What pension plans disclose about their private-market funds

Pensions: all tracked plans · Australian super in numbers · Private market performance · the PE statistics · what plans disclose · where they diverge

US public pension plans publish fund-by-fund private-market performance — but not all of them publish the same fields, or as recently. Some give you the net IRR, the multiple and every cash flow for each fund; others give the cash flows but leave you to work out the return. This page reads 31 plans' own reports and shows, plan by plan, exactly what each one puts on the table at fund level — then runs two screens the cash flows make possible: which funds look great on paper but have handed little back, and which have actually paid out.

What each plan publishes, fund by fund

Each plan-lane below lists funds with the money committed. The columns are the share carrying a fund-level net IRR (the yearly return after fees), a net multiple (dollars back per dollar in), and the cash flows (money contributed and distributed) — plus whether it labels each fund's vintage year, and its overall DPI (cash returned per dollar in, across the whole portfolio). The net-IRR share counts funds vintage 2024 or earlier (funds committed in the last year or two are legitimately too new to show a meaningful IRR — the J-curve).

The "through" line under each IRR share is the key: it is the freshest vintage that plan publishes a per-fund IRR for most of that year's funds (a stray later mark or two does not move it). That is where a lower share comes from — not secrecy, but a marking policy. CalPERS publishes IRRs through vintage 2020 and marks everything newer "not meaningful", so its share reads lower; CalSTRS publishes through 2024. Both are legitimate — the "through" column is what tells them apart from a plan that simply publishes less.

A few plans leave out a whole field — and because it is the whole field, not a J-curve dip, it shows up as a flat gap: New Jersey DOI publishes the money in and out of each fund but no per-fund net IRR at all — the return has to be worked out from the cash flows; Colorado PERA publishes a net IRR but no per-fund multiple; CCCERA (Contra Costa County) does not label each fund's vintage year (shown as — in the IRR column).

¹ Share of funds vintage 2024 or earlier carrying a fund-level net IRR; the "through" line is the freshest vintage the plan publishes an IRR for most of that year's funds. — = the plan labels no vintage years. ² Whole-portfolio DPI: total distributed ÷ total contributed — how much cash the plan's private-markets book has actually returned per dollar in.

One field these reports do not break out per fund is fees: the management fee and carried interest are already taken out of the net IRR the plan reports, and the reports publish no separate per-fund fee line — so there is no honest fee-disclosure column to score, and inventing one would measure our reading of the report rather than the report itself. Where a plan does not publish a field, the column above says so.

Great on paper vs actually paid out — two screens on the cash flows

A fund can report a high net IRR while having returned very little actual cash. The measure of cash back is DPI — distributions ÷ contributions, dollars paid out per dollar called in. A DPI of 0.3 means 30 cents back on the dollar so far; the rest of the reported return is unrealised — the manager's own mark on what it still holds. Not wrong, and mid-life funds distribute late — but it is the difference between a gain banked and a gain on paper, and only fund-level cash flows show it. Both screens below run over the same eligible set: mature funds (vintage 2019 or earlier, over $50m contributed, not flagged by the plan as not-comparable). 39 of 50 plan-lanes contribute at least one fund the screen can rate — old enough (vintage 2019 or earlier), over $50m contributed, with both cash-flow columns; the rest have no fund clearing all three and are left out.

Of 1,966 eligible funds, 100 are paper-heavy (net IRR 12.0%+ but DPI below 0.6) and 684 are proven (DPI 1.5× or more — the cash is genuinely back). The per-plan split, with each plan's whole-portfolio DPI for context — a plan early in its cycle naturally carries more paper:

The funds furthest into paper territory

Lowest DPI first — a high reported IRR sitting on the least cash actually returned (each links through to the fund in the performance tool):

Showing 30 of 100 paper-heavy funds (lowest DPI first) · Email us for the full list.

The other side — funds that have paid out

Highest DPI first — mature funds where the cash is genuinely back, not a mark:

Showing 30 of 684 proven funds (highest DPI first).

The gotcha: a low DPI on a high IRR is a prompt, not a verdict. A 2019 buyout fund should still be holding most of its companies — early distributions would be the surprise. Read it as "how much of this return has actually been paid out", check it against the fund's age and the plan's whole-portfolio DPI, and treat a fund that is both old and still unrealised as the one worth a question — not any single number on its own.

The underlying figures: the private market performance tool (search a fund, read a plan, or compare plans) · the PE statistics · where two plans disagree on the same fund · all tracked plans.