Internal rate of return is the most widely cited single metric in private capital. It appears on every investment paper, every fund-level performance summary, and every secondary marketing document. It is also, on a careful reading, the metric most likely to flatter a transaction. None of this is news, and yet investment committees continue to anchor on it. A short note, then, on where the construction breaks and what should be read alongside it.
What IRR is, and what it is not
The internal rate of return is the discount rate at which the net present value of a stream of cash flows equals zero. It is, mathematically, a property of the cash flow profile. It is not a return in the colloquial sense; it is a single number that summarises a multi-period profile, and the summary is lossy in specific and predictable ways.
Three of those losses matter to the committee.
Reinvestment assumption
The construction assumes that interim distributions are reinvested at the IRR itself. In a high-IRR deal (anything above twenty per cent), this is almost never realisable. A 35 per cent IRR achieved through an early distribution, followed by a long hold on the residual, is not, in fact, a 35 per cent return on the committed capital. The modified internal rate of return addresses this, but is rarely shown.
Time sensitivity
IRR is more sensitive to the timing of cash flows than to their magnitude. A small early distribution can move the headline figure by several percentage points without changing the multiple of invested capital. This is the well-known phenomenon of "IRR engineering": subscription lines that defer the capital call, dividend recapitalisations that accelerate an early return of capital, and partial realisations that crystallise a flattering interim figure. None of these are improper; all of them inflate the reported IRR relative to the economic return.
Multiple solutions
Cash flow streams with more than one sign change can produce multiple mathematically valid IRRs. The convention is to report the one the spreadsheet finds first, which is not the same as the one that means anything. Development projects with reinvestment phases, leveraged structures with refinancing events, and infrastructure assets with major capex programmes all routinely produce multi-solution profiles. The committee paper rarely says so.
What is happening in practice
The drift towards IRR as the dominant metric has had two visible effects on investment committee practice.
First, deals with materially different risk profiles are being compared on a single axis. A 22 per cent IRR on a three-year clean realisation is not the same proposition as a 22 per cent IRR on an eight-year hold with three rounds of follow-on capital, and yet the committee paper compares them side by side as if they were. The risk-adjusted reading is not available without further work.
Second, the metric is being used to defend transaction structures that are themselves designed to optimise the metric. Subscription credit facilities, used judiciously, are a sensible treasury tool. Used to shift the J-curve, they convert a fund-level IRR into something closer to a marketing artefact. The limited-partner community has, to its credit, begun to notice. Net-of-line and gross-of-line IRRs now appear in better-disciplined reporting; they should be the default.
What to read alongside
None of this argues for abandoning IRR. It argues for refusing to read it on its own. Three companion figures should appear on every committee paper where IRR is cited.
- Multiple of invested capital (MOIC, or TVPI at fund level). This is insensitive to timing and gives the committee a clean read on the magnitude of return relative to the capital actually at risk. A 2.4x MOIC is a 2.4x MOIC regardless of whether it was earned over four years or eight.
- Duration of capital at risk. The weighted average period during which the investor's capital was committed. Two deals with the same IRR and the same MOIC are not the same deal if one held the capital for half the time of the other; the opportunity cost is different and the risk exposure is different.
- Direct alpha or public market equivalent. The return the same capital, deployed on the same schedule, would have earned in a chosen public benchmark. This is the figure that tells the committee whether the strategy actually outperformed the alternative use of capital, which is the question the committee is being asked to answer.
A note for sponsors
The sponsor that voluntarily reports MOIC, capital-at-risk duration, and a public market equivalent alongside the headline IRR is doing the limited partner a service and, over time, builds the relationship that produces the next fund commitment. The sponsor that reports only the headline figure is asking the limited partner to do the work of stripping out the noise, and a serious limited partner will do exactly that.
The metric is not the problem. The convention of reading it alone, on its own terms, without the companion figures that disclose its limits, is the problem.
What the committee should ask
Three questions, asked in the room, tend to surface most of the structural distortions:
- What is the IRR on a fully called, no-line basis, with the same realisation pattern?
- What is the MOIC, and over what weighted duration of capital at risk?
- What public benchmark, deployed on the same schedule, would have produced the same dollar outcome, and how does that compare?
Investment committees that ask these questions consistently end up with portfolios whose realised returns more closely match the returns underwritten at commitment. That is, after all, the point of the exercise.