The Net Present Value (NPV) and the Internal Rate of Return (IRR) are two central criteria of the investment decision, but they are based on distinct financial logics. NPV measures absolute wealth creation: it is the sum of future cash flows discounted at the Weighted Average Cost of Capital (WACC), minus the amount of the initial investment. If the NPV is positive, the project creates value because it yields more than what investors demand to compensate for their risk.

IRR, on the other hand, is a relative performance measure. It is defined as the discount rate that cancels out the NPV. This is the intrinsic rate of return of the project. If this rate is higher than the WACC (the “hurdle” rate), the project is considered acceptable because it generates a return greater than the cost of the resources mobilized.

However, the IRR has a major conceptual flaw related to the assumption of reinvestment of intermediate flows. Mathematically, the IRR calculation implicitly assumes that the cash flows generated by the project over its lifetime are reinvested at the same rate (the IRR) until the end of the project. This assumption is often unrealistic, especially for highly profitable projects with high IRRs (e.g. 30% or 40%). It is unlikely that the company will systematically find reinvestment opportunities that offer such profitability.

Conversely, NPV assumes that intermediate flows are reinvested in the WACC, which is a much more conservative and realistic assumption: the firm can always, at a minimum, repay its sources of financing (debt and equity) or reinvest in assets with similar risk. This is why, in the event of a conflict between NPV and IRR (for example for projects of different sizes or durations), financial theory favours NPV. To compensate for the IRR shortcoming, it is recommended to use the Modified Rate of Return (MRR), which calculates the return by capitalizing the positive flows at the WACC and discounting the negative flows at the WACC, thus providing a more “realistic” and robust measure of profitability.