3. Valuation multiples: P/E, P/B, EV/EBITDA and their traps

Theory 3: Asset classes and how to read a company

A P/E of thirty can be extremely expensive. The same P/E of thirty, in another company, can be cheap. And the number has not changed.

We now know, broadly, how to read a company's three financial statements. The natural next step is to ask whether the price at which a share trades is reasonable relative to those numbers.

That is what valuation multiples are for: ratios that compare the market price with some fundamental metric of the company.

The P/E, and what it does not tell you

The most quoted by far is the P/E ratio (price to earnings). It is calculated by dividing the share price by earnings per share (EPS).

A P/E of 20 means, in simple terms, that the market is willing to pay 20 times the company's current annual earnings to own a share. Put another way: if earnings stayed constant, it would take 20 years to 'recover' that price from earnings alone.

What is interesting is that every reading admits two opposite interpretations.

• A high P/E may reflect that the market expects strong future growth, which would justify paying more for each euro of current earnings. Or it may simply reflect that the stock is overvalued.
• A low P/E may reflect a genuine bargain. Or it may reflect that the market anticipates a deterioration of the business that current earnings do not yet show.

The P/E on its own does not distinguish between these scenarios. It merely compares price and earnings, saying nothing about why that relationship is what it is.

The P/B, and where it stops working

P/B (price to book) compares the market price with the shareholders' equity per share we saw on the balance sheet in the previous module.

It is especially useful in sectors intensive in physical assets, such as banks or industrials. There, book value bears a reasonable relation to what the company actually owns.

It is far less useful in companies whose value lies mainly in intangible assets that are hard to account for: brand, patents, talent, data. In those cases, book value can massively understate the real economic value of the business.

The EV/EBITDA, and the two things it corrects

EV/EBITDA (enterprise value over EBITDA) is the multiple preferred by many professional analysts, precisely because it corrects two limitations of the P/E.

First, it uses enterprise value instead of market capitalization. EV adds the company's net debt to the market cap, reflecting the real cost of 'buying' the whole business, not just the part belonging to shareholders. That allows a fairer comparison between companies with very different debt structures.

Second, it uses EBITDA instead of net income. In doing so it removes the effect of two decisions:

• How each company finances its debt, that is, the interest.
• How it accounts for the depreciation of its assets.

These are two items that can vary a lot between companies in the same sector for purely accounting or financing reasons, without reflecting real differences in the quality of the operating business.

The trap: comparing out of context

The most common trap when using any of these multiples is comparing them out of context.

There lies the contradiction we opened with. A P/E of 30 can be extremely expensive for a mature utility growing at 2% a year, and reasonable — even cheap — for a technology company growing at 40% a year.

What matters is not the number in isolation, but the number relative to expected growth, to the sector and to the point in the economic cycle.

There are two ways of comparing that avoid much of this trap:

• Comparing a company's P/E against its own historical average.
• Comparing it against companies in the same sector with a similar business model.

And there is one that is almost a guarantee of reaching the wrong conclusion: comparing a bank's P/E directly against a software company's, with no further adjustment.

It is worth fixing the underlying idea. Multiples are a tool for quick comparison, not a calculator of intrinsic value: they say 'expensive or cheap relative to what', never 'expensive or cheap in absolute terms'.

The PEG: putting growth into the equation

One variant tries to correct exactly the P/E's central limitation, which is that it says nothing about expected growth. It is the PEG ratio, which divides the P/E by the expected annual earnings growth rate.

Two examples make it clear:

• A company with a P/E of 30 and expected growth of 30% a year has a PEG of 1.
• Another with a P/E of 15 but expected growth of only 5% a year has a PEG of 3.

Through this lens, the first company — with the 'higher' P/E in absolute terms — turns out to be relatively cheaper once adjusted for how much it is expected to grow. Exactly the nuance an isolated P/E cannot capture.

The PEG has its own limitations. Starting with the fact that it depends entirely on how reliable the growth estimate used is. Unlike the P/E, that figure is not a verifiable historical fact, but a projection subject to all kinds of error.

Even so, it remains a useful reminder: no multiple should ever be read in isolation, without asking what other relevant variable it is leaving out of the equation.