ly/1jBlLi7. (I'll be following that up shortly with a new regular piece on national indices.) Since then, I've been investigating whether CAPE could also be used for weighing up UK industry sectors.
CAPE's best period
Professor Robert Shiller - the doyen of the cyclically-adjusted PE approach - famously uses 10 years of earnings when calculating CAPE for the S&P 500 since the 19th century. Data for the FTSE 350's 40 sectors is comparatively limited, with prices only going back only to 1985. So I've spliced that data with Datastream's lookalike indices, which start in 1965.
Where CAPE works best
| Covers 1974-2004 | Best CAPE | Holding period | Explanatory power |
|---|---|---|---|
| Industry | (# of yrs earnings) | (# of years) | R-sq, higher better |
| Aerospace & defence | 10 | 7 | 0.772 |
| Support services | 10 | 10 | 0.702 |
| Food producers | 10 | 10 | 0.697 |
| Healthcare equipment & services | 10 | 10 | 0.620 |
| Construction & building materials | 10 | 10 | 0.612 |
For each of the sector groupings, I worked out 10 PE ratios, starting with a single year's earnings, extending right up to a whole decade's. I then tested each sector's 10 ratios to see which one had been best for forecasting real returns thereafter, also over one- to 10-year horizons. Specifically, I asked how what proportion of subsequent returns was explained - in the statistical sense - by each ratio.
As was the case with whole-country indices, 10-year CAPE was typically the most powerful valuation tool. It worked best both for shorter holding periods of up to five years and for ones of up to 10 years. The longer the holding period, the more predictive CAPE was. So a 10-year CAPE has generally told us much more about likely returns over the next decade than a two-year CAPE has told us about next year's outlook.
And where it doesn't
| Covers 1974-2004 | Best CAPE | Holding period | Explanatory power |
|---|---|---|---|
| Industry | (# of yrs earnings) | (# of years) | R-sq, higher better |
| Software | 1 | 6 | 0.030 |
| Personal goods | 10 | 10 | 0.097 |
| Mining | 9 | 6 | 0.109 |
| Chemicals | 10 | 7 | 0.147 |
| Reits | 10 | 6 | 0.17 |
To give one example, pharmaceuticals & biotech's 10-year CAPE has explained around half of its returns over a 10-year holding period, with a low CAPE leading to high returns and vice versa. Therefore, knowing what pharmaceuticals' CAPE was would have given you a vast amount of information about the prospects for your returns.
Low CAPE, healthy returns
CAPE hasn't been useful for all sectors, however. Even on a multi-year view, CAPE explained just one-tenth of returns in personal goods & mining. For software, it was virtually irrelevant, explaining a pathetic 3 per cent of real returns over a six-year timeframe.
Today's cheapest sectors
In the accompanying table, I show the recent 10-year CAPE valuations for the FTSE sectors. Miners, banks, oil & gas are the cheapest, while chemicals, personal goods and technology hardware comes out very dear.
| Sector | Price | 10-yr CAPE |
|---|---|---|
| Mining | 17,440 | 9.6 |
| Banks | 4,668 | 10.9 |
| Oil & gas producers | 7,919 | 12.9 |
| Fixed line telecom | 4,585 | 13.5 |
| Electricity | 9,300 | 17.3 |
| Non-life insurance | 1,840 | 17.6 |
| Food & drug retail | 4,300 | 17.6 |
| Food producers | 7,121 | 17.8 |
| FTSE All-Share | 3,616 | 18.5 |
| Gas, water & multiutilities | 5,695 | 20.3 |
| Construction & building materials | 4,555 | 21.0 |
| Travel & leisure | 7,430 | 21.9 |
| General retailers | 2,816 | 22.7 |
| Pharnaceuticals & biotech | 12,577 | 23.6 |
| Tobacco | 34,673 | 24.9 |
| Aerospace & defence | 4,886 | 25.5 |
| Life insurance | 7,031 | 25.6 |
| Industrial transport | 3,348 | 27.6 |
| General industrials | 4,493 | 27.7 |
| Media | 6,308 | 30.8 |
| Support services | 6,412 | 33.1 |
| Beverages | 13,306 | 35.0 |
| Healthcare equipment & services | 5,257 | 35.3 |
| Financial services | 11,892 | 36.0 |
| Auto & parts | 9,633 | 37.9 |
| Industrial engineering | 10,358 | 39.2 |
| Software | 1,160 | 40.0 |
| Electronic & electrical equipment | 4,324 | 41.2 |
| Chemicals | 11,005 | 43.5 |
| Personal goods | 24,632 | 50.4 |
| Technology hardware | 1,133 | 102.4 |
Source: FTSE via Datastream
Now, there are some caveats here. First of all, half a century of data isn't very much when you're analysing whole decades' returns using 10 years of earnings each time. Secondly, the Datastream earnings data is flawed because it bizarrely excludes loss-makers. Despite these shortcomings, the results are decent enough.
I will be publishing an update on how to use CAPE sector data in your portfolio planning shortly.
