Lately, I have been interested in investigating “uncertainty” as a way of assessing the desirability of purchasing shares. My idea is based on the observation that returns on low P/B stocks come from “the long tail” – in other words, the majority do badly, but it is the few surprise winners that produce the outperformers. It is a theme I started in my blog about a Schroder article. Piotroski approached the problem low P/B stocks by trying to score for improving metrics. I am interested in approaching it from a different angle: does high variability in the low P/B stocks correlate positively with return? The reasoning is that by investing in high uncertainty stocks, you are moving out along the tail. The more that nobody knows how to guage the future, the better off you are.
How to measure this “uncertainty”, or “unpredictability”? I took a look at 4 metrics which seemed sensible candidates: ROE, OPM (Operating Profit Margin), and Beta. I selected 7 shares: BARC (Barclays), BATS (Brit American Tobacco), VOD (Vodafone), TSCO (Tesco), BWY (Bellway), ULVR (Unilever) and Glaxo (GSK). The metrics I present below are sample standard deviations over an extended time period, usually a decade. The exception is Beta, which I took from Digital Look.
Here is the table I compiled:
ROE OPM PER BETA BARC 7.46 N/A 1.55 1.35 BATS 11.05 10.61 1.65 0.76 VOD 1.71 3.52 1.96 1.26 TSCO 0.91 0.77 2.54 0.64 BWY 6.90 5.36 8.94 1.62 ULVR 15.60 1.16 2.69 0.78 GSK 12.72 2.65 2.27 0.77
My expectation is that defensive, predictable companies would have low standard deviations, whilst the cyclic, unpredictable ones would have high standard deviations. So I would expect low figures for BATS, VOD, ULVR and GSK, but high values for BARC and BWY.
Whilst somewhat true, there is considerable muddying of the waters. BARC, for example, showed a very narrow range of PERs. ULVR, a solid defensive company, showed the highed variability of ROEs. BATS showed the highest variability in OPM. Its margins have been rising steadily over a decade, and now appears to be the highest they have ever been.
In fact, the single-most convincing statistic to separate the defensives from the roller-coaster stocks appears to be beta. The others appear to be limited in their discriminative abilities.
Update 19-Jun-2014: To clarify my thinking, what I am aiming for is greater “optionaility” in the low P/B stocks. I also found an interesting PDF available at NYU Stern with the title “Stable Earnings, Better Investment?”. I reckon there’s good odds that it was written by Aswath Damodaran.