IRR Bridge

What is an IRR bridge?

An IRR bridge is a decomposition of the change in a fund's internal rate of return between two measurement dates into named components that sum to the change. It starts with the prior period's since-inception IRR, adds and subtracts the effect of each driver, and ends with the current IRR, in the same waterfall form a revenue bridge uses.

What an IRR bridge is not

It is not the relationship between IRR and multiple on invested capital. Those two measure different things about the same cash flows, and a chart showing how a given multiple maps to an IRR at different holding periods is a sensitivity table, not a bridge. An IRR bridge is strictly a time series: two dates, one delta, and an account of what produced it.

It is also not an attribution of IRR by company. A per-company breakdown does not add up to the fund IRR, because IRR is not additive across cash flow streams. A bridge can attribute the change in IRR to events at particular companies, which is a weaker and more honest claim.

The components

Five drivers cover almost every real bridge.

• Valuation changes on unrealized positions, split into markups and markdowns rather than netted, because a bridge that nets them hides the two events a reader most wants to see.
• Realizations, measured against the carrying value the position held at the prior date. A sale above carrying value lifts IRR, a sale at carrying value moves it much less than people expect, and a sale below carrying value lowers it.
• The passage of time. This is the component most people omit and it is frequently the largest. Holding a mark flat for another year lowers a since-inception IRR mechanically, because the same gain is being annualized over a longer period.
• New capital deployed at cost. Fresh contributions invested and held at cost add weight to the denominator of the money-weighted calculation without adding gain, which pulls a positive IRR down.
• Fees and expenses charged in the period, on a net IRR bridge. Management fees and fund expenses are cash out with no corresponding value in, so they reduce net IRR every quarter. This is also where the gross-to-net distinction becomes a compliance matter: Rule 206(4)-1 prohibits any presentation of gross performance in an advertisement unless the advertisement also presents net performance.

Why time alone moves IRR: a worked example

Figures are hypothetical. A fund calls 10,000,000 dollars on January 1, 2024 and buys a single position. At December 31, 2025 the position is carried at 14,000,000 dollars, two years after the outflow. The since-inception IRR solves 10,000,000 times one plus r, raised to the power of two, equals 14,000,000. So one plus r squared equals 1.4, one plus r is the square root of 1.4, or 1.1832, and r is 18.3 percent.

A year passes. Nothing happens at the company, and the position is still carried at 14,000,000 dollars at December 31, 2026, now three years after the outflow. Now one plus r cubed equals 1.4, one plus r is the cube root of 1.4, or 1.1187, and r is 11.9 percent.

The IRR fell 18.3 minus 11.9, or 6.4 percentage points, with no change in value whatsoever. The multiple is 14,000,000 divided by 10,000,000, or 1.40 times, at both dates. Any bridge that cannot show this line is not a bridge.

A worked example: bridging 18.0 percent to 14.5 percent

Figures are hypothetical. A fund reports a net since-inception IRR of 18.0 percent at December 31, 2025 and 14.5 percent at December 31, 2026, a change of negative 3.5 percentage points.

• Opening net IRR: 18.0 percent
• Markups on two portfolio companies: plus 2.1 points, giving 20.1 percent
• Write-down of one portfolio company: minus 4.4 points, giving 15.7 percent
• One realization above carrying value: plus 0.9 points, giving 16.6 percent
• Passage of time on unrealized value held flat: minus 1.7 points, giving 14.9 percent
• New capital called and deployed at cost: minus 0.4 points
• Closing net IRR: 14.5 percent

Check the components on their own: 2.1 minus 4.4 is negative 2.3; plus 0.9 is negative 1.4; minus 1.7 is negative 3.1; minus 0.4 is negative 3.5. And 18.0 minus 3.5 is 14.5, which ties.

The sentence that bridge supports: one write-down at 4.4 points cost more than every markup and the realization combined at 3.0, and the two mechanical components, 1.7 points of elapsed time and 0.4 of new capital held at cost, account for 2.1 of the 3.5 point decline, which is arithmetic rather than news.

How to build one that ties

• Fix both endpoint IRRs from the official performance record first and never let the bridge change them.
• Run the calculation once per component, cumulatively, in a stated order. Recompute the full cash flow series with only the first change applied, then with the first and second, and so on. The difference at each step is that component's contribution.
• State the order, because the contributions are order-dependent. Reversing the sequence gives different individual figures for the same total, and a bridge that does not disclose its order cannot be reproduced.
• Force the residual into a named interaction line rather than smearing it across the components.
• Keep gross and net bridges separate. A gross bridge has no fee line; mixing the two produces a fee effect that appears twice or not at all.

ILPA's Performance Template exists partly because this arithmetic was being done inconsistently. ILPA describes it as developed to standardize return calculation methodologies by creating a framework for capturing performance metrics and the corresponding contributions and distributions, and publishes it in two versions, a Granular Methodology for managers who use fund-to-investor cash flows and itemize each capital call, and a Gross Up Methodology for managers who use fund-to-investment cash flows or do not itemize calls. Which version a manager uses changes the cash flow series and therefore changes every component of an IRR bridge built on it.

Where the subscription line goes

A fund-level credit facility delays capital calls, which shortens the apparent holding period and raises since-inception IRR without changing any multiple. So a bridge on a fund that uses a line has an additional component: the change in the facility's effect between the two dates.

The regulators and the standard-setters have both landed on the same answer, which is to show both figures. ILPA issued guidance on subscription lines of credit in 2017 and follow-on guidance in June 2020 recommending specific quarterly and annual disclosures so limited partners can monitor the impact of the lines on both exposure and performance. The Securities and Exchange Commission's 2023 private fund adviser rule would have required advisers to illiquid funds to disclose levered and unlevered returns, meaning with and without the impact of fund-level subscription facilities, after the Commission commented in the proposing release that levered returns often do not reflect a fund's actual performance and have the potential to mislead investors. The United States Court of Appeals for the Fifth Circuit vacated that rule on June 5, 2024, so the disclosure is now voluntary, but the analytical point survives the vacatur.

Common mistakes

• Omitting the time component, which makes every other line wrong by construction.
• Netting markups against markdowns into a single valuation bar.
• Attributing the whole IRR to individual companies and presenting the parts as if they summed.
• Bridging a gross IRR to a net IRR and calling it a period bridge. That is a gross-to-net reconciliation, a different exercise done at one date rather than across two.
• Presenting an IRR bridge without the accompanying multiple. If the multiple is flat while the IRR falls, the honest headline is that nothing happened and a year passed.

Adjacent terms

Internal rate of return is the figure being bridged, and its money-weighted, order-sensitive nature is exactly why the decomposition needs stating rather than assuming. Multiple on invested capital is the control: read alongside an IRR bridge it separates value creation from timing. A subscription line of credit is the component most likely to be missing from a bridge you are handed, and the one worth asking about first.