Damodaran draws a line worth keeping for life. When you build a DCF, you value an asset. You estimate what it's worth from its own cash flows, growth, and risk, and you accept that your answer may disagree with the market for years. When you use a multiple, you price an asset. You look at what similar assets sell for and mark this one to that market. Pricing is a different activity, and no lesser one. A house buyer who checks the price per square foot of the last five sales on the street is pricing, and sensibly. But pricing assumes the street is sanely priced.
In 1999, Cisco looked reasonable against other networking stocks. The whole street was mad.
Relative valuation can only ever tell you that something is cheap or dear relative to its comparables. It can't tell you whether the comparables themselves make sense. That's the job of your DCF and reverse DCF, and it's why the phase capstone puts all three on one page.
Advertisement
Why use multiples at all?
So why bother with multiples at all? Four reasons hold up:
- Speed. A teardown gives you minutes. You won't get days. A well-read multiple gets you 80% of the pricing answer in 2% of the time.
- A check from the market. A DCF is a hall of mirrors for your own assumptions. A comps table makes you face what a hundred thousand other investors currently believe, and they're pricing similar businesses with real money.
- It's how the trade talks. Deals are negotiated, research is written and markets are discussed in multiples. If you can't speak it, you can't work.
- Reversibility. Every multiple compresses a DCF, so you can run it backwards. Put a multiple through the driver algebra and out come the market's implied assumptions. That's a five-minute reverse DCF.
What is the danger in relying on multiples?
The danger is just as real. Multiples let you sound rigorous while assuming everything. "It trades at 12× against a peer median of 15×, so it's 25% cheap" hides assumptions about growth, returns, risk, accounting comparability, and the sanity of the peer group, and nobody has examined any of them. The machinery for examining them is in valuation multiples.
Take the Gordon model from the terminal-value work, P0 = D1/(r−g), and divide both sides by any fundamental you like: earnings, book value, sales, EBITDA. What falls out is that every standard multiple equals a ratio of fundamentals. Each is some mix of growth, payout or reinvestment, returns on capital, margins, and the discount rate. No multiple has a "right" level that floats free as a market custom. Each one is a compressed DCF, and there are few enough compression formulas to memorize.