1. Home
  2. Blog
  3. DPI vs TVPI vs RVPI

DPI vs TVPI vs RVPI

Total value to paid-in is cash already returned plus remaining NAV, each divided by capital paid in. A 1.8x that is mostly DPI is not the same fund as a 1.8x that is mostly a mark.

13 min read
Two stacks of unprinted paper on a zinc desk

Total value to paid-in (TVPI) is distributions to paid-in (DPI) plus residual value to paid-in (RVPI). DPI is cash the limited partner has already received, divided by capital paid into the fund. RVPI is the remaining net asset value, divided by the same paid-in figure. Add them and you have the fund's total multiple. A 1.8x TVPI with 1.4x DPI is not the same fund as a 1.8x TVPI with 0.3x DPI. The first has mostly paid. The second is still a mark.

Private equity reports these three because none of them is a rate. They do not care when the dollars moved. They care what has already been wired, what is still sitting in companies, and what those two add up to, per dollar actually called. Internal rate of return answers a different question. Multiple on invested capital answers a related one at the deal.

What DPI, TVPI, and RVPI measure

Distributions to paid-in is cumulative cash (and in-kind distributions valued at the date they went out) sent to limited partners, divided by cumulative capital those partners have paid in. A 0.40x DPI means forty cents of cash per dollar called. A 1.0x DPI means the fund has returned paid-in capital in cash. It is not yet a profit after time. It is capital back.

Residual value to paid-in is the fund's remaining net asset value divided by the same paid-in capital. Net asset value is the general partner's current mark on what is still held, net of fund-level accruals. Carta's RVPI note (6 October 2025) treats residual value as that net asset value: a fair-value estimate under the fund's valuation policy, often ASC 820 for US books. It moves when a company is written up, written down, or sold. It is not a wire.

Total value to paid-in is the two added together, which is the same as (cumulative distributions plus remaining net asset value) divided by paid-in capital.

TVPI = DPI + RVPI

The identity is the point. You never read the headline multiple without splitting it. A 1.6x TVPI that is 1.3x DPI and 0.3x RVPI has already put most of the claimed value in the limited partner's account. A 1.6x TVPI that is 0.2x DPI and 1.4x RVPI has not. The second fund can still be a good fund. It is a different claim.

On the usual limited-partner presentation, all three are net of management fees, fund expenses, and carried interest that has been paid or accrued. Gross versions exist. They answer a different question: what the deals did before the partnership took its share. If a pitch quotes one number, ask which.

The limited partner signs a commitment. That is a promise. The general partner then issues a capital call. Paid-in capital is the cash that has actually been called: money for investments, and money for management fees and partnership expenses. Uncalled capital is still at the limited partner.

The three multiples divide by paid-in, not by the commitment. A $100 million fund that has called $60 million and distributed $24 million has a 0.40x DPI ($24 million / $60 million). Divide by the $100 million commitment and you print 0.24x. That is a smaller number for a reason that has nothing to do with how the companies did. Early in the investment period the commitment is a poor denominator because most of it has not been used.

Paid-in includes fee calls. That is why a young fund often prints TVPI below 1.0x before any company has been marked down: cash has gone to the management company, and the remaining book has not yet grown enough to cover it. Carta's DPI explainer (Kyler Thomas, 14 July 2024) also notes that general-partner contributions to the fund do not count toward the limited-partner paid-in figure used in these multiples.

Recallable distributions cut the other way. If the partnership can call a distribution back, some reports net that recall against paid-in, and some show it gross. If DPI and RVPI do not add to TVPI on a statement, the statements are mixing those conventions, or mixing paid-in with committed.

A $100 million sketch

A reporting date, not a forecast. Commitment $100 million. Paid-in $60 million. Distributions $24 million. Remaining net asset value $54 million.

DollarsMultiple on $60m paid-inIf you divide by the $100m commitment
Paid-in$60,000,000
Distributions$24,000,000DPI 0.40x0.24x
Remaining NAV$54,000,000RVPI 0.90x0.54x
Total value$78,000,000TVPI 1.30x0.78x

DPI is 0.40x. RVPI is 0.90x. TVPI is 1.30x. Forty cents of each paid-in dollar is cash. Ninety cents is a mark. The 1.30x is the sum. The right-hand column is what you get if you put the unused $40 million of commitment in the denominator. It is not how the industry writes the score.

How the mix moves as the fund ages

Early in the fund, DPI is zero or close to it. Companies have not been sold. RVPI is almost the whole of TVPI, and TVPI is often below 1.0x because fees have been paid. That cash trough is the J-curve. A zero DPI in year two is how the vehicle is built. It is not a failed fund.

In the middle years RVPI usually peaks. Marks rise if the companies are worth more. A dividend recapitalization or an early sale can put a first distribution on the page. Most of the book is still illiquid.

In harvest, exits convert residual value into distributions. DPI rises. RVPI falls. TVPI moves only if the exit comes in above or below the last mark. At wind-down, residual value is supposed to be about zero, and TVPI equals DPI. The multiple that remains is cash.

A teaching clock on the same $100 million commitment. Paid-in, distributions, and net asset value are round numbers so the identity is visible. They are not a vintage forecast.

YearPaid-inDistributionsRemaining NAVDPIRVPITVPI
2$40,000,000$0$36,000,0000.00x0.90x0.90x
5$80,000,000$16,000,000$96,000,0000.20x1.20x1.40x
8$90,000,000$72,000,000$63,000,0000.80x0.70x1.50x
12$90,000,000$162,000,000$01.80x0.00x1.80x

Year 2 is fees and a first mark below cost. Year 5 is still mostly residual value. Year 8 has returned eighty cents on the dollar in cash and still holds seventy cents of book. Year 12 is all cash. The 1.80x at the end is the same 1.80x you could have quoted as TVPI in a year when most of it was paper. Only the mix changed.

A 0.20x DPI in year 5 or 6 is ordinary on a buyout clock. The same 0.20x in year 11 is not. A six-year buyout clock that prints 0.2x DPI next to a 1.5x TVPI is ordinary. The calendar is the context.

The same TVPI, two different funds

Hold the headline at 1.80x and change only the split.

Fund AFund B
DPI1.40x0.30x
RVPI0.40x1.50x
TVPI1.80x1.80x

Fund A has already wired $1.40 per paid-in dollar. Forty cents is still in companies. Fund B has wired thirty cents. A dollar fifty sits on the general partner's marks. Same TVPI. Different liquidity, different valuation risk, different story for the next fundraising.

Residual value is an opinion until a buyer pays. It can be a last round, a comparable, or a model. It can be conservative. It can be hopeful. Limited partners who have to fund new commitments from distributions cannot spend RVPI. Secondaries exist in part because someone wants to turn that residual into cash before the fund does.

Why limited partners now read DPI first

Bain & Company's 17th Global Private Equity Report (23 February 2026). Buyout funds sit on a record $3.8 trillion of unrealized value. Average holding periods at exit have drifted toward seven years. Distributions as a percentage of net asset value are well below historical norms. It is already harder, Bain writes, to produce both acceptable returns and strong distributions to paid-in.

That is why limited partners now split the TVPI before they talk about the rate. New commitments are partly paid from distributions on old funds. A mark does not recycle. Cash does. A high internal rate of return on delayed calls does not pay the next capital call at another firm.

The slogan going around the fundraising circuit is that DPI has replaced IRR. It has not. The rate still prices speed, and the preferred return in the partnership is still a rate. What changed is the order of the questions. Cash first, then the mark, then the clock. A general partner who sells a good company early only to print DPI has not done the limited partner a favor. The multiple still has to be earned.

On the general-partner side, the same split is a staffing fact. Associates live with marks. They build the model that supports the residual value. Carried interest waits on realizations. Fund III is judged, in this market, on whether Fund I and Fund II have wired cash. Open investing seats sit on Private Equity Jobs. The companies directory is where those firms are listed. The scoreboard in the quarterly is still DPI, RVPI, and TVPI.

TVPI, MOIC, and IRR

Multiple on invested capital is usually a deal-level ratio: realized and unrealized value divided by equity invested in that company. It is often quoted gross of fund fees. TVPI is a fund-level ratio: distributions plus remaining net asset value, divided by paid-in capital, usually net. Paid-in includes fee calls. Invested equity at the deal does not. The two line up when the fund is fully paid in and you are looking through to the same net-of-fee pile of value. In the middle of the investment period they do not, for reasons that have nothing to do with whether the deals worked.

Internal rate of return folds the calendar into a percentage. The multiples do not. A 1.8x that took four years and a 1.8x that took nine years share a TVPI and almost nothing else. Subscription facilities delay the limited partner's cash out, so the rate rises while the multiple barely moves, except for interest and unused fees. That is why the multiples belong next to the rate, and why a delayed call can make the rate look better than the cash.

The deal that produces most of these cash flows, on a buyout fund, is a leveraged buyout.

Gross, net, and what gets reported

Gross multiples and rates are deal performance before the fund's fees and carry. Net is what limited partners keep. Public benchmarks (Cambridge Associates, Preqin, PitchBook) quote net unless they say otherwise. Marketing decks sometimes quote gross. The gap is large enough to flip a comparison.

The Institutional Limited Partners Association's Performance Template (January 2025) is built for funds that commence operations on or after 1 January 2026. The fund-performance table asks for net internal rate of return and net TVPI both with and without the impact of fund-level subscription facilities. The pair exists because a facility changes when paid-in is recorded. With the facility, paid-in is later, so early IRR is higher and early TVPI can look larger on a smaller denominator. Without it, the limited partner is treated as if the call had come when the investment was made. Both belong on the page. One of them is not a substitute for the other.

Net asset value loans that fund distributions raise a separate question. A distribution that is borrowed against remaining marks can raise DPI without a sale. If the quarterly shows DPI jumping while remaining companies are unchanged, ask whether the cash came from an exit or from a facility.

Common mistakes

Using the commitment as the denominator. The industry score is paid-in. Commitment in the denominator makes a half-called fund look worse than it is, or, later, mixes unused dry powder into a performance ratio.

Reading TVPI as cash. TVPI includes the mark. Only DPI has hit the limited partner's account.

Comparing vintages without the clock. A 2018 fund at 1.5x TVPI and a 2023 fund at 1.5x TVPI are not peers. One is in harvest. One is still calling.

Mixing gross and net. A 2.2x gross MOIC next to a 1.6x net TVPI can be the same portfolio after fees and carry.

Treating a recap as the whole story. A dividend recapitalization raises DPI. It also leaves more debt on the company. The residual value has to be read with that stack.

Printing a "good DPI" as a single number. A 1.5x at year 12 and a 1.5x at year 5 are different claims. End-of-life DPI above 1.5x is a common rule of thumb, not a vintage quartile. Compare against the same strategy and the same year.

Common questions

What is a good DPI?

It depends on age and strategy. Early years are supposed to be near zero. Many buyout funds are still below 1.0x at year six. At the end of the fund, limited partners want DPI above 1.0x, which is capital back, and then some, which is profit. "Good" is a peer comparison for that vintage, not a universal 2.0x.

Can DPI be higher than TVPI?

Not on the usual presentation. Residual value is not negative in a solvent fund, so DPI cannot exceed TVPI. If a statement shows that, the numerators or the denominators are not the same pile of cash.

Is TVPI the same as MOIC?

When the fund is fully paid in, and both are net (or both are gross), they describe the same pile of value. Before that, TVPI divides by paid-in, which is smaller than the commitment and not identical to deal-level invested equity. MOIC is also often gross at the deal. Ask.

Does paid-in include management fees?

Yes. Capital called to pay fees and expenses is paid-in. That is one reason TVPI starts below 1.0x.

Why is DPI zero for years?

Because nothing has been sold, or because early proceeds were recycled into new deals rather than sent out. Zero DPI in the investment period is the vehicle. Failure is a fund that never distributes, or that distributes less than was contributed after the companies are gone.

Sources

Bain & Company, 17th Global Private Equity Report, 23 February 2026. Institutional Limited Partners Association, Performance Template, January 2025 (funds commencing on or after 1 January 2026). Carta, How Distributions to Paid-In (DPI) Works, Kyler Thomas, 14 July 2024, and RVPI explained, 6 October 2025.