Private credit: big enough to matter, too opaque to measure

Published estimates of Australian private credit run from $50bn to $250bn, and nobody publishes a register. What can be measured shows super funds' private debt up almost three-fold in four years, banks still growing in development lending, and non-banks taking a bigger share.

01 · How big is it?

Nobody can say, and the estimates run from $50bn to $250bn

Private credit is lending that does not come from a bank: a fund raises money from investors, lends it out and passes the interest back. In Australia a large share of it goes to property, including construction and development. The RBA says the full size is not directly observable. Published estimates differ by a factor of five, because each counts something different.

Against the $2.6tn of Australian housing credit (RBA, Aug 2026), even the top of the range is just under a tenth, and a $200bn market is 8%. It is small next to the banks, and large enough to matter for the parts of the economy it lends to.

02 · Who supplies the money

Super funds hold $46bn of private debt, and a handful of funds hold most of it

Super is the one group of investors whose private debt we can measure. APRA's tables show $46.2bn in Jun 2026, up from $16.4bn in Jun 2022. That is 1.5% of super assets, up from 0.8%. The largest holder is Australian Retirement Trust at $6.5bn; the top five funds hold 52% and the top ten 77%.

Industry funds hold $28bn of the total, retail funds $13bn and public sector funds $5bn. As a share of the fund, some mid-sized funds hold much more than the big ones: CareSuper 4.8%, MLC 4.0%.

A note on the latest quarter. The rise from $36.7bn in Mar 2026 to $45.7bn in Jun 2026 is 91% accounted for by three funds (AustralianSuper +$3.3bn, Cbus +$2.8bn, CareSuper +$2.1bn). Cbus's figure moves between $0.7bn and $3.5bn over the last six quarters, which looks like funds reclassifying assets between categories, not money moving. The long climb is real. A single quarter's jump should not be read as growth.

03 · What it lends to, and where banks stand

Banks are still growing too. Non-banks are growing from a smaller base.

The picture that private credit replaced the banks does not fit the tables. Banks' exposure to land development and subdivision is $34.5bn, and to other residential property (which includes apartment development) $66.7bn, both higher than two years earlier ($24.0bn and $49.5bn). Bank business lending to residential building construction rose from $15.6bn in Jun 2023 to $21.4bn in Jun 2026.

The bank series steps up in March 2023 (land development from $17.1bn to $23.6bn in one quarter), which looks like a change in how it is reported, so the comparison starts at June 2023.

At the same time, non-bank lenders have taken a larger share of new housing loans: from about 5% in the years to 2023 to 11% in Jun 2026, with new commitments of $10.8bn. Commitments to construct dwellings, across the lenders that report to the ABS, went from $2.5bn in the Jun 2023 quarter to $6.7bn in the Jun 2026 quarter, so demand for construction finance has more than doubled.

These ABS series cover lenders that report to the ABS, which are not the private credit funds themselves. They show the pattern, non-banks gaining share, and cannot size private credit.

04 · Signs of stress

Insolvencies are flat once one group collapse is set aside; listed managers have been sold hard

1,540 NSW construction companies entered external administration in 2025-2026, against 1,567 the year before and 513 in 2021-2022. That is three times the level of four years ago, but it has stopped rising.

The monthly figures need care. In August 2026, 710 NSW construction companies entered external administration, against 164 a year earlier. 545 of them were voluntary administrations by one administrator on 25 August 2026, the day Bathla Construction Pty Ltd entered voluntary administration. A large group counts as hundreds of companies. Without that cluster the month is 165, level with the year before. Counts of companies overstate how many separate businesses failed when a group collapses.

In prices, the pattern since mid-August is uneven. The first reported freeze on a fund's withdrawals was on 14 August (press reports); Bathla's administration followed on 25 August 2026. From 13 August to 29 September 2026, four listed managers fell between 21% and 31% (Australian shares fell 4%), while the listed credit trusts moved between −9% and +1% and corporate bond ETFs about −2%. The market sold the managers far harder than the funds that hold the loans. One lender, Liberty Financial, rose.

This is a pattern in prices, not proof of why they moved. Several of these companies have other businesses, and trusts can trade at a discount or premium to the value of their loans, which we do not observe here.

05 · What we do not know

The gaps are the point

• The size. The published estimates run from $50bn to $250bn. The one group we can measure directly, super funds, is $46bn.
• The official series stopped. The RBA's table of lending by non-bank financial institutions ends in Jun 2019.
• Who borrows. Super funds' holdings files list some of the managers they use, but private debt itself appears as a single line in the fund we checked, with no borrowers and no breakdown of property lending.
• No register. ASIC does not publish a scheme-level register with fund sizes and strategies. A Treasury consultation on 23 September 2026 proposes collecting that data, which would close the gap.
• The overlap. Nobody publishes how much of the lending sits with the same borrowers across many funds, which is the concentration question the Bathla case raises.

06 · A claims check

Five things in circulation, tested against the tables

07 · Sources

The Bathla group is in administration. This report states only facts from ASIC's published statistics and does not characterise any party's conduct.

What this does not show

• The size estimates are reported, not measured. The RBA, ASIC, Foresight and Tanarra figures are as reported and should be checked against the primary documents. They count different things, and none is a complete count.
• Super is one slice. The APRA tables cover super funds with more than six members, and some cells are masked. Private debt is the funds’ own classification, which changes, so single quarters can jump.
• The ABS series are not private credit. They cover lenders that report to the ABS. Non-ADI lenders are mostly non-bank lenders, not funds, and the construction finance series has no split by lender.
• Company counts are not business counts. ASIC counts companies, so a group collapse adds hundreds. We show the cluster separately.
• Prices are a pattern. The price changes show what happened, not why. We do not observe trust discounts to the value of loans, and some companies have other businesses.
• The price window ends on the last day all series have a price. 29 September 2026. Later moves are not included.
• Events are from public sources and press. The date of the first fund freeze (14 August) is from press reports. The Bathla administration date and the size of the administrator cluster are from ASIC data.