Trading comps price minority stakes changing hands on an exchange. Precedent-transaction comps price something else, entire companies changing hands in M&A. That's a claim-level distinction, as real as equity versus enterprise:
- A share price is the market's price for a non-controlling claim on existing management's plans.
- A deal price buys control: the right to redirect cash flows, replace management, re-lever the balance sheet, cut costs, and capture synergies.
That right is worth paying for, so deal multiples systematically exceed trading multiples. The gap is the control premium. US studies of premia paid over the pre-announcement ("unaffected") price have historically clustered around 25 to 40% at the median, swinging with the cycle. Treat the range as an order of magnitude. Before you use a number, verify it against a current study. FactSet Mergerstat publishes the standard series. And decompose the premium before you trust it:
Deal price = standalone value + value of control (run it better) + synergies (run it together) − bargaining failuresAdvertisement
How much of a control premium is rational?
The premium is only rational up to standalone value + control value + synergies to this buyer. Auctions push prices toward that ceiling, and past it. That's the winner's curse: the highest bidder is, by construction, the one with the most optimistic assumptions. That gives you two working rules. (1) Precedent multiples are a seller's benchmark. They're the right tool for "what could this company fetch in a sale?" and the wrong tool for "what should I pay for 100 shares?" Pay a control premium for a minority stake and you've bought a premium with no control. (2) When you are valuing a takeover, cap the defensible price with your own synergy estimate. What the last overexcited acquirer paid is no cap at all.
How do you build a precedent transactions table?
The grammar is the same, with deal-specific hygiene on top. Precedent tables use EV/EBITDA and EV/Sales, because the equity is being re-levered anyway, plus the sector's unit multiple (EV per tonne of cement capacity against replacement cost, EV per bed, EV per subscriber). Then screen each deal on five things. Its age matters, because a 2021-vintage multiple prices a different rate regime. So does the consideration, because stock struck at an inflated currency overstates "price". Ask whether the sale was an auction or a negotiation, whether the seller was distressed, and whether the buyer was a strategic with synergies or a financial buyer priced off leverage math. Adjust or exclude accordingly. Five clean precedents beat fifteen polluted ones.
The data is free in both markets. In the US, the merger proxy (DEFM14A on EDGAR) for any public deal contains the bankers' fairness opinions. Those reproduce their entire comps and precedent tables, which makes them a working analyst's cheat sheet hiding in plain sight. The 8-K announces the terms. In India, look at stock-exchange announcements and scheme-of-arrangement documents. For open offers, there's the letter of offer on sebi.gov.in, which discloses how the offer price was built up. SEBI's Takeover Regulations (2011) set floor-price rules (Reg 8: among other tests, the 60-trading-day volume-weighted average), so Indian open-offer "premia" are partly shaped by regulation. Control works differently in India, too. With promoters typically holding 50 to 75%, true control rarely trades. Deals are negotiated stake sales plus a mandatory 26% open offer, and delisting attempts carry their own premium dynamics.
A pair of recent deals is worth studying, on approximate headline terms you should check against the filings. Adani–Holcim (Ambuja/ACC, 2022, ~$10.5bn) priced near ~$80 per tonne of capacity, against an Indian replacement cost then estimated at ~$110 to 120/tonne. That's a control deal below replacement cost, and it told you something about seller motivation. Mars–Kellanova (2024, ~$36bn) went at roughly mid-teens EV/EBITDA including synergies, a strategic ceiling-setter for packaged food. You'll apply all this in Worked example 4.