Tell when a multiple makes a stock look cheap or expensive for the wrong reasons.
Before you act on a low P/E, look up the company's earnings per share for each of the last five years and check whether the current year is the highest. A multiple can measure a company at the worst possible moment even when nothing about the company is hidden and the arithmetic is right. Lesson 9.1, "P/E, EV/EBITDA, P/B, P/S and FCF yield: what each measures", set out what each multiple compares. This happens in three situations: the business cycle, debt, and accounting choices.
A friend of Marcus bought shares in a container shipping line (an invented company) at a P/E of 6. Freight rates were at a record and profit had quadrupled, so the shares looked absurdly cheap against the market. A year later freight rates had halved, profit had fallen by four fifths and the shares were a third lower. On the new, smaller profit the P/E was 20.
Cyclical companies look cheapest at the top of their cycle. As a rule of thumb, when a cyclical company's profit is at a peak, investors expect it to fall, so they pay a low multiple of it. At a trough they expect recovery, so they pay a high multiple of a small number. Buying a cyclical because the P/E is low and selling because it is high does exactly the wrong thing. The fix is to value the company on earnings across a whole cycle. Normalised earnings are the profit a business would earn in an average year of its cycle.
The simplest method takes earnings per share across a full cycle, averages them, and divides the share price by that average. The shipping line earned 40 cents a share in the boom year, then 8, 12, 20 and 15 cents, an average of 19 cents. The peak share price of S$2.40 was a P/E of 6, and the price a year later, S$1.60 on earnings of 8 cents, was a P/E of 20. On normalised earnings, S$2.40 / 19 cents is about 12.6 times, not 6, and S$1.60 / 19 cents is about 8.4 times, not 20. Once earnings were normalised, the boom-year price was the expensive one.
A better method averages margins rather than profit, because revenue often grows across the cycle. Average the operating margins across a full cycle and apply that average to current revenue to get normalised operating profit. Deduct interest and tax to get normalised earnings per share, then compare the P/E on that figure with the P/E on actual profit to see whether the current year flatters the multiple.
Larkspur is a mild cyclical. Lesson 7.2, "Income statement: revenue quality, margin stack and one-offs", showed its operating margin at about 11% in the FY4 trough and 16% at the FY3 peak, averaging about 13.4%. FY5 revenue is S$400 million, and 13.4% x S$400 million gives normalised operating profit of about S$53.6 million, against S$56 million reported. After interest and tax, normalised EPS is about 13.7 cents a share. The normalised P/E is about 11.7, against 11.4 on actual underlying profit. FY5 sits close to the middle of Larkspur's cycle, so its P/E is not badly flattered. Check where a company sits in its cycle before comparing it with a peer measured at that peer's own peak.
Debt makes P/E look low. Company A and Company B (both invented) each have operating profit of S$50 million and an enterprise value of S$500 million, and trade at 10 times EBIT. A has no debt. B has borrowed S$300 million at 5% and has no spare cash. The tax rate is 17%.
A's profit after tax is about S$41.5 million on equity worth S$500 million, a P/E of about 12. B pays S$15 million in interest, leaving profit of about S$29 million on equity worth S$200 million (S$500 million less S$300 million of debt), a P/E of about 6.9. Two identical businesses with the same enterprise value end up on very different P/E ratios because one carries debt.
B's lower P/E is the market charging for risk, because B's shareholders stand behind the lenders. Interest does not shrink when operating profit falls, so debt magnifies every swing in profit for shareholders. If operating profit falls 20% to S$40 million at both companies, A's profit falls 20%, while B's falls to about S$20.8 million, a drop of about 29%. Levered equity deserves a lower multiple for this reason. EV/EBIT and EV/EBITDA show A and B as the same price for the same business.
Multiples compare one line of the accounts across companies, so anything that changes that line without changing the business distorts the comparison. Two accounting choices do most of the damage: lease accounting, and capitalising and depreciating costs.
Under SFRS(I) 16 and IFRS 16, most lease costs leave operating expenses and reappear as depreciation and interest, both below EBITDA. Under US GAAP, an operating lease still shows a single lease cost inside operating expenses, above EBITDA, so a US-GAAP reporter shows lower EBITDA than an SFRS(I) 16 / IFRS 16 reporter with the same lease. A US-reporting twin of Larkspur (invented) pays S$8 million a year for its factory land, and its EBITDA is about S$72 million against Larkspur's S$80 million. If a data provider leaves the lease liability out of the twin's enterprise value, the twin trades at about 6.7 times EBITDA against Larkspur's 6.25 times, and looks more expensive purely because of the rulebook. Lesson 6.4, "Notes to the accounts: revenue, segments, leases and debt", showed where to find the lease figures that put both companies on the same basis.
Capitalising development spending means recording it as an asset and depreciating it over several years, as opposed to expensing it as it is incurred. A company that capitalises reports higher EBITDA than one that expenses the same spending as it goes. Depreciation lives move EBIT and P/E, because a shorter life means a larger yearly charge. S$100 million of machines over ten years gives a S$10 million yearly charge, and over five years the charge doubles to S$20 million. Neither company is lying when policies differ, but their P/E ratios are not comparable until you adjust. The accounting policies note says which choices each company made.
When a company trades well above or below its peers, ask all three questions:
Is the company's profit at a peak or a trough of its own cycle? Does it carry much more debt than its peers? Do its accounting policies on leases, development costs and depreciation differ from its peers'?
Most large valuation gaps between peers turn out to be explained by cycle position, debt or accounting policy. Gaps that survive all three checks are worth your time.
For the activity, pick a company whose profits clearly rise and fall with a cycle, such as a shipping line, a commodity producer or a chipmaker, and find its earnings and share price in its last peak year and its last trough year.
Find a cyclical company. Compare its P/E at its last earnings peak with its P/E at its last trough.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).