Internal Rate of Return
The internal rate of return (IRR) is the discount rate that brings the net present value of a cash-flow series to zero: the rate at which an investment just reaches break-even. It is read off the same cash-flow table as the net present value, but it asks the question the other way round. Where NPV fixes a rate and computes how much value the project creates in francs, IRR fixes the value at zero and looks for the rate that leads there. This single rate compares directly against the threshold an organisation requires of its investments, the hurdle rate: above it, the project creates value at the chosen cost of capital; below it, it destroys value. A percentage rather than an amount is at once the strength of IRR, it communicates and ranks effortlessly, and its weakness, it hides the scale of what is at stake.
Goal
The internal rate of return answers a single question: at what rate of return does this project exactly reach break-even, and does that rate exceed what we require of our capital. It is the green-light question. The net present value answers a neighbouring but distinct question, how much value in francs, and both are read off the same cash-flow table produced by cost-benefit analysis. IRR is the root of the NPV function, the particular discount rate at which that function equals zero. Everything IRR adds to NPV lies in converting the result into a percentage and in what that percentage makes possible.
What that percentage makes possible is twofold. First, it enables an accept decision by simple comparison: if the IRR exceeds the hurdle rate, the project clears the bar; if not, it does not. Second, it enables a ranking: between two solution approaches of the same duration, the one with the higher IRR is, under conditions, the better. This is the use BABOK puts forward explicitly, comparing two solution approaches over the same period and recommending the one with the higher internal rate of return.
The deliverable is a percentage per option, together with the hurdle rate it is measured against and a sentence on the cash-flow assumptions it follows from. An IRR delivered without its hurdle rate is a number with no verdict, and an IRR delivered without the cash flows that produced it cannot be verified.
Usage
When to use
- The cash-flow table already exists: the cost-benefit analysis is done, and a single percentage comparable to the hurdle rate is wanted rather than an amount.
- Ranking several approaches of the same duration and comparable scale: at similar horizon and outlay, the highest IRR points to the option to keep.
- Communicating to a decision-maker who thinks in rates: a percentage compares at once to the cost of credit or the expected return, where an amount in francs needs a frame to be interpreted.
- Conventional cash flows: an initial outlay followed by positive net benefits, a single sign change, guarantee a unique and reliable IRR.
When not to use
- Non-conventional cash flows, several sign changes: the solution may admit several IRRs or none, none is authoritative, switch to the net present value evaluated at the hurdle rate.
- Mutually exclusive options of unequal scale or duration: ranking by IRR can contradict the value created, decide with the net present value, which compares francs and not rates.
- No hurdle rate set: with no agreed cost of capital, IRR has nothing to measure against, set the hurdle rate first or fall back on the payback period in the meantime.
Description
IRR is the root of the NPV function
The whole mechanism sits in one equation, the net present value read another way. The NPV of a cash-flow series at a discount rate r is written NPV(r) = −I + Σ CFt / (1 + r)t, where I is the initial outlay and CFt the net cash flow of period t. IRR is the value of r for which this expression equals zero. BABOK states it in these exact terms: the IRR calculation is based on the interest rate at which the NPV is zero. There is therefore nothing to rediscount or to re-demonstrate about the time value of money, which is the proper domain of the net present value: IRR consumes the same formula and asks it a different question, solve for r instead of evaluating at a given r.
This difference of orientation has an immediate practical consequence. NPV is computed directly: the rate is known, the formula is applied. IRR, by contrast, cannot be computed directly beyond two periods, because the equation becomes a polynomial in (1 + r) with no convenient closed-form solution. It is found by successive approximation, a spreadsheet function (IRR, XIRR) or a manual interpolation between two trial rates, one giving a positive NPV, the other negative, the IRR sitting between the two. The one step that never changes is verification: the found rate is put back into the formula and the NPV is checked to fall back to zero. It is the check a mis-keyed spreadsheet fails first.
The decision rule and its precondition
The accept rule is BABOK's, unqualified: if the initiative's IRR is below the hurdle rate, the investment should not be made; if it exceeds it, the project is acceptable. The ranking rule follows: once the bar is cleared, between two approaches of the same duration, the higher IRR is preferred. BABOK states the equal-duration condition explicitly. It leaves a second one tacit, and this is the one that trips up beginners: the scale must also be comparable. Two projects of the same duration but very different outlay cannot be ranked reliably by their rate alone, because a high rate on a small sum can create less value than a modest rate on a large one. Ranking by IRR is safe only at similar duration and scale; the moment either of the two diverges, it is the net present value that decides.
The internal character and its limits
IRR is called internal because it depends only on the project's own cash flows. BABOK stresses this as a limit to know: IRR is internal to an organisation, it does not account for external influences such as inflation, fluctuating interest rates or a changing business context. An IRR of 12% is a nominal rate computed on nominal cash flows; it says nothing about the reality of those flows against an inflation that would erode them nor about a cost of capital that would rise over the project's life. That is precisely why the hurdle rate it is compared against must incorporate these elements: it is in the hurdle rate that the external context enters the decision.
Points to watch
IRR is quick to compute and inspires a confidence that runs past its domain of validity.
The first pitfall is that of non-conventional cash flows. The equation NPV(r) = 0 is a polynomial of degree n in (1 + r), and Descartes' rule of signs bounds the number of positive roots by the number of sign changes in the cash-flow series. An outlay followed by benefits, a single sign change, gives a unique IRR. But a series that turns negative again, a large decommissioning or restoration payment at end of life, a renewal investment mid-course, can give several. A series of CHF −100'000 in year 0, +CHF 310'000 in year 1 and −CHF 220'000 in year 2 has two sign changes and two mathematically valid IRRs: 10% and 100%. This is confirmed by setting x = 1 + r, which gives −100'000 x² + 310'000 x − 220'000 = 0, that is 10 x² − 31 x + 22 = 0, whose roots are x = 1.1 and x = 2. At 10% as at 100%, the NPV of this series is exactly zero. Neither rate is authoritative, and the exit is to drop IRR and evaluate the NPV at the organisation's real hurdle rate, which does give a single verdict.
The second pitfall is that of mutually exclusive options of unequal scale, where ranking by IRR contradicts ranking by value. Take a pilot and a full rollout of the same one-year initiative, to be compared against the 8% hurdle rate.
| Mutually exclusive option (1 year) | Investment | Year 1 cash flow | IRR | NPV at 8% |
|---|---|---|---|---|
| X, pilot | CHF 10'000 | CHF 15'000 | 50% | CHF 3'888.89 |
| Y, full rollout | CHF 100'000 | CHF 130'000 | 30% | CHF 20'370.37 |
These two options have the same duration, one year, so BABOK's rule would pass its duration test. It fails nonetheless, because the scale differs by a factor of ten. The tie-breaker is the net present value.
The third pitfall is the reinvestment assumption. Ranking by IRR implicitly assumes that each interim cash flow is reinvested at the IRR itself until the end of the project. A project computed at 12% therefore assumes the organisation will place each annual receipt back at 12%, which is rarely true: its real reinvestment opportunity is closer to its cost of capital. The modified internal rate of return (MIRR) corrects this bias by reinvesting each interim cash flow at the hurdle rate to the end, summing these amounts into a terminal value, then taking the n-th root of the ratio to the outlay. On option A (CHF 50'000 per year over three years, hurdle rate 8%), the terminal value is 50'000 × 1.08² + 50'000 × 1.08 + 50'000 = CHF 162'320, and the MIRR is (162'320 / 120'100)1/3 − 1 ≈ 10.56%, against an IRR of 12.00%. The MIRR falls below the IRR, exactly in the direction the critique announces: reinvesting at the realistic 8% rather than at the project's 12% lowers the annualised return. When a project's IRR comes out well above the cost of capital, it is the MIRR that gives the defensible number.
AI considerations
The gain on the mechanical part is clear. A language model connected to a spreadsheet builds and checks the cash-flow formula, writes the IRR or XIRR function and, above all, runs the verification the technique requires: put the found rate back into the formula and confirm the NPV falls to zero. It also catches what a hurried human overlooks, the number of sign changes in the cash-flow series, and flags at once a non-conventional series that calls for caution over the uniqueness of the IRR. It computes the MIRR effortlessly and places the two numbers side by side, IRR and MIRR, which make the reinvestment assumption visible. On this arithmetic ground, the machine is faster and more reliable than the end-of-workshop manual computation.
Three decisions escape it: the ones that make the analysis. The choice of the hurdle rate is a business judgement: it incorporates the cost of capital, the project's risk and the external context that IRR, by construction, ignores. The plausibility of the projected cash flows themselves is a question about this organisation and its market: an impeccable IRR computed on optimistic benefits is a correct number on wrong data. And the accept decision at the margin, when the IRR grazes the hurdle rate, engages an appreciation of risk that a gap of a few tenths of a point does not settle on its own. The model produces the percentage; it does not decide whether to go ahead.
Examples
A precision-engineering SME in the Vaud stretch of the Jura Arc is evaluating two approaches to automate incoming-goods quality inspection, done by hand today. Both run over the same three-year horizon. The company's hurdle rate is 8%, its blended cost of bank credit and equity.
Option A is a full machine-vision system: CHF 120'100 invested in year 0, a net benefit (labour saved and scrap avoided) of CHF 50'000 per year. Option B is camera-assisted sampling inspection: CHF 88'600 invested, a net benefit of CHF 35'000 per year, lower because it catches fewer defects.
| Year | Option A | Option B |
|---|---|---|
| 0 | −CHF 120'100 | −CHF 88'600 |
| 1 | +CHF 50'000 | +CHF 35'000 |
| 2 | +CHF 50'000 | +CHF 35'000 |
| 3 | +CHF 50'000 | +CHF 35'000 |
Finding the IRR means looking for the rate that zeros the NPV of each column. It is 12.00% for option A and 9.00% for option B. Verification consists in discounting the three benefits at that rate and confirming that their sum equals the outlay.
| Year | Cash flow, option A | Discount factor at 12% | Present value |
|---|---|---|---|
| 1 | CHF 50'000 | 0.89286 | CHF 44'642.86 |
| 2 | CHF 50'000 | 0.79719 | CHF 39'859.69 |
| 3 | CHF 50'000 | 0.71178 | CHF 35'589.00 |
| Present value of benefits | CHF 120'091.55 | ||
| Year | Cash flow, option B | Discount factor at 9% | Present value |
|---|---|---|---|
| 1 | CHF 35'000 | 0.91743 | CHF 32'110.09 |
| 2 | CHF 35'000 | 0.84168 | CHF 29'458.80 |
| 3 | CHF 35'000 | 0.77218 | CHF 27'026.41 |
| Present value of benefits | CHF 88'595.30 | ||
Both IRRs exceed the 8% hurdle rate: both investments are acceptable. Since the duration is identical, three years, and the scale similar, the ranking rule applies with no risk of conflict, and option A wins, its IRR of 12.00% exceeding option B's 9.00%. The net present value confirms the same order, without contradicting it: at 8%, the NPV of option A is CHF 8'755 against CHF 1'598 for option B. Ranking by IRR and ranking by value coincide because neither scale nor duration diverges, the situation where IRR is a safe ranking criterion.
Visualizations
The NPV profile is the representation that shows what IRR is. A bar chart of two percentages gives the ranking, it says nothing of its meaning or its fragility. The curve, by contrast, carries four readings in a single image.
- The crossing of the zero axis is the IRR: it is the definition made visible, the discount rate where the NPV passes from positive to negative.
- The vertical hurdle-rate line settles acceptance: a curve that cuts zero to the right of this line clears the bar, a curve that cuts it to the left fails.
- The slope of the curve at its crossing tells the sensitivity: an NPV that plunges steeply around the IRR signals a project whose verdict changes quickly with the rate, so one to examine more closely if the cost of capital is uncertain.
- A second crossing of the same curve makes the non-conventional-cash-flow pitfall immediately visible: two passes through zero, two IRRs and the sign that the percentage alone is no longer enough.
The cash-flow table and the verification tables are the other half of the visualization. The first shows the series that feed the calculation, the others show the one operation that is authoritative, putting the rate back into the formula and seeing the NPV fall to zero. Together, the curve and the tables state the definition, the verification and the decision.
Cost
| Phase | Level | Rationale |
|---|---|---|
| Preparation | Moderate | IRR has no preparation cost of its own when its two inputs already exist: the cash-flow table from cost-benefit analysis and an agreed hurdle rate. The cost shifts there, onto these two prerequisites. Building a credible cash-flow table is a piece of work in its own right, and setting a hurdle rate is a financial-policy decision that IRR presupposes without producing. When these two elements are in place, the preparation specific to the technique is low. |
| Execution | Low | A spreadsheet function (IRR, XIRR) gives the rate instantly once the cash flows are entered, and verification fits in one line. The only real effort is checking the sign changes and, where needed, computing the MIRR, a few minutes more. |
| Documentation | Low | The result is a percentage per option, the hurdle rate it compares against and the cash-flow assumptions behind it. A short note suffices, provided the link to the source cash-flow table is kept, without which IRR becomes an unverifiable number. |
Tooling
The spreadsheet is the default tool, and it suffices in almost every case. Excel, Google Sheets and LibreOffice Calc carry the functions IRR (cash flows at regular intervals), XIRR (irregular dates) and MIRR (modified internal rate of return, with an explicit reinvestment rate). All three do in one formula what manual calculation does by approximation, and the MIRR function is the only simple way to make the reinvestment assumption explicit rather than borne unseen.
The NPV-profile chart, drawn in the same spreadsheet, deserves to be known as a diagnostic tool as much as an illustration. By computing the NPV for a range of rates and plotting it as a curve, one sees at a glance the crossing, hence the IRR, and above all detects a second crossing, hence a multiple-IRR case that a spreadsheet function would signal at best by an error message or, at worst, by returning only one of the roots depending on the seed rate supplied.
The financial-planning and project-appraisal software with investment-analysis modules takes over for a portfolio of options to compare and track over time, where maintaining dozens of spreadsheet models becomes a risk in itself. Beyond that, IRR calls for no specialised business-analysis tooling: it is read off the same cash flows as the net present value and cost-benefit analysis, and it is those techniques that carry the tooling cost.
Sources
- IIBA, A Guide to the Business Analysis Body of Knowledge (BABOK Guide) v3, §10.20 Financial Analysis: the definition of IRR as the rate at which an investment breaks even, the rule of comparison to the hurdle rate, the rule of ranking options of the same duration, the internal character of the measure and its formula (the rate at which the NPV is zero).
- Aswath Damodaran, NYU Stern School of Business, Capital Budgeting: IRR as the discount rate that zeros the NPV, the accept rule (accept if IRR above the discount rate), the problem of multiple IRRs on non-conventional cash flows, the reinvestment assumption and the modified internal rate of return (MIRR) as its corrective.
- Project Management Institute, PMBOK Guide, Eighth Edition, §2.4.1 Finance Performance Domain: IRR as one of the financial value measures named by PMI, alongside return on investment and payback period.