Bottom-Up Beta from Comparable Companies

A regression beta needs years of stock-price history against a market index to mean anything, which is a problem for a company that's private, recently IPO'd, or simply too thinly traded for the regression to be statistically reliable. CAPM still needs a beta to produce a cost of equity, so analysts build one from the outside in: borrow the betas of public companies that do the same kind of business, and adjust for the fact that those companies don't share the target's exact debt load.

The three steps

1. Unlever each comparable's beta, removing the effect of its own debt:

$$\beta_{unlevered} = \frac{\beta_{levered}}{1 + (1-t)\times \frac{D}{E}}$$

In words: divide the observed equity beta by a factor that grows with how much debt, relative to equity, that comparable carries — more debt means more of its equity beta is really just leverage amplifying the same underlying business risk.

2. Average the unlevered betas across the comparable set. Because leverage has been stripped out, this average now represents pure business risk in that industry, comparable across companies with very different balance sheets.

3. Relever to the target's own capital structure:

$$\beta_{target} = \beta_{unlevered,avg} \times \left(1 + (1-t)\times \frac{D}{E}\right)_{target}$$

In words: put back in the effect of however much debt the target company actually carries (or is projected to carry), scaling the pure business-risk beta back up to reflect the target's own financial risk.

Worked example

Three public comparables have levered equity betas of 1.4, 1.1 and 0.9, with debt-to-equity ratios of 0.8, 0.3 and 0.1 respectively, and a common 25% tax rate.

Unlevering: Comp A: $1.4 / (1+0.75\times0.8) = 1.4/1.6 = 0.875$. Comp B: $1.1/(1+0.75\times0.3) = 1.1/1.225 \approx 0.898$. Comp C: $0.9/(1+0.75\times0.1) = 0.9/1.075 \approx 0.837$.

Average unlevered beta: $(0.875+0.898+0.837)/3 \approx 0.870$.

The target company plans to carry a debt-to-equity ratio of 0.5, at the same 25% tax rate. Relevering: $\beta_{target} = 0.870 \times (1+0.75\times0.5) = 0.870 \times 1.375 \approx 1.196$.

That 1.196 — not any single comparable's raw beta, and not a simple average of the three raw levered betas — is the beta that belongs in the target's own CAPM cost-of-equity calculation.

What this means in practice

This is the standard way to estimate beta for private companies, IPO candidates, and any division being valued on a stand-alone basis with a different capital structure than its parent. It also lets an analyst model how a company's cost of equity would change under a hypothetical recapitalization, simply by relevering to a different target D/E.