Australian super, in APRA's own numbers

Compare any funds on any reported measure — assets, cost, returns and flows — straight from APRA's published statistics. Every reported figure cited; nothing ranked or recommended. Estimates exist only in the clearly-marked Modelled section, with their error published.

Deep-dive a single fund

This quarterMar 26

The quarterly collection — whole-of-fund measures APRA publishes for the large funds. Definitions drafted, pending review.

Total fund investments

A$ · Mar 26

A$ billions

Estimated median account balance

A$ · Mar 26

A$ thousands

Median member age

Count · Mar 26

Fund investments, quarter by quarter

Total fund investments of the six largest funds each quarter since June 2022; the rest of the published large funds pooled as “Other”.

AustralianSuperAustralian Retirement T…Aware SuperUnisuperHOSTPLUS Superannuation…Colonial First State Fi…Other

How the big funds invest

Asset allocation of the largest funds · quarter ending Mar 26 — each bar is a fund's reported investments split by asset class. APRA publishes this table for large funds only. Class definitions are drafted from APRA's column headers, pending review.

EquityFixed incomePropertyInfrastructureAlternativesCashUnclassified

AustralianSuper

652012

$418.4bn

ART

582210

$356.3bn

Aware Super

581811

$209.2bn

UniSuper

5119101315

$165.3bn

Hostplus

6310

$149.0bn

CFS FirstChoice

5627

$108.7bn

Cbus

5620913

$107.6bn

Rest

59219

$105.7bn

HESTA

572710

$103.3bn

MLC

5923

$93.2bn

Mercer

641310

$81.4bn

Asgard

572014

$79.5bn

IOOF

59279

$74.8bn

AMP North

602112

$70.3bn

Active positions of the MySuper defaults

Each fund's MySuper default option: the actual asset allocation it disclosed in its own portfolio holdings disclosure (Dec 25) minus the benchmark allocation APRA publishes for the same product (Quarterly super product statistics, Table 8a, quarter ending Dec 25). Positive is an overweight. Asset classes are bucketed to make the two taxonomies comparable; which option is the MySuper default is a curated, sourced mapping — hover a bar for it. Funds with lifecycle MySuper products aren't shown: APRA's benchmark can't be tied to one lifecycle stage.

Percentage points vs benchmark

OverweightUnderweight

Actual allocation is the option's own s1017BB disclosure (fund websites); benchmark is APRA's published benchmark asset allocation. Differences of a point or two are ordinary rebalancing drift.

Two defaults, side by side

Actual asset allocation of two MySuper default options, class by class, from each fund's own portfolio holdings disclosure (Dec 25). The dashed line is the median across all 14 defaults shown above — the industry's centre, not a benchmark.

Per cent of option assets

AustralianSuperUniSuperMedian of shown defaults

Options “Balanced” and “Balanced”. The median is computed over the 14 MySuper defaults with a usable disclosure, not the whole industry.

Who owns what

Cross-fund look-through over 26 funds' portfolio holdings disclosures · as at Dec 25. Each bar is one holding, split by the funds that report it. Names merge only when a shared security id proves them identical or a reviewed synonym says so. Compiled from the disclosures funds publish on their own websites under s1017BB — not an APRA collection.

Commonwealth Bank of Australia

$66.0bn · 23f

IFM Investors

$59.0bn · 23f

National Australia Bank Ltd

$42.0bn · 24f

Westpac Banking Corp

$33.8bn · 23f

BHP Group Ltd

$29.6bn · 20f

Nvidia Corp

$24.0bn · 23f

Australia & New Zealand Banking Group Ltd

$22.6bn · 22f

Microsoft Corp

$19.9bn · 23f

Alphabet Inc

$19.0bn · 23f

Apple Inc

$18.9bn · 23f

Amazon.Com Inc

$13.8bn · 22f

Rio Tinto Plc

$13.2bn · 22f

Macquarie Group Ltd

$13.2bn · 24f

BlackRock Investment Management (Australia) Ltd

$11.9bn · 6f

Taiwan Semiconductor Manufacturing Co Ltd

$11.9bn · 23f

CSL Ltd

$11.7bn · 23f

Wesfarmers Ltd

$9.9bn · 23f

Blackrock

$9.3bn · 20f

Meta Platforms Inc

$8.9bn · 21f

Morgan Stanley Investment Management (Australia)

$8.7bn · 22f

AustralianSuper

Aware Super

Cbus

UniSuper

HESTA

CFS FirstChoice

Hostplus

Rest

Mercer

CareSuperBrighter Super

Equip

Modelledestimates — not APRA data

A theoretical model: each fund's reported asset allocation × listed market proxies implies a quarterly return — including the current quarter, before APRA publishes anything. These are computed estimates with known error, shown below; they are not reported figures and say nothing about any fund's actual performance.

Implied return, quarter to Jun 26

Start-of-quarter allocation × proxy ETF returns · largest modelled funds

Per cent

How wrong is the model?

Served model vs APRA's reported MySuper quarterly returns · 376 fund-quarters · MAE 2.0pp · corr 0.81 · fitted layers tested OOS, not better — details below

Method, proxies and limitations — read before quoting anything above

Method: Baseline: implied quarterly return = start-of-quarter Table 4 leaf allocation weights × listed proxy total returns. Fitted layer: ridge on residuals from that baseline (so shrinkage pulls toward allocation×market, never toward a constant) against APRA-reported MySuper quarterly returns, λ chosen by leave-one-quarter-out correlation; plus a shrunk per-fund intercept anchored to reported expense ratios. Fund-quarters where allocation leaves explain <90% of reported investments are skipped, never imputed.

Fit: lambda 3 · intercept (annual drag) 0.7% · au_equity 0.976, intl_equity 0.958, fixed_income 0.998, property 0.988, infrastructure 0.993, alternatives 0.999, cash 1, property_lag 0.005, infrastructure_lag -0.002, alternatives_lag 0.001, unlisted_eq_lag 0.003 · 376 training pairs · leave-one-financial-year-out (lambda chosen by OOS error). Fund level: one shrunk per-fund intercept (mean residual × n/(n+1)); a full per-fund regression is not defensible on ≤3 FY observations per fund.

Targets: quarterly_mysuper.table_2a.net_return_rep_member_quarterly (to Sep 2023) · qsps.table_4a.net_return_rep_member_quarterly (MySuper products, from Mar 2025) · Dec 2023 – Dec 2024 quarters have no published target. Targets are the fund's MySuper representative-member return, not the whole fund; the FY whole-fund check below quantifies the difference.

Whole-fund FY check: compounded modelled quarters vs whole-fund annual returns — MAE 5.0pp · corr 0.147 · 241 fund-years (different measure: MySuper option ≠ whole fund).

Proxies: AAA.AX (BetaShares Australian High Interest Cash); DJRE.AX (SPDR Dow Jones Global Real Estate); IFRA.AX (VanEck FTSE Global Infrastructure (hedged)); VAF.AX (Vanguard Australian Fixed Interest); VAP.AX (Vanguard Australian Property Securities); VAS.AX (Vanguard Australian Shares (ASX 300)); VGAD.AX (Vanguard MSCI Index Intl Shares (AUD hedged)); VGS.AX (Vanguard MSCI Index Intl Shares (unhedged)); VIF.AX (Vanguard Intl Fixed Interest (hedged))

Limitations: Unlisted assets proxied by listed equivalents — no smoothing or lag. Alternatives proxied by a 50/50 global equity / bond blend. Proxy ETF fees included; fund-level fees and tax enter only via the fitted intercept. The fit is trained on very few financial years; betas are unstable. Prototype uses free Yahoo Finance data, not licensed index series.

Validation: The SERVED model is the parameter-free baseline (allocation × market) — its error needs no cross-validation because nothing is fitted. Ridge betas, lagged-proxy features and per-fund effects were all tested under leave-one-quarter-out CV against APRA's reported MySuper rep-member quarterly returns and did not beat the baseline (pooled and fund-effect figures shown); they are reported, not served. The FY figures compare compounded quarters against whole-fund annual returns — a different measure, kept as a sanity check.

Generated: 2026-08-05T17:06:17+00:00

Annual deep-diveFY to Jun 25

The annual collection — returns, cost and the full balance sheet for every APRA-regulated fund. The next financial-year release lands late in the year.

Total assets

A$ · Jun 25

A$ billions

One-year rate of return

Per cent · Jun 25

Per cent

Five-year rate of return

Per cent · Jun 25

Per cent

Cost per member account

A$ · Jun 25 · derived

Assets under management, over time

Total reported assets of the six largest funds across annual releases; the rest of the market pooled as “Other”. Band thickness is assets that year — the consolidation story.

AustralianSuperAustralian Retirement T…Aware SuperUnisuperColonial First State Fi…Public Sector Superannu…Other

Where the assets are

Tile size is total assets · colour is the chosen measure · masked funds greyed

Explore →

Value map

One point per fund · bubble is total assets · dashed lines mark medians · each axis is its own measure

Axes clipped to the 2nd–98th percentile for readability; 2 funds with extreme values fall outside the frame and are not shown.

How this is built

Figures are ingested from APRA's published statistics through a registry-driven pipeline — no hand-typed numbers. A masked cell stays masked; it is never imputed or shown as zero. Derived measures are labelled as computed, not APRA-published. Every value on this site travels with the APRA reference it must be quoted with. Orderings describe reported figures; they are not assessments of fund quality.

Source: Australian Prudential Regulation Authority, CC BY 3.0 AU · general information, not financial advice.