Dallas Police and Fire has spent nine years getting out of private assets

Illiquid assets were about half of this $2.3 billion pension in 2018. Private assets were 17.6% at the end of 2025 and 10.4% in August 2026, and the fund is still not at its target mix. Over the ten years to December 2025 it returned 5.6% a year against a policy benchmark of 8.6%. It never ran a secondaries sale. It waited, and its own management has written the same sentence about the wait in five consecutive annual reports.

Key numbers, and what they mean

9 years

How long the fund has been reducing illiquid holdings, from about 50% of the portfolio at the end of 2018 to 10.4% in August 2026. It is still not finished.

5.6%

Its ten-year return to December 2025, against a policy benchmark of 8.6%. Compounded over the decade that gap turns a dollar into $1.72 rather than $2.28.

80 to 1

Distributions received against capital calls paid across 2021 to 2025: about $497 million in, about $6.2 million out. The fund was not investing, it was waiting to be repaid on its managers' schedule.

1.0x

Distributions to paid-in capital on the legacy private book as of March 2026, with a since-inception net internal rate of return of 1.3% across 92 fund commitments and 35 manager relationships. In aggregate the book has returned the capital and little else.

Asset mix, August 2026

Public equity56.3%

Fixed income and cash19.9%

Public credit13.4%

Real assets7.6%

Private equity2.0%

Private credit0.4%

Legacy energy fund0.4%

Preliminary. Private and other illiquid assets total 10.4%. Having spent nine years above target, the fund now sits below its 6% private equity and 4% private credit targets, and its policy sets no rebalancing range for illiquid classes.

The read-through

From the report

Takeaway: how to apply it to a portfolio

The same sentence appears in its annual report in 2021, 2022, 2023, 2024 and 2025: it will take several years to reduce private assets to the levels in the asset allocation. Across those years the weight went 33%, 32%, 26%, 22%, 17.6%.

Several years is a description, not a plan. If you are over-allocated to private funds, build the expected distribution schedule fund by fund and then assume it slips. The exit runs on your managers' timetable, not yours.

Across 2021 to 2025 the fund took in about $497 million of distributions against about $6.2 million of capital calls, and the private weight still fell only about 15 points.

The percentage falls slowly because the denominator falls too. Benefits and spending come out of the liquid side, which concentrates the problem while you wait. Size the liquid sleeve to your outflows first. This fund's policy now holds roughly 18 months of net benefit payments in cash and short bonds.

The legacy private book shows a since-inception net return of 1.3% across 92 commitments and distributions equal to 1.0 times paid-in capital, and the fund now sits below its private equity target and is committing again.

The cost of being stuck is not only the bad assets. It is the years you cannot participate in anything else, and then the overshoot when you finally get free. Decide your steady-state private weight now and pace back toward it, because nothing pulls you there automatically.

The Argus lensThe problem here was never one bad fund. It was 92 commitments across 35 relationships whose exit timing the fund did not control, and a book that returned its capital and little else. Knowing what a fund can call, when it can call it and what it is likely to return is a documents question, and it is ours. Benchmark your deals in the Portal →

About the source

The fund's 2025 Annual Comprehensive Financial Report, audited in June 2026, covers the year ended 31 December 2025: the 17.6% private-asset weight, the ten-year return against the policy benchmark, and the distribution and capital-call figures. The August 2026 allocation is from its own portfolio update and is preliminary. The private-book performance figures are from its consultant's private markets report as of 31 March 2026.

View the source →