DPI Is the Only Number an Allocator Should Trust Right Now

For private equity funds of the 2018, 2019 and 2020 vintages, PitchBook's global benchmarks put median DPI at 0.56x, 0.33x and 0.16x as of March 31, 2025, against median TVPI of 1.66x, 1.53x and 1.38x and median net IRR between 13% and 16%. Most of the value those funds report is still a valuation mark. DPI is the one figure already settled in cash.

TVPI and IRR are not wrong, and the standards that define DPI define them too. They answer a different question: what a portfolio is worth on a manager's books and how quickly that value appears to have accrued. DPI records what has come back to the limited partner. When exits slow and holding periods lengthen, the distance between the two widens, and only the number counted in cash cannot widen with it.

01The Vintage Record

The table below sets out the public record for global private equity by vintage year. Every figure is net of fees and accrued carried interest and is taken from the quantile and pooled tables in PitchBook's Q1 2025 private equity benchmarks, with data as of March 31, 2025. PitchBook assigns a fund's vintage by the year of its first investment, falling back to the year of its final close. Medians describe the middle-ranked fund of each vintage. Pooled figures aggregate the cash flows and net asset values of every fund of the vintage, so large funds weigh more heavily.

Cambridge Associates is the other public benchmark of record, and it is absent from the vintage columns for a specific reason. Its free quarterly benchmark statistics publish index returns by horizon, not distribution multiples by vintage; the vintage-level detail sits in its paid reports. For the US private equity index, Cambridge reports a 10-year return of 14.99% and a 5-year return of 14.08% as of September 30, 2025, across a universe of 1,720 US funds. Those horizon returns rest on the same interim valuations that drive TVPI, and they say nothing about how much value has been paid out.

Three patterns stand out. The first is the median DPI column. Funds of the 2012 through 2016 vintages, roughly nine to thirteen years old at the measurement date, have returned more than their paid-in capital at the median. The 2017 median has not, at 0.83x, and every later vintage sits well below one.

The second is the stability of the IRR column. Median net IRR stays between roughly 13% and 18% for every vintage from 2012 through 2020, even as median DPI falls from 1.57x to 0.16x. The third is the spread between median and top-quartile DPI. For the 2019 vintage the top-quartile fund has returned 0.64x, almost twice the median, which means manager selection shows up in cash long before it shows up in a final multiple.

Growth equity, the strategy between venture capital and buyout, shows the same pattern in sharper form. PitchBook reports the growth and expansion subset separately, on smaller samples.

The 2017 growth vintage carries a median TVPI of 2.11x and a median net IRR of 20.55%, the strongest interim figures in the subset, while its median fund has returned 0.68x of paid-in capital. With 21 funds in the sample, a single strong or weak fund moves the median, so the figure is best read as direction rather than precision.

02The Arithmetic of Interim Returns

The three measures are defined together, which is why the gap between them can be read so exactly. The 2020 GIPS Standards for Firms define the realization multiple, DPI, as since-inception distributions divided by since-inception paid-in capital. PitchBook's methodology describes DPI as the capital distributed back to limited partners as a proportion of paid-in capital, "also known as the cash-on-cash" return, and states that TVPI is found by adding DPI and RVPI, the residual value to paid-in multiple. RVPI is the fund's reported net asset value divided by paid-in capital. It is the mark.

The pooled figures make the composition visible because pooled DPI and RVPI sum exactly to pooled TVPI. In the PitchBook tables, the 2018 vintage shows pooled TVPI of 1.56x, made up of 0.52x distributed and 1.04x still held at reported value. The 2019 vintage shows 1.55x, of which 1.15x is unrealized. The 2020 vintage shows 1.32x, of which 1.09x is unrealized. Put as shares, roughly two-thirds of the reported value of the 2018 vintage, three-quarters of the 2019 vintage and more than four-fifths of the 2020 vintage is valuation rather than cash.

The medians cannot be decomposed the same way, because the median DPI fund and the median TVPI fund need not be the same fund, but they point in the same direction.

Internal rate of return compounds the effect. A pooled IRR needs a terminal value, and the convention is to treat the remaining net asset value as if it were distributed on the measurement date. PitchBook's methodology states that in its pooled calculations, remaining unrealized value "is treated as a distribution in the most recent reporting period," which "explains why some vintages show high IRRs but low DPI values." The 2022 vintage illustrates the point at its limit. Its pooled net IRR is 13.08% while its pooled DPI is 0.07x. Nearly all of that return is the mark, discounted as though it had been received.

None of this makes a mark wrong. Many marks will be realized at or above their carrying value. The point is narrower. A mark is an estimate made by the manager under a valuation policy, and its accuracy is known only when an asset is sold. GIPS recognizes the distinction directly. For pooled funds that present money-weighted returns, provision 7.A.3 requires the percentage of total fair value valued using subjective unobservable inputs. That disclosure exists because the standard setter treats the share of value resting on judgment as material information in its own right.

03Subscription Facilities and Reported IRR

The second reason DPI has gained standing is that IRR responds to financing decisions that do not change what a limited partner ultimately receives. A subscription line of credit lets a fund buy assets with borrowed money and call capital from investors later. The shorter period between a capital call and a distribution raises IRR without adding a dollar of profit.

The Institutional Limited Partners Association documented the size of the effect in its guidance on subscription line disclosure, which states that these facilities "have the potential to increase time-sensitive return measures substantially and can thereby also distort fund quartile rankings." The guidance cites an analysis of 498 funds in which the median IRR increase was 206 basis points by year three, falling to 35 to 45 basis points by the end of the fund's life. The boost is largest early in a fund's life, when the record is shortest and the mark carries the most weight.

The reporting standards now require the two versions to be shown side by side. ILPA's performance template guidance asks general partners to report net IRR and TVPI both with and without the impact of fund-level subscription facilities. The same document gives ILPA's reading of the SEC Marketing Rule: whenever performance calculated with the impact of such facilities appears in advertising or marketing materials, it "must be accompanied by performance metrics that remove the impact."

GIPS provision 7.A.2 sets a parallel requirement for firms that claim compliance. A pooled fund that uses a subscription line must present its since-inception money-weighted return with and without the line, unless the principal was repaid within 120 days using capital drawn from investors and no principal was used to fund distributions.

DPI carries no time term, so the timing benefit that a facility gives to IRR does not flow through to it. A dollar distributed in year eight counts the same as a dollar distributed in year four. That indifference to timing is a limitation when the question is the speed of a return. It is an advantage when the question is whether the return has happened.

04Continuation Vehicles and the Quality of Distributions

DPI is not immune to engineering. It measures cash paid, and not every distribution is the product of a sale to an independent buyer. The most significant source of distributions that are not conventional exits is the GP-led secondary, in which a manager sells one or more portfolio companies from an older fund to a continuation vehicle that the same manager controls, financed by new secondary investors. Limited partners who elect to sell receive cash, which counts toward the selling fund's DPI. Limited partners who elect to roll their interest receive no distribution and carry the asset forward at the transaction price.

The price in that transaction is set inside a process the manager sits on both sides of. ILPA's continuation fund guidance addresses the conflict with three principles: rolling limited partners "should be no worse off than if a transaction had not occurred," "a true status quo option should always be offered," and the limited partner advisory committee should vote to waive the conflict.

The SEC adopted a rule in 2023 that would have required a fairness or valuation opinion for adviser-led secondaries, as described in its fact sheet on the private fund adviser reforms. That rule did not survive. The Fifth Circuit vacated the private fund adviser rules, including the adviser-led secondaries rule and the quarterly statement rule, on June 5, 2024, in National Association of Private Fund Managers v. SEC, No. 23-60471, as the SEC confirmed in an announcement on the private fund advisers rules. A fairness opinion in a continuation transaction is therefore market practice and a matter for the partnership agreement, not a federal requirement.

The practical consequence is a second question to ask of any DPI figure: where did the cash come from? Distributions from a strategic sale or a public offering reflect a price set by an independent buyer. Distributions from a continuation vehicle reflect a price negotiated in a conflicted process, even when that process is well run. Distributions funded by borrowing against the remaining portfolio are cash in hand but not realizations at all. All three raise DPI. Only the first resolves the uncertainty in the mark. The mechanics of these transactions are covered in the reference entry on secondaries.

05The Reporting Standard

The demand for cash-based evidence has a regulatory backdrop that is easy to misread. With the quarterly statement rule vacated, no SEC rule requires the standardized quarterly performance statement that rule would have mandated. What governs is the limited partnership agreement, the ILPA reporting and performance templates where a manager adopts them, and the GIPS Standards where a firm claims compliance. Provision 7.A.4 of the 2020 GIPS Standards requires a pooled fund presenting money-weighted returns to show paid-in capital, distributions, committed capital, the investment multiple (TVPI), the realization multiple (DPI) and the paid-in capital multiple. The standard places DPI beside TVPI by design. Neither is presented alone.

The market itself has begun to return cash. In its commentary on the first half of 2025, Cambridge Associates reported that US private equity funds distributed more capital than they called, $78.9 billion against $67.6 billion, and that the US private equity index returned 3.9% for the half, with growth equity at 4.9% against 3.6% for buyouts.

Cambridge also reported that venture capital funds have called 1.6 times more capital than they distributed since 2022. One half-year of positive net cash flow in buyout and growth does not close the gap in the vintage table. It does mean that the question of which managers can convert marks into cash is being answered now, fund by fund, and DPI is where the answer is recorded.

For an allocator, the reading is specific. Compare a fund's DPI with the median and top-quartile DPI of its own vintage, not with its TVPI, because funds of the same age have converted very different shares of value into cash. Read TVPI and IRR with and without subscription facilities, as ILPA and GIPS now expect. Ask for the share of net asset value resting on unobservable inputs, which GIPS treats as a required disclosure. And ask how much of the distributions came from independent exits rather than continuation vehicles or portfolio borrowing. Both multiples are explained in the reference entries on DPI and TVPI.

LeverVenture is an operator-led, mid-market growth equity firm investing across Biopharma & Therapeutics, Diagnostics & Precision Medicine, Devices & Robotics, Digital Health & Delivery, and Longevity & Neuro, with AI as the accelerant across all five. In that position between venture capital and buyout, value is created through operating work inside companies, and it is confirmed only when a company is sold to a buyer who sets the price. The growth subset of the PitchBook table shows how long that confirmation can take. The broader case for the strategy is set out in the growth equity thesis.

06Frequently asked questions

What is DPI in private equity?

DPI, or distributions to paid-in capital, is the cash a fund has returned to its limited partners divided by the capital they have paid in. The GIPS Standards call it the realization multiple. A DPI of 1.0x means investors have received back what they contributed, net of fees and carried interest.

Why can a fund show a strong IRR and a low DPI?

IRR treats the fund's remaining net asset value as if it were distributed on the measurement date, so unrealized marks count toward the return. A young fund with high marks and few exits will show a strong IRR and a low DPI. Subscription lines of credit can raise early IRR further without changing DPI.

Is a higher DPI always better?

Higher DPI means more cash has been returned, but the source matters. Proceeds from a sale to an independent buyer confirm a valuation. Proceeds from a continuation vehicle or from borrowing against the portfolio are cash, yet the price behind them was not set by an independent buyer. Read DPI alongside the source of each distribution.

Nothing in this piece is investment, legal, tax or accounting advice, and nothing in it is an offer to sell or a solicitation of an offer to buy any security.