TVPI, DPI, RVPI and MOIC

Key takeaways

• DPI is cumulative distributions divided by paid-in capital, and it is the only one of the three fund multiples made entirely of cash that has already been received.
• RVPI is residual net asset value (NAV) divided by paid-in capital, and its numerator is a valuation the GP prepares, so it can move without any cash changing hands.
• TVPI equals DPI plus RVPI by construction, because all three ratios share the same denominator.
• Net fund multiples use paid-in capital that includes management fees and fund expenses, which is why net TVPI is always lower than the gross MOIC of the underlying deals.
• Over a fund's life the TVPI mix migrates from almost all RVPI to almost all DPI, so a young fund's TVPI is mostly a forecast and an old fund's TVPI is mostly a fact.
• A multiple ignores time: it cannot tell a fund that doubled money in three years from one that took twelve, which is why it is always read next to an IRR.

Paid-in capital: the shared denominator

All three fund multiples divide by the same number, paid-in capital. Paid-in capital is the cumulative amount the LPs have actually transferred to the fund in response to capital calls, measured from the first close to the report date. It is not the commitment (the total the LPs promised), and it is not invested capital (the amount the fund put into portfolio companies). It sits between the two.

Paid-in capital includes everything the LPs were called for. Most of it funds investments, but a meaningful share pays management fees, organizational costs and ongoing fund expenses such as audit, legal and administration. Where a fund calls capital to pay interest on a subscription line, that is paid-in as well. The limited partnership agreement (LPA) defines which costs the fund may call for, and the capital account statement shows the split.

Using this full figure is deliberate. The LP's economic question is 'what did I get back for everything I paid?', and fees are part of what it paid. A multiple that divided only by invested capital would describe the deals, not the LP's experience. That is why fund-level net multiples use paid-in capital, and why they are called net: they are net of the costs that were funded out of the same calls.

Two practical wrinkles are worth knowing now and are treated in depth in the next topic. First, some distributions are recallable, meaning the GP may call the money again, which can push cumulative paid-in above what a simple reading of the commitment suggests. Second, some reports compute LP multiples on contributions net of fees paid outside the fund. Always read the definitions page before comparing two funds' multiples.

DPI: cash in hand

DPI, distributions to paid-in, measures how much cash the fund has returned for each dollar paid in. Distributions are cumulative from inception: every dollar of sale proceeds, dividends, interest or recapitalization proceeds that the fund has passed to the LPs, after the GP's carried interest has been taken out. A DPI of 1.00x means the LPs have received back exactly what they paid in. Everything above 1.00x is realized profit.

DPI is the hardest multiple to argue with. The numerator is a sum of wire transfers the LP can reconcile to its own bank statements. It cannot be revised downward by a later valuation, a change of auditor or a market crash. That is why practitioners describe it as 'cash in hand' and why allocators who have been through a downturn often screen on DPI before anything else.

DPI does have limits. It says nothing about what remains in the fund, so a young fund with excellent companies will show a low DPI for structural reasons, not because it is failing. It also treats a distribution received in year three the same as one received in year twelve, even though the earlier cash is worth more to the LP. And a GP can raise DPI early by selling its easiest winners first, which may or may not be the best decision for total value.

RVPI: the GP's mark

RVPI, residual value to paid-in, measures what is still inside the fund. The numerator is the fund's net asset value attributable to the LPs: the fair value of the remaining investments, plus cash and other assets, minus liabilities, and in most reports minus the carried interest that would be payable if the fund were liquidated at those values. The result is divided by paid-in capital.

The critical fact about RVPI is who produces the numerator. For private investments there is no quoted price, so the GP estimates fair value, typically each quarter, under an accounting framework such as ASC 820 (US GAAP) or IFRS 13, and often following the IPEV Valuation Guidelines. Auditors review year-end values and many funds use third-party valuation agents, but the estimate still begins with the manager whose fundraising depends on it.

That makes RVPI a measure of expected value, not realized value. It can be accurate, conservative or optimistic, and the LP usually cannot tell which until the assets are sold. It can also change without anything happening to the business: if public comparable companies fall 20 percent, a GP applying a market multiples method may mark its holdings down even though the companies' earnings are unchanged.

TVPI: the sum of cash and marks

TVPI, total value to paid-in, adds the two together: what has come back plus what is still held, over what was paid in. Because DPI and RVPI share a denominator, TVPI is exactly their sum. It is the fund-level answer to 'how many times my money is this worth today?' and it is the multiple most often quoted in fundraising materials.

The appeal of TVPI is that it is complete. A fund that has sold nothing yet is not penalized for patience, and a fund that sold its best company early is not flattered relative to one that held on. The weakness is the mirror image: TVPI blends a hard number (DPI) with a soft one (RVPI) into a single figure, and the reader cannot see the blend unless both components are shown.

The honest way to read a TVPI is therefore always as a decomposition. A 1.8x TVPI made of 1.5x DPI and 0.3x RVPI is a mostly proven result. A 1.8x TVPI made of 0.2x DPI and 1.6x RVPI is a mostly unproven one. Both are legitimate numbers. They are not the same claim.

Ten years of Fund A: how the mix shifts

The table below follows Fund A, a hypothetical $100m fund, across its ten-year term. Capital is called heavily in years 1 to 4 (the investment period), more slowly afterward for follow-on investments and fees, and the last call lands in year 9. Distributions begin in year 3 with an early exit and accelerate in years 6 to 9 as the portfolio is harvested. NAV rises while capital is being deployed and value is being created, peaks in year 5, then falls as companies are sold and their value leaves the fund as cash.

Three patterns stand out. First, TVPI starts below 1.00x. In years 1 and 2 the fund has paid fees and expenses and has not yet written anything up, so total value is less than paid-in. This is the multiple's version of the J-curve. Second, the TVPI climbs steeply in the middle years and then flattens: from year 8 onward it barely changes, because the remaining value is mostly being converted rather than created. Third, and most important for an LP, the composition of TVPI turns over completely. In year 3, DPI is 0.08x of a 1.05x TVPI. By year 10, DPI is 1.51x of a 1.63x TVPI.

This migration is what makes a TVPI from year 3 and a TVPI from year 10 different kinds of information. The early number is mostly a GP forecast. The late number is mostly a bank statement. When comparing funds of different vintages, compare the mix, not only the total.

MOIC: the deal-level multiple

MOIC, multiple on invested capital, applies the same idea to a single investment or to a portfolio of investments before fund-level costs. The numerator is total value from the deal: realized proceeds already received by the fund plus the unrealized fair value of whatever is still held. The denominator is the capital the fund invested in that deal, including follow-on investments, but not the management fees or fund expenses that the LPs paid separately.

Because MOIC excludes fees, expenses and carried interest, it is a gross measure. It describes how well the GP picked and managed companies, which is exactly what an LP wants to know during due diligence on a track record. It does not describe what the LP received. That is the job of net TVPI.

MOIC is often split into realized and unrealized pieces, and the terms are used loosely. Realized MOIC sometimes means the multiple on fully exited deals only, and sometimes means proceeds received divided by the invested cost of all deals. Ask which. A deal can also be partially realized, for example after a dividend recapitalization, and then its realized proceeds and remaining value both count toward total value.

Gross MOIC versus net TVPI

The gap between a GP's gross deal multiple and the LPs' net fund multiple comes from three layers. First, fees and expenses: they are part of paid-in capital but were never invested, so they enlarge the denominator without adding to value. Second, carried interest: the GP's share of profits reduces what the LPs receive. Third, any other fund-level items, such as subscription line interest or broken-deal costs. The worked example below keeps the model simple so each layer is visible.

The size of the gap depends on the fee base, the fund's expense load, the carry terms and how successful the fund is. It is not a constant. A later topic in this section, on gross versus net returns, builds a full year-by-year bridge with a hurdle and catch-up and shows why the gap is proportionally larger for weaker funds.

Reading the multiples together

No single multiple is enough. A disciplined reading takes TVPI as the headline, decomposes it into DPI and RVPI, and then asks how old the fund is. A young fund should have a low DPI; the questions are about the quality of the marks. An old fund should have a high DPI; if it does not, the question is why the GP has not been able to sell what it still holds, and whether those marks would survive a sale.

Multiples also ignore time. Fund A's 1.63x after ten years and another fund's 1.63x after five years are the same multiple and very different results. That is the reason every serious report pairs multiples with an internal rate of return, covered later in this section, and the reason neither metric should be read without the other.

Finally, remember what the multiples cannot see: the LP's opportunity cost. A 1.5x TVPI looks like a gain, but if the same money in a public index over the same dates would have produced more, the private fund underperformed. Public market equivalent (PME) methods exist to answer that question and are covered elsewhere in the Institute.