IRR

How IRR works

IRR is the single annualized rate that makes a series of dated cash flows worth exactly zero today. It is defined implicitly, by an equation rather than by a recipe.

Sum over all periods t of CF(t) / (1 + IRR)^t = 0

Cash out of the investor's pocket, meaning capital calls, enters as a negative CF. Cash back, meaning distributions, enters as a positive. The fund's remaining net asset value enters as a final positive flow on the measurement date, which is why a since-inception IRR on a live fund is partly a statement about unrealized marks.

The Institutional Limited Partners Association defines IRR as the discount rate at which the present value of future cash flows equals the cost of the investment. Cambridge Associates describes it as the standard measure of returns for private investments, based on cash-on-cash returns over equal periods and modified for the residual value of the investment, and notes the argument for it: dollar weighting the cash flows properly accounts for a manager's ability to draw down capital at its own discretion.

That last point is the whole reason private markets use IRR instead of a time-weighted return. A limited partner does not choose when its money is at work; the general partner does. A money-weighted measure holds the manager responsible for that timing. It also means IRR is sensitive to timing in ways a multiple is not. The same 2.0x gross multiple is a 25.99 percent IRR if it lands in three years and a 10.41 percent IRR if it takes seven.

Several variants travel under the same three letters:

• Gross IRR is computed on fund-to-investment cash flows, before management fees, fund expenses and carried interest. Net IRR is computed on fund-to-investor cash flows, after all of them. The gap is not a rounding difference.
• Since-inception IRR runs from the first capital call to the measurement date. End-to-end IRR, in Cambridge Associates' definition, is an IRR over a specified window that takes beginning net asset value, the period's cash flows and ending net asset value.
• Pooled IRR aggregates the cash flows of many funds and solves once, so large funds dominate. A median IRR ranks individual fund IRRs and takes the middle. They answer different questions and rarely match.

IRR also has known failure modes. A cash flow series that changes sign more than once can admit more than one mathematically valid solution. Very early distributions on small amounts of drawn capital can produce enormous percentages that mean very little. And because the calculation discounts by elapsed time, the denominator of "time" can be shortened deliberately.

Worked example

Suppose a limited partner's cash flows with a single fund, in millions of dollars, are: 5.0 called at the start of year 0, 3.0 called at the start of year 1, 4.0 distributed at the end of year 3, and 12.0 distributed at the end of year 5, at which point the position is fully realized. All figures are hypothetical.

Step one: write the equation. Minus 5.0, minus 3.0 discounted one year, plus 4.0 discounted three years, plus 12.0 discounted five years, set equal to zero.

Step two: solve for the rate. There is no closed form; a spreadsheet or solver iterates. The answer is 18.53 percent.

Step three: verify it. At 18.53 percent, the discount factors are 1.1853 for one year, 1.6653 for three, and 2.3396 for five. The present values are minus 5.0000, minus 2.5310, plus 2.4020, and plus 5.1291. Those sum to approximately zero, which is the definition being satisfied.

Step four: compare it to the multiple. Total in is 8.0, total out is 16.0, so the multiple is exactly 2.0x. But because half the capital went in a year late and a quarter of the proceeds came back two years early, the IRR is 18.53 percent rather than the 14.87 percent a flat 2.0x over five years would produce.

Step five: change one thing. Move the 12.0 distribution from year 5 to year 4 and nothing else. The multiple is unchanged at 2.0x. The IRR rises above 22 percent. Nothing about the businesses changed; only the calendar did. That sensitivity is the metric's main virtue and its main vulnerability.

Where it shows up

In the quarterly report, IRR appears as since-inception net IRR to limited partners, usually next to gross IRR, DPI, RVPI and TVPI. ILPA's reporting standards and its more recent performance template exist to make those figures comparable between managers; the performance template explicitly distinguishes a granular approach using fund-to-investor cash flows from a gross-up approach using fund-to-investment cash flows.

In the limited partnership agreement, the mechanism that most often turns on an IRR is the preferred return. ILPA's Principles state that the preferred return should be calculated from the date capital is called from limited partners to the point of distribution, and that where capital is drawn from a subscription facility collateralized by uncalled commitments, it should be calculated from the date capital is actually at risk rather than the date it is finally called.

That guidance exists because a subscription line shortens measured elapsed time and lifts reported IRR without improving any investment. ILPA's Principles say such lines should serve the partnership rather than chiefly enhance reported IRR to accelerate carry, should be short, for example no more than 180 days, limited to a maximum percentage of commitments such as 20 percent, and should not fund early distributions.

In marketing materials and the Form ADV brochure, IRR is regulated performance. The Securities and Exchange Commission's marketing rule requires that any presentation of gross performance be accompanied by net performance with at least equal prominence, over the same period and using the same return type and methodology, and the staff's guidance names internal rate of return among the metrics that count. It also states that gross and net IRR cannot use different methodologies, and in particular that the effect of a subscription facility must be treated consistently in both.

In benchmark books, IRR appears as pooled, arithmetic mean, median and quartile-break figures within a vintage year. Cambridge Associates describes pooled returns as calculated on the aggregate of all cash flows and market values reported by managers, net of management fees, expenses and carried interest.

Common mistakes

• Comparing a gross IRR to a net IRR. This is the single most common apples-to-oranges error in fund diligence.
• Reading an early-life IRR as a forecast. In the first two or three years the number is dominated by fees and unmarked holdings and can be deeply negative, which is the J-curve, not a verdict.
• Ignoring the subscription line. A fund that delays calls and pays distributions from a facility can report a materially better IRR on identical investments.
• Treating IRR as achievable on the full commitment. IRR is earned only on capital actually at work; uncalled commitment sitting in Treasuries earns something else entirely.
• Confusing pooled and median benchmark figures. A fund can beat one and lose to the other in the same vintage.
• Assuming IRR is unique. With multiple sign changes in the cash flow series, more than one rate can satisfy the equation.
• Quoting IRR without the multiple. A very high IRR on a small amount of capital returned quickly can coexist with a mediocre multiple, and vice versa.

Related terms

IRR supplies the timing that a multiple leaves out, so it is read against TVPI, DPI and MOIC, and its early-life shape is the J-curve. Comparisons belong inside a vintage year. The metric drives real money through the hurdle rate and the distribution waterfall, which is where carried interest is earned, and it is the number limited partners scrutinize most closely when a general partner comes back to raise the next fund.

Frequently asked questions

What is a good IRR for a venture capital fund?

The only defensible answer is relative: compare against funds of the same strategy and the same vintage year, using published quartile breaks, and use net IRR to limited partners. An absolute target quoted without a vintage and without a gross-or-net label carries no information, because entry pricing and exit conditions differ enormously between vintages.

What is the difference between gross IRR and net IRR?

Gross IRR is calculated on the cash flows between the fund and its investments, before management fees, fund expenses and carried interest. Net IRR is calculated on the cash flows between the fund and its investors, after all of those. The difference is the cost of the fund structure. When an adviser advertises gross performance, the marketing rule requires net to be shown with equal prominence, over the same period and on the same methodology.

How do you calculate IRR for a private fund?

Build a dated list of every capital call as a negative amount and every distribution as a positive, add the current net asset value as a final positive amount on the measurement date, and solve for the rate that sets the present value of the whole series to zero. Spreadsheets do this with an iterative function; the daily-dated version is the one most administrators use, since private fund cash flows do not land on neat annual boundaries.

Why is IRR criticized as a private markets metric?

Because it can be improved without improving anything. Delaying capital calls behind a credit facility, distributing early from borrowings, and choosing when to realize assets all move the rate. It also assumes nothing about what happens to distributed cash, ignores the drag of holding uncalled commitments in liquid assets, and can produce more than one solution. None of that makes it useless; it makes it a number that has to be read with a multiple and with the cash flow schedule behind it.

What is the difference between IRR and MOIC?

MOIC is total value divided by invested capital, with no reference to time. IRR is the annualized rate implied by when each dollar moved. A 3.0x over ten years is an 11.61 percent IRR; a 1.5x over one year is 50 percent. Managers with different strategies can look better on one measure than the other, which is why serious diligence requires both.

Does a subscription line of credit change IRR?

It can, substantially. Drawing on a facility and calling limited partner capital later shortens the measured period during which investor money was outstanding, which raises since-inception IRR while leaving the multiple untouched. ILPA's Principles recommend calculating the preferred return from the date capital is at risk rather than the date of the eventual call, precisely to neutralize this effect in the economics.