CalPERS asset allocation: growth and risk versus MSCI World

Executive summary

CalPERS compounded at 8.57% a year after investment costs over July 2016-June 2026, versus 13.14% for MSCI World net total return in US dollars. It reported milder annual fluctuations and higher annual-observation Sharpe and Sortino ratios. Its $635.4bn pension portfolio is built to pay benefits, with 36.3% in private assets at 30 June 2026. Yet public and private equity contributed 84% of its forecast volatility at that date, and CalPERS put its equity-equivalent exposure at 75.2% on 30 September.

Copio reads this as lower realised growth alongside lower reported volatility. The slightly higher risk ratios do not establish geometric-growth optimality or crash protection. The charts below separate annual performance, a restricted public-market frontier and CalPERS' own correlation assumptions. Private-asset valuation smoothing makes the apparent risk advantage less conclusive (performance report; risk report).

What is in the fund?

CalPERS' Public Employees' Retirement Fund, or PERF, combines listed and private businesses, bonds, private lending and real assets. The examined 30 June 2026 allocation has more private equity and less private debt than its policy benchmark. The negative strategic-leverage line records financing that supports additional asset exposure; it is not an investment with a negative market value that offsets losses automatically.

Actual allocation: public equity 38.1%; private equity 19.5%; income 28.0%; real assets 12.6%; private debt 4.3%; strategic leverage -2.4%. Source: CalPERS GIPS report, PDF p. 7; allocation data (CSV). Weights total 100.1% after rounding.

The 28% income allocation includes 8.3% Treasuries, 5.0% mortgage-backed securities, 4.5% investment-grade corporate bonds, 5.1% high yield and 5.1% emerging-market sovereign debt. Much of it therefore carries credit or interest-rate risk. Real assets include property, infrastructure and forestland. Private equity, private debt and real assets bring illiquidity as well as different investment opportunities (GIPS report, PDF pp. 7-8).

CalPERS' first quarterly report under its new Total Portfolio Approach regroups the fund into reference-portfolio assets, active programmes and diversifiers. At 30 September 2026 it valued PERF at $651.3bn, with private equity at 22.4%, real estate at 8.5%, infrastructure at 4.1% and private debt at 4.8% (quarterly portfolio report, PDF p. 4). Those four lines total 39.8% by our sum, against 36.3% in private assets reported for June under the earlier categories.

The new allocation table nets financing and liquidity to -6.2% of the fund. CalPERS' broader measure of financing and derivatives that create recourse to PERF put total fund leverage at 11.8%, against a 20% limit. Equity-equivalent exposure was 75.2%, within an operating range of 70-80%, and forecast volatility was 14.2% (quarterly portfolio report, PDF pp. 2 and 5).

Why does CalPERS use this structure?

CalPERS defines success as paying promised benefits while maintaining adequate funding. Its investment beliefs tie the asset structure to growing cash requirements and inflation-sensitive liabilities. That explains the emphasis on income, inflation exposure and liquidity alongside appreciation. The fund's actuarial return target was 6.8% at June 2026, with allocation reviewed through a four-year asset-liability cycle (GIPS report, PDF p. 7).

Its justification for private assets is conditional: a long horizon permits illiquid investments if they earn an appropriate premium and leave enough cash for obligations. CalPERS also says it should diversify across distinct return drivers, take risk only when it expects compensation, and use indexing where active management lacks supporting evidence. Its scale should improve fee negotiations (investment beliefs, PDF pp. 61-63).

The March 2024 allocation review makes its diversification case more specific: compare portfolios targeting the same projected 6.8% return, then add assets and financing to reduce modelled drawdown risk. CalPERS intended to retain expected growth while reducing risk. That is its modelling rationale; the realised comparison below tests the outcome over a different period.

In July 2026, CalPERS introduced its Total Portfolio Approach, with more flexibility across asset classes and a 75% global-equity, 25% US Treasury reference portfolio as a public-market yardstick. It is separate from the June holdings snapshot and from the MSCI World comparison below. For July-September 2026, CalPERS reported 1.5% against 0.2% for the reference portfolio (quarterly portfolio report, PDF p. 3); one quarter does not test the new approach.

Historical performance versus MSCI World

Over the ten fiscal years ending June 2026, a hypothetical $100 invested at CalPERS' published net returns grew to $227.65. The same amount at MSCI World's net USD total returns grew to $343.63. CalPERS' geometric return was 4.56 percentage points lower per year. Both calculations reinvest returns without contributions or withdrawals; the index includes withholding-tax adjustments but no fund implementation fee.

June 2016-June 2026: $100 becomes $227.65 at CalPERS and $343.63 at MSCI World; the year to June 2022 lost 6.09% and 14.34%. Sources: CalPERS GIPS report, PDF p. 6 and MSCI World NETR USD; annual returns and wealth (CSV).

Our calculations use the CalPERS annual net-return table, MSCI's monthly index levels and Kenneth French's monthly cash-return series. CalPERS' annual-observation Sharpe was 0.91 versus 0.85 for MSCI World, and its Sortino was 3.32 versus 2.55.

Sharpe measures excess return relative to variability; Sortino uses only shortfalls below cash in its risk denominator. Both use the same ten July-June observations, rather than mixing annual fund returns with monthly index ratios.

At CalPERS' measured annual excess-return risk, a hypothetical 52.1% MSCI World / 47.9% cash mix compounded at 8.12%, versus the fund's 8.57%. Sources: the same CalPERS, MSCI and French inputs; index/cash frontier data (CSV). Annual rebalancing, no leverage or implementation costs; these are comparisons, not proposed allocations.

The plot uses arithmetic mean return against annual excess-return variability, with CAGR, Sharpe and Sortino beside each portfolio. CalPERS sits slightly above this restricted index/cash frontier on the reported observations. That is a more favourable result than the full-equity CAGR comparison alone suggests, but private valuation smoothing can improve its apparent position. It is not a frontier for CalPERS' entire investment opportunity set.

CalPERS' higher ratios accompany a smaller reported annual loss in 2021/22, but also much less participation in the preceding equity surge. Only one fiscal year fell below cash for each series, making Sortino particularly sensitive to that observation. Annual returns conceal intra-year crashes, while CalPERS itself warns that smoothed private valuations bias realised volatility downwards (risk report, PDF p. 24).

Copio's reading

Is it optimal for geometric growth?

This record does not make the case for a geometric-growth optimum. The portfolio compounded materially less wealth than global equities, despite higher reported Sharpe and Sortino ratios. Those ratios describe return relative to measured variability; they do not show which allocation maximises compounded wealth. The distinction matters here because private-asset smoothing improves the appearance of risk without creating additional economic wealth.

MSCI World alone would also expose a pension fund to a different loss and liquidity profile. A credible improvement must meet the same benefit payments and capital calls, include private-asset fees and financing costs, and increase expected log wealth over a stated horizon. The historical comparison establishes a growth shortfall against equities; it does not establish that replacing the fund with equities would solve its pension obligations better.

Is it crash aware and protected?

The measured outcome is lower annual variability with substantially lower compounded growth. CalPERS' annual-return standard deviation was 7.14%, versus MSCI World's 13.97%, while CAGR was 8.57% versus 13.14%. The record includes less participation in the 2020/21 equity surge and a smaller loss in 2021/22. A smaller reported annual loss does not establish that capital or liquidity was protected during a crash's worst weeks.

CalPERS' 2024 review explicitly counts private-asset valuation smoothing and lags among the features moderating fluctuations. With 36.3% in quarterly-valued private assets, the reported path differs from immediately saleable wealth. Benefits, capital calls and derivatives or repos still require cash (GIPS report, PDF p. 8; liquidity definitions, PDF p. 59). Copio would require evidence of payouts and usable liquidity under stress before crediting this design with crash protection.

Is it diversified without giving up growth?

It diversifies asset ownership, but equity remains the dominant risk and growth lagged the equity comparator. At 30 June 2026, public and private equity made up 57.5% of the fund's value in CalPERS' risk model and contributed 84% of its 13.9% forecast volatility (risk report, PDF p. 26). CalPERS calls these estimates forward looking and notes that lagging private valuations can make outcomes differ, so they show shared equity exposure, not a measured crash correlation (quarterly portfolio report, PDF p. 5).

The correlation matrix gives the specific diversification assumptions behind CalPERS' earlier modelling. Its March 2024 review assumed correlation of 0.6 between market-cap global equity and private equity, compared with 0.1 between global equity and long US Treasuries. These dated assumptions support a clearer diversification case for Treasuries than for another equity allocation; they are not observations of how the assets moved during a 2026 crash.

Modelled correlation with global listed equity: private equity 0.6; private debt 0.4; US real estate 0.3; long US Treasuries 0.1; high yield 0.4. Source: CalPERS March 2024 asset-liability review, PDF p. 10; full 11-asset assumption matrix (CSV). US real estate represents that source's property category, not all CalPERS real assets.

There is a case for diversification in those assumptions, but it has not preserved equity-like growth in the measured decade. Nor do average correlations establish protection when markets fall together. Copio's concern is the combination of lower compounded wealth, substantial shared equity exposure and illiquidity. The higher annual Sharpe and Sortino temper that criticism; they do not settle whether the diversification improves long-run economic wealth after all costs.

CalPERS cannot receive a faithful unlevered UCITS ETF replica grade because private assets, derivatives and financing materially affect its portfolio. Our studies of CPP Investments and Japan's GPIF examine other pension structures. For the underlying comparison, observed asset behaviour during crashes matters alongside the amount of wealth retained at the endpoint.