A DCF has two halves. The cash flow forecast gets weeks of work, management interviews and three scenarios. The discount rate usually gets an afternoon and a number copied from someone else’s website. That allocation of effort is backwards. On a long-dated asset, one percentage point of discount rate can move the value by 15-20%, which is more than most of the assumption debates a board spends its time on. The rate prices two things: the investor’s appetite for risk, and the uncertainty sitting in the cash flows. The higher it is, the lower the value. What follows is a list of the discount rate errors I keep running into, mostly from valuations in markets where the textbook inputs are hard to find.
The discount rate converts future cash flows into a present value. It represents the return an investor requires for carrying the risks of a particular stream of cash flows. The word to hold on to is “particular”. Different streams carry different risks, so they carry different rates.
People often use WACC and discount rate interchangeably. The discount rate is the general concept. The weighted average cost of capital is one specific discount rate: the blended return required by all providers of capital, debt and equity together. That makes it the right rate for one job, which is discounting enterprise-level free cash flows, the cash that belongs to everyone who funded the business.
The WACC formula itself is simple. The judgement sits in the components:
Kd is the easy half because it is observable. Take the rate at which the business can borrow today, not the coupon on debt raised three years ago, and apply the marginal tax rate. Weights should be at market values, ideally at a target capital structure rather than whatever the balance sheet shows on the valuation date. Ke is where the valuer’s perspective comes in, and where two competent professionals can land a full point apart. CAPM prices systematic risk through beta. The build-up approach stacks observable premia instead and needs no beta at all. Which of the two you should trust depends on whether a reliable beta exists, and in many of the markets I work in it does not. The last early-stage clean-energy valuation I reviewed in North America carried an 18.5% discount rate built from stacked premia over a 3.5% risk-free rate; the debate that decided the number was a single input, the commercialisation risk premium, and it moved the value by 22%.
This is the error I see most often. Each stream of cash flows has its own risk structure and the rate has to match it. If the cash flows you are discounting are the ones distributable to equity holders, after debt service, the right rate is Ke, not WACC. WACC is almost always lower than Ke, so discounting equity cash flows at WACC overstates equity value, and the error compounds through every year of the forecast. The check takes one line in a model review: free cash flow to the firm goes with WACC, free cash flow to equity goes with Ke. No exceptions.
Nominal cash flows carry expected inflation inside them and need a nominal discount rate. Real cash flows need a real rate. Mining is where real-terms models are most common, and also where this mistake is most common. A real forecast discounted at a nominal WACC undervalues the asset by roughly the inflation rate, every single year. The same test applies to currency. A discount rate built from dollar inputs, a US treasury Rf and a dollar market premium, belongs to dollar cash flows. Apply it to local-currency flows without adjusting for the inflation differential and you have built an error into the machinery of the model. No sensitivity table will surface it, because it is not an assumption anyone looks at.
Government bonds are the standard reference for Rf, but in practice valuers reach for 5, 10, 15, 20 or 30-year paper almost interchangeably, and each choice produces a different discount rate. The selected Rf should match the duration of the underlying cash flows. A going concern valued into perpetuity wants a long-term rate. A mine that runs out of ore in ten years wants something like a 10-year rate. When the yield curve is steep, the gap between the 5-year and the 30-year can be more than a full percentage point. On a perpetuity, that gap alone can change the conclusion of the whole exercise.
Beta is estimated by regressing a stock or a sector against the market, which assumes there is a deep, liquid market to regress against. In developed markets that holds. Across MENA, Africa and parts of Eastern Europe it does not. Thin trading and shallow exchanges produce betas that are statistically meaningless, often biased towards zero, which flatters the discount rate in exactly the places where risk is hardest to price. Models still get populated by copying an industry beta from Damodaran’s tables, built mostly from developed-market comparables, into a Cairo or Casablanca valuation. Where there is not enough liquidity to estimate a defensible beta, and that is the normal situation in Middle Eastern markets, the build-up approach is the better instrument. It replaces one unmeasurable input with a stack of premia that can each be sourced and challenged, and it limits how much judgement hides inside the WACC.
The same measurement gap distorts the country risk premium. In developed markets country risk is traded and observable. In frontier markets there is often no liquid sovereign debt or CDS market to read it from, so what cannot be measured simply gets assumed to be extreme. The second problem is that the CRP is routinely applied based on where a company is incorporated or listed rather than where it earns its cash flows. Both directions matter. A Lagos-listed exporter earning 80% of its revenue in euros does not deserve the full Nigerian premium. A London-listed company whose only producing asset sits in Algeria does not deserve none of it. Weight the CRP by the company’s genuine operational exposure to the country, through revenue, assets and cost base, and give credit for the part of the risk the business has effectively diversified away.
A subtler version of the same problem is pricing a risk in the discount rate and in the cash flows at the same time. If the forecast already haircuts revenue for expropriation scenarios, payment delays or convertibility restrictions, adding a full CRP on top charges the investor twice for the same exposure, and the valuation collapses for reasons nobody can trace to any single assumption. Every material risk should live in exactly one place. Either model it in expected cash flows and keep the rate clean, or leave the base case alone and price it in the premium. Decide which, write it down, and hold the model to it.
A discount rate assembled from six inputs deserves a cross-check from outside its own logic. Where possible, compute it under both CAPM and build-up and understand any gap between the two. Compare it against observed industry discount rates. The most useful check is to invert it: the implied capitalisation rate on terminal-year cash flow is a multiple, and you can hold that multiple up against where broadly comparable companies trade. If your DCF implies an exit at 14x EBITDA in a sector trading at 7x, look in your rate and your growth assumption, and find the problem before the investment committee does.
One structural observation from working across Africa and MENA. Investments there tend to be equity-heavy, and often all equity, because local debt markets are shallow, tenors are short and security packages are hard to enforce. As Wd approaches zero the WACC converges on Ke, and since no defensible beta exists, that Ke is in practice built through the build-up approach rather than CAPM. The whole valuation then rests on a stack of judgement-driven premia. The sensible response is to document every premium, source it and expose it to challenge, rather than distrust the exercise. There is also an opportunity in this. In markets where risk is assumed rather than measured, the investor who does the work of measuring it gets paid for other people’s blind spots.
The forecast shows you understand the business; the discount rate shows you understand its risk. Most valuations put nearly all of their effort into the first, and that imbalance, more than any single input, is the most common mistake of all.