Choose peers by business model, growth and risk, from SGX and from overseas listings.
A stock screener lets you type in an industry label and sort every company under it by a multiple. The list shows what the market pays for that industry in general. It says little about what the market should pay for one specific business inside it, because companies with the same label can make money in very different ways.
A peer group is the set of companies you compare your company with on multiples. Choosing it is where most of the judgement in a comparable companies analysis sits. If the peers are wrong, their median gives you a precise answer to the wrong question.
Marcus is valuing Larkspur, a semiconductor parts supplier. He types "semiconductor" into a stock screener and sorts by P/E. Larkspur sits near the bottom of a list running from 8 times to 60 times earnings (example figures). For a moment it looks like a bargain. Then he reads the names above it: the world's largest chip foundry, a lithography machine maker whose machines Larkspur's customers buy, two memory chipmakers and a chip designer. None of them does what Larkspur does.
Good peers share three things with your company: how they make money (business model), how fast they grow, and what can go wrong (risk). So Marcus starts with Larkspur's economics. It is a precision manufacturer selling parts and modules to a few large equipment makers. It is paid per part, its demand follows chipmakers' spending, and it has two big customers.
He tests each candidate against those three points. The chip designer fails on business model because it sells its own products to device makers. The equipment maker fails on position in the value chain: it is Larkspur's customer, with pricing power and a global service business that Larkspur lacks. An SGX precision engineer making medical device parts matches the "precision engineering" label but fails on risk, because its demand does not follow the chip cycle.
Marcus ends up with five peers. He writes one line on why each chosen peer fits and one line on why each rejected candidate does not. He records the same six figures for every peer, in the same order: P/E, EV/EBITDA, P/B, free cash flow yield, revenue growth per year over five years, and operating margin.
Peer A, an SGX precision engineer supplying chip equipment makers: P/E 10.5, EV/EBITDA 6.0, P/B 1.4, FCF yield 6.5%, growth 6% a year, margin 12%. Peer B, an SGX maker of parts for chip testing equipment: P/E 13.0, EV/EBITDA 7.5, P/B 1.9, FCF yield 5.5%, growth 9%, margin 15%. Peer C, a contract manufacturer for equipment makers, listed on Bursa Malaysia: P/E 18.0, EV/EBITDA 10.0, P/B 2.6, FCF yield 3.5%, growth 12%, margin 13%. Peer D, a Tokyo-listed precision parts maker with slow growth and a March year end: P/E 12.0, EV/EBITDA 6.5, P/B 1.1, FCF yield 5.0%, growth 4%, margin 10%. Peer E, a US-listed supplier of whole subsystems to the same equipment makers, reporting under US GAAP and the most richly valued of the five: P/E 24.0, EV/EBITDA 14.0, P/B 3.5, FCF yield 3.0%, growth 11%, margin 14%.
On the same basis, Larkspur has a P/E of 11.4, EV/EBITDA of 6.25, P/B of 1.7, FCF yield of 5.8%, growth of about 7.5% a year and an operating margin of 14%.
A multiple from one company's annual report and another company's data provider page may not measure the same thing, so line them up before you compare.
Use the same kind of earnings for every company: either all underlying or all reported. Lesson 9.1, "P/E, EV/EBITDA, P/B, P/S and FCF yield: what each measures", gave this warning. Put leases on the same footing for every company too. This matters for Peer E, which reports under US GAAP; lesson 9.2 showed how US GAAP lease treatment moves EBITDA.
Then align financial years. Peer D's year ends in March and Larkspur's ends in December, so Marcus has to calendarise: build a December-year figure from the two March years it overlaps. Three months come from the year ending this March and nine months from the year ending next March, so the calendar figure is 1/4 of the first year plus 3/4 of the second. If you skip this in a fast-moving cycle, you can end up comparing one company's boom year with another company's slump year.
The two SGX peers have an average P/E of 11.75 ((10.5 + 13.0) / 2). The three overseas peers, C, D and E, have a median P/E of 18. Part of that gap comes from the businesses themselves. Peers C and E grew at 12% and 11% a year, so investors pay more for each dollar of today's profit.
Part of the gap may come from the market. Smaller SGX companies trade thinly, are followed by few analysts, and are too small for many large funds to own. Module 12 returns to this in lesson 12.7, "Survivorship bias, SPIVA and where your edge is and is not". Investors who may need to sell in a hurry tend to demand a lower price for shares that are hard to sell. This is the illiquidity premium, from lesson 2.7, "Risk premia: what you are paid to hold equity, term, credit and illiquidity".
A gap between Larkspur's multiple and Peer E's is therefore no proof that Larkspur is cheap. Part of it is the price of being small and listed in Singapore, and that part may never close. When you explain any gap between peers, separate business differences such as growth from market differences such as liquidity, analyst coverage and fund ownership.
The median is the middle value once the figures are sorted. Medians resist outliers, which is why analysts use them for peer groups. The mean P/E of Marcus's five peers is 15.5 (77.5 / 5) and the median is 13. Peer E, at 24 times earnings, pulls the mean up by more than two points on its own.
Larkspur's underlying EPS is 14 cents a share. The mean P/E gives 15.5 x 0.14, about S$2.17 a share. The median gives 13 x 0.14, about S$1.82 a share. Using the mean adds 35 cents, all because of one richly valued American company.
With only five peers, the result can hinge on a single name. Drop each peer in turn and see how far the median moves. A small move means the group passes; a big jump means the group is too small or too mixed to trust. When Marcus drops Peer E, the median P/E falls only to 12.5, so his group passes.
The peers' median EV/EBITDA is 7.5 times. Applied to Larkspur, that implies about S$1.93 a share after net debt of S$20 million is deducted.
Marcus's central DCF range from lesson 8.8, "Value your modelled company with a DCF", was about S$1.58 to S$2.08 a share. The P/E median value of S$1.82 and the EV/EBITDA median value of S$1.93 both fall inside it. That agreement is reassuring but not proof, because both methods carry the same view of the industry.
For the activity, list the five companies you'll compare yours with, at least two of them listed outside Singapore, each with one line on why it belongs.
Pick five peers for the company you modelled, with at least two listed outside Singapore. Write one line for each on why it fits.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).