| Method | Use when | Breaks when |
|---|---|---|
| DCF | Cash flows are forecastable for 5-10 years: mature business, stable margins, predictable capex (think Coca-Cola, Costco, utilities) | Early-stage companies (no profits yet), deeply cyclical businesses (next-year earnings could be peak or trough), or terminal-value-dominated stories where TV is 80%+ of total value |
| Multiples (P/E, EV/EBITDA, P/S) | Public peers exist with similar growth and capital intensity; you need a sanity check or a quick screen | No real peers, accounting differences between peers distort the comparison, or peers themselves are mispriced (a sector-wide bubble) |
| Net Asset Value (NAV) | Asset-heavy businesses where book value approximates economic value: REITs, BDCs, banks (with adjustments), closed-end funds | Asset-light businesses (software, consulting) where most value sits in intangibles the balance sheet doesn't capture |
| Replacement Asset Value (RNAV) | Capital-intensive businesses where the book value of plant/property is stale: pipelines, ports, telecom towers, steel mills | When 'replacement cost' is itself ambiguous (no one has built a new one in 30 years), or when the market for those assets is illiquid |
| Sum of the Parts (SOTP) | Conglomerates and holding companies where the units have different economics (Berkshire, IAC, LVMH) | When inter-segment transfers and shared corporate costs aren't disclosed cleanly enough to value each part; you also need a holding-company discount |
A useful rule of thumb: start with the method that requires the FEWEST assumptions you can't defend. For a mature consumer-staple with 10 years of stable margins, a DCF needs growth, discount rate, and terminal value -- all defensible. For a pre-revenue biotech, those same inputs are hallucination. A revenue multiple of comparable trials in the same phase, or an option-pricing model on the drug pipeline, both rest on a smaller set of assumptions. The valuation method is downstream of the business's earnings profile, not upstream of it.
Multiples are seductive because they're fast: one number, one peer set, done. The trap is that they import every assumption baked into the peer set without showing the work. If peers are mispriced (sector bubble) or if your target has structurally different economics (better margins, lower capex), the multiple gives you a confidently wrong answer. A DCF forces you to write down growth, margins, and discount-rate assumptions; a multiple lets you skip them, and the assumptions don't go away, they just hide.
Match the method to the business. DCF for predictable cash flows. Multiples for sanity-checking and screening when real peers exist. NAV and RNAV for asset-heavy. SOTP for conglomerates. Triangulate when you can; investigate the gaps when the methods disagree. The biggest mistake is using one tool for everything because it's the tool you know.
Sit with the ideas.
A holding company owns three businesses: a mature industrial unit (steady cash flow), an early-stage software unit (no profit yet, fast revenue growth), and a real estate portfolio. What is the most appropriate valuation method for the parent?