What do IRR, TVPI, DPI, RVPI, MOIC, and PME mean?

The six measures, one at a time

For each: what it measures, what it does not, and what it can hide.

• IRR. The annualised rate of return implied by the timing and size of cash flows. Useful because it accounts for when money moved. It does not measure how much money was made, it is highly sensitive to early distributions, it can be flattered by a subscription line that delays capital calls, and in a young fund it is computed largely from unrealized marks and is correspondingly unstable.
• TVPI. Total value, meaning distributions plus remaining value, divided by paid-in capital. Useful as the headline of what a fund has produced so far. It says nothing about timing, and it treats a dollar of unrealized mark as equal to a dollar returned, which they are not.
• DPI. Distributions divided by paid-in capital. The only measure that is entirely cash and entirely realized. Useful because it cannot be argued with. It is late-arriving in venture, so a young fund's DPI is usually near zero and that is not a criticism of it.
• RVPI. Remaining value divided by paid-in capital. Useful as the explicit measure of what is still an opinion. It is exactly as reliable as the marks behind it, which is why the valuation basis matters more here than anywhere else.
• MOIC. Multiple on invested capital, typically at the investment level: current value plus proceeds, over cost. Useful for describing a single company or the portfolio gross. It ignores timing entirely, and at portfolio level it is a gross measure unless stated otherwise, so it is not what an investor received.
• PME. Public market equivalent, comparing a fund's cash flows to the same flows invested in a public index. Useful for asking whether the illiquidity was worth it. Cambridge Associates, which developed a modified version to address the negative net asset value problem in earlier methods, cautions in its benchmarking framework against apples-and-oranges comparisons, noting that IRRs should not simply be benchmarked against public time-weighted returns.

A hypothetical worked example, with every input stated

The numbers below are invented to make the arithmetic legible. They are not a benchmark, not a target, and not drawn from any fund. Every output follows only from the inputs stated.

• Hypothetical input: paid-in capital of 40 million dollars.
• Hypothetical input: distributions to date of 30 million dollars.
• Hypothetical input: remaining value of 45 million dollars.
• Hypothetical result, TVPI: 30 plus 45, divided by 40, is 1.88 times.
• Hypothetical result, DPI: 30 divided by 40 is 0.75 times. Three quarters of what investors paid in has come back as cash.
• Hypothetical result, RVPI: 45 divided by 40 is 1.13 times. That portion is still an estimate.
• Check: 0.75 plus 1.13 equals 1.88, so DPI plus RVPI equals TVPI, as it always does.
• Hypothetical result, MOIC on one position: 2 million dollars invested, now valued at 10 million, is 5.0 times on that investment. Note it says nothing about the other positions or about time.
• Hypothetical result, IRR, timing illustration one: if the 40 million had gone out in a single payment and 75 million came back exactly five years later, the annualised rate would be about 13.4 percent.
• Hypothetical result, IRR, timing illustration two: the same 40 million out and 75 million back, but after three years rather than five, is about 23.3 percent.

The last two lines are the whole argument for showing more than one measure. The multiple is identical in both cases and the annualised rate differs by ten percentage points, purely because of when the money moved. A manager quoting only IRR is emphasising speed; one quoting only TVPI is emphasising magnitude; an allocator wants both plus the DPI split.

Gross, net, fund level, and portfolio level

Four words that change what a number means, and the ones most often left off.

ILPA's performance template guidance draws the distinction precisely: fund level refers to the transactions occurring between the fund and its investors, while portfolio level refers to the cash flows between the fund and its investments. It also observes that in practice most private fund general partners present gross IRRs and MOICs using portfolio-level cash flows and net IRRs and TVPIs using fund-level cash flows.

That has a consequence managers routinely miss. Fund-level net calculations typically begin with the fund's first capital call and include the impact of any fund-level subscription facility, while portfolio-level gross calculations typically begin at the time of investment regardless of how it was funded. The two therefore start on different dates. A gross portfolio IRR and a net fund IRR are not two views of one number; they are two different measurements.

ILPA's guidance also describes how a net portfolio-level figure can be produced at all: by applying the ratio or spread between the fund-level net and fund-level gross metrics to the portfolio-level gross figure, which is a synthetic calculation rather than a directly observed one. A manager with no fund, and therefore no fund-level net, has nothing to apply that ratio to, which is why a pre-fund record cannot be converted into a net figure by any standard method.

What each measure can hide

Stated as the question an allocator asks when they see it quoted alone.

• A high IRR with a low DPI. How much of this is realized, and how much is annualisation over a short period on marked positions?
• A high TVPI early in a fund's life. What is the RVPI, and what is the basis of those marks?
• A strong MOIC on a named company. What did the rest of the portfolio do, and is this figure extracted from a larger set?
• An IRR quoted with a subscription facility in the calculation. What is it without the facility? ILPA's guidance requires that metrics calculated with facility impact be accompanied by metrics that remove it.
• Any gross figure. What is the net, over the same period, on the same methodology? United States marketing rules require net at equal prominence and in a format designed to facilitate comparison.
• A PME. Against which index, over which period, and using which methodology, since the choice of all three moves the answer.

Which measure matters when

The emphasis shifts with the fund's age and the manager's situation.

• A first-time manager with a pre-fund record. Investment-level multiples with the sample stated, plus realized and unrealized separated. Annualised rates over few positions and short periods invite more scepticism than they earn.
• A young Fund I. TVPI with the DPI and RVPI split shown, and the valuation basis for the unrealized portion.
• A mid-life fund. DPI starts to matter as the question shifts from what is it worth to what has come back.
• A fund seeking a re-up. DPI and PME, because the allocator is deciding whether the illiquidity was worth it against what public markets did over the same period.
• An established manager. Consistency across funds, which is a comparison of the same measures computed the same way over multiple vintages.

What these measures do not prove

None of them measures skill. All six are outputs of a portfolio, a market, and a period, and the same manager produces different numbers in different vintages without becoming better or worse. That is the reason allocators spend more time on attribution and process than on the numbers themselves.

The worked example above is hypothetical throughout. It is not a benchmark, a target, or a representation of what any fund has achieved or could achieve, and no figure on this page is drawn from a real fund.

This page is educational and general. It is not legal, tax, accounting, valuation, or investment advice. Performance calculation and its presentation may need review by an auditor, an administrator, or counsel before materials are used.