Command Palette

Search for a command to run...

Blog

Five Mistakes I See in Almost Every DCF

A discounted cash flow model is only as honest as its assumptions. These five errors make it lie.

In my first months of the MBA, I built and reviewed a lot of discounted cash flow models. The same mistakes appeared again and again, in student work and in published research alike.

1. Growth without reinvestment

A model that grows revenue at fifteen per cent a year while capital expenditure stays flat is describing a company that does not exist. Growth has to be paid for with capex, working capital or both. If reinvestment does not match growth, the free cash flow is fiction.

2. A terminal value that is the whole valuation

When seventy or eighty per cent of the value sits in the terminal year, the model is really a bet on one perpetual growth rate. That is fine if you say so. It is dangerous if the explicit forecast years create a false sense of precision.

3. Mixing nominal and real

Rupee cash flows that include inflation must be discounted at a rate that includes inflation. Mixing a nominal forecast with a real discount rate, or the other way round, quietly inflates value.

4. A discount rate chosen to hit a number

If the cost of capital moves every time the answer looks wrong, the model is a justification, not an analysis. Fix the rate first, with its reasoning written down, and then look at the result.

5. No sensitivity table

A single point estimate hides how fragile it is. A simple grid of value against growth and discount rate tells the reader far more than the headline number. Often the honest conclusion is a range, and the decision depends on which side of a threshold the range falls.

A quick self-check before you share a model

Before sending any DCF to a reviewer, I now run five questions against it:

  • Does reinvestment rise with growth, and is return on new capital plausible?
  • What share of value comes from the terminal year, and is that share stated openly?
  • Are cash flows and discount rate both nominal, or both real?
  • Was the discount rate fixed before looking at the output?
  • Is there a sensitivity table, and does the decision change anywhere inside it?

It takes ten minutes and catches most of the problems a reviewer would otherwise find.

What good models have in common

The best models I have reviewed share a few traits. Assumptions sit on one sheet, labelled and sourced. Each forecast line has a short note explaining its driver. The terminal value is cross-checked against an implied exit multiple, so the reader can see whether it is reasonable. And the conclusion is stated as a range with the key sensitivities named, rather than a single figure presented with false confidence.

A DCF is not a machine that produces a price. It is a structured way of writing down what you believe about a business, so that someone else can disagree with you precisely.

Command Palette

Search for a command to run...