Valuation & Cost of Capital

Every valuation is a forecast wearing a formula. The formula is the easy part.

One idea, three disguises

• Every valuation is the same statement: an asset is worth the present value of what it pays you. Discounted cash flow says it explicitly; multiples say it in shorthand; comparables say it by pointing at someone else who already said it.
• The disagreement is never about the formula. It is about the cash flows (a forecast) and the discount rate (a judgement about risk).
• The discount rate is where the market's price of risk enters. That is why the cost of capital, not the spreadsheet, is the thing worth understanding.

CAPM: the standard price of equity risk

The Capital Asset Pricing Model prices risk that cannot be diversified away — and only that risk, which is its whole point:

• Diversifiable risk earns nothing. If you can remove a risk for free (see diversification), no one will pay you to carry it. Only the market exposure β is compensated.
• The equity risk premium is not observable. Historical averages run ~4–6% over bonds depending on country and window; forward-looking estimates from dividend yields plus growth are usually lower. Any valuation is a hostage to which one you pick.
• β is estimated, unstable and mean-reverting. Practitioners commonly shrink raw regression betas toward 1 for exactly this reason.
• The model is empirically contested — low-beta stocks have historically outperformed what CAPM predicts, which is the origin of factor investing. It survives as a shared language, not as settled truth.

WACC: blending debt into the rate

A company is financed by both equity and debt, and interest is usually tax-deductible. The blended rate is what projects must beat:

• The tax shield is the reason debt looks cheap, and the reason it is only partly cheap: the deduction is worth τ × interest, and worth nothing to a company with no taxable profit.
• Leverage does not create value by lowering WACC alone. More debt raises the risk of the remaining equity, pushing r_e up — the Modigliani–Miller insight. What is left is the shield, minus the rising expected cost of distress.
• Use market values, not book values, for E and D. Book equity is an accounting residual; the market's number is the one the market discounts.
• The WACC is a hurdle, not a target. A project earning exactly WACC creates zero value — it has paid its financiers and nothing more.

Multiples: valuation compressed to one number

• Enterprise value is what the whole business costs, independent of how it is financed: market cap + net debt (+ minorities, pensions and other debt-like claims in practice).
• EV/EBITDA compares like with like across capital structures, which P/E cannot. Its weakness: EBITDA ignores the capital expenditure that keeps the assets alive, so it flatters capital-hungry businesses.
• Every multiple is a compressed DCF. A P/E of 20 asserts a set of growth and discount-rate assumptions; the multiple simply refuses to show them.
• Leverage moves equity multiples mechanically. Two identical businesses with different debt loads have the same EV/EBITDA and very different P/Es — and very different fragility.

Where valuations go wrong

• The terminal value is most of the answer. In a typical DCF, 60–80% of the value sits beyond the forecast horizon, governed by a growth rate assumed forever. Small changes there swamp everything in the modelled years.
• Growth above the discount rate is impossible in perpetuity — the formula divides by (r − g) and returns nonsense, which is the model correctly refusing an incoherent assumption. See the dividend discount model.
• Reverse-engineer instead. Rather than producing a price, take the market's price and solve for the growth it implies — then judge whether that is plausible. This is the single most useful habit in the discipline.
• Comparables import the market's mistakes. Valuing by peer multiple during a bubble reproduces the bubble with a citation.
• Precision is not accuracy. A value to two decimals built on a ten-year forecast is a rounding of a guess. Ranges and sensitivities are the honest output.

The full DCF, and where its weight actually sits

Everything above assembles into one calculation: forecast the cash flows, discount them, and add a terminal value for everything after the forecast ends.