Not a peg to grasp

For such folk it would be equally nice to think that the pursuit of success in investing in shares could be distilled to a single digit – and it’s not as if there would be a shortage of numbers to choose from; or, at least, of simple formulae that each churn out a single figure.

High on the list of contenders would be PEG factors because they take a ubiquitous measure – the price/earnings (PE) ratio – then turbo-charge it, while still holding it down to a single figure, by dividing it by a growth rate in earnings. This two-in-one feature was supposed to endow PEG factors with marvellous properties. After all, if just one number can show how a stock is cheap just so long as likely earnings growth is outstripping its rating, then it must be very special.

Well, maybe not that special. Three weeks ago (Bearbull, 5 October 2018) I explained how – with some heaving and shoving – PEG factors could be pushed into a conventional arbitrage pricing model, thus giving the appearance of intellectual respectability. However, bigger imponderables remained: are PEGs any use? Do they help predict share price performance? These are vital questions to which academia has paid little attention possibly because of PEG's debatable status.

I can’t claim to supply a definitive response. However, using data for the components of the FTSE 100 index, I have churned PEG factors of varying sorts for the four years 2014 to 2017 inclusive and found barely a tenuous relationship between PEGs and share price performance.

The table shows the key data, but it needs explanation. A big difficulty in assessing the usefulness of PEGs is that they have no settled definition. Sure, we are dividing a PE ratio by a growth rate in earnings, but it’s largely a matter of choice which numerator to use and, especially, which denominator (ie, the growth rate).

So I have constructed both my own PEG factors and used those provided off the shelf by S&P Capital IQ, the database we often use here at Investors Chronicle. First, let’s discuss the table's Capital IQ PEGs because they are simpler, only offering factors based on 2017’s earnings growth as forecast by City analysts. This also provides a chance to explain what the table is about.

Capital IQ’s data excludes negative PEG factors. That’s good because, in a sense, those can’t exist within a logical framework for assessing PEGs where the lower the PEG factor, the more attractive the stock in question. That logic would make negative PEGs better still, yet that couldn’t be so since the negative value implies either a loss-making company or, more likely, shrinking earnings. It also aids simplicity that Capital IQ’s PEGs are based on forecast earnings and the City’s sell-side analysts rarely forecast that a company’s earnings – massaged free of inconvenient blips – will fall.

This might make Capital IQ’s PEGs better than my manufactured ones. If so, it doesn’t show in the data. The key columns in the table are for ‘Slope of line’ and ‘R-squared’. These show the extent to which a company’s PEG factor ‘drives’ its share price. In theory, higher price returns should correlate with lower PEGs so, in a conventional scattergram, the regression line showing the average relationship between a company’s PEG factor and its share price return should slope down, meaning that the value should be negative. However, for Capital IQ PEGs, the regression line’s slope is upwards (or positive) – in a sense, the wrong way. No matter, since the line bears almost no relation to the wildly scattered dots anyway. That scattering is quantified by R-squared, where a perfect correlation would produce a score of 100. Just 0.4 for the Capital IQ PEGs is another way of saying 'forget it'.

Much the same could be said of the PEGs constructed using historic earnings data, although there is one bright spot. For my home-made PEG factors, built using an estimated forecast PE ratio and where all negative PEGs were eliminated, for each time period the line slopes downwards, meaning that better share price performance was linked – albeit very loosely – with lower PEG factors.

But should we be surprised that PEGs have so little predictive value? Not really. In investment analysis, to put faith in a single ratio – whatever it may be – is touchingly naive. Meanwhile, PEGs may have some use. At least, if I find a low PEG based on a sensible forecast of a company’s earnings where the denominator is an intuitively-sustainable medium-term growth rate in future earnings, then I may have something worthy of examination. Apply that rule of thumb by all means, but don’t expect it to be the answer to everything.