CalPERS replaced strategic asset allocation with a total portfolio approach on July 1, 2026, the first US pension fund to do so. Under the framework, every investment competes for capital on its contribution to the whole portfolio. The reported 1.3 percentage point advantage comes from 26 of the world's largest funds, so it plausibly measures organizational capability rather than the technique.
The longest-running unresolved question in portfolio construction received its most consequential live test this year. On July 1, 2026, the California Public Employees' Retirement System replaced strategic asset allocation with a total portfolio approach, becoming the first US pension fund to do so. The portfolio, worth roughly $620 billion, had about 36% in private markets, below the 40% target set in March 2024.¹
The change raises the question of whether capital should be allocated top-down to asset classes that then compete internally, or bottom-up to individual opportunities that compete with everything else the portfolio could own.
What did CalPERS actually change?
Under strategic asset allocation, a board sets percentage targets by asset class, and each class is managed against its own benchmark in its own silo. Under a total portfolio approach, every investment competes for capital based on its contribution to the whole portfolio's objective, evaluated through a common risk lens and measured against a reference portfolio. CalPERS adopted a seventy-five/twenty-five equity-and-bond reference portfolio and a 400-basis-point active risk limit.²
For managers, the implication is clear. The CAIA Association's description of the framework notes that under a total portfolio approach, a fund must demonstrate how it contributes to portfolio-level risk, diversification, and resilience.³ An asset-class label no longer entitles a strategy to a share of capital. That is a higher bar than simply clearing an asset-class benchmark, and it initially reads as a decisive outcome for bottom-up selection.
Does the performance evidence support the framework?
The performance case cited in support is specific. Research referenced by CalPERS found that total portfolio approach users outperformed strategic asset allocation users by 1.3 percentage points annually over ten years, across a sample of twenty-six of the world's largest pension and sovereign wealth funds. A separate PGIM analysis estimates a risk-adjusted advantage of about one percentage point, attributing part of it to more dynamic allocation shifts through market cycles.⁴
The sample is problematic if read as a verdict on the method. Twenty-six of the world's largest and best-resourced investment organizations are not randomly assigned to treatment. They are a group selected for capability, and the framework is what capable organizations adopt.
The framework's architects describe it that way. Roger Urwin, who helped develop the approach, has characterized it as governance reconsidered through a systems lens rather than as an investment method. Russell Investments observes that adoption has concentrated among mega-funds, where scale, tooling, and organizational silos create the greatest room for improvement, and that smaller investors may neither require nor be able to manage the complexity. It also notes that most adopters retain a reference portfolio and that compensation and accountability become materially harder once asset-class benchmarks no longer define success.⁵ TIFF has argued that the framework is largely established practice under a new name.⁶ CalSTRS is pursuing a comparable one-fund model.⁴
Read that way, the 1.3 point difference is plausibly a measurement of organizational capability rather than of an allocation technique, and the causal arrow runs opposite to the one usually implied.
Is top-down allocation really different from bottom-up selection?
The allocation-versus-selection debate assumes the two are alternatives. The total portfolio approach makes a different claim: that they are the same decision examined at different levels of resolution, and that the binding constraint is whether an organization can evaluate a specific position's marginal contribution to the total portfolio.
Most organizations cannot, so strategic asset allocation exists for that reason. It is a governance technology for making allocation decisions when that capability is absent. Percentage targets, asset-class silos, and benchmark-relative evaluation are not intellectual errors. They are a functioning substitute for an analytical capacity that is expensive to build, and they produce accountable decisions with far less machinery. A firm that adopts total portfolio language without the underlying capability has not upgraded its process. It has removed the constraint that was compensating for the gap.
What can an advisory firm use from this framework?
The transferable element is not the framework, which is calibrated for institutions with dedicated risk teams and proprietary analytics. It is the diagnostic the framework forces, which requires no infrastructure.
- For each private markets position held, what does it contribute that the rest of the portfolio does not already provide?
- What is the full cost of that contribution, measured by fee, liquidity, transparency, and governance attention rather than fee alone?
- If the position were removed, what specifically would the portfolio lose, stated in terms a client would recognize?
- Which positions could answer those three questions today, and which are occupying an allocation slot rather than earning one?
You can ask those questions without a reference portfolio, a risk-factor model, or a four-hundred-basis-point active risk budget. They are the substance of the framework; the machinery is the implementation.
What does this mean for private markets allocations?
This series began with a structural question — what an investor is compensated for when capital is locked — and it closes on the same point from the other direction. From a total-portfolio lens, a perpetual vehicle's narrower distribution and a locked vehicle's wider one are not competing products to be ranked. They are distinct contributions at different prices, and the correct choice depends on what the rest of the portfolio already holds.
Every philosophy examined across these six posts shares a defect: each was formulated as a rule, and each is a conditional. Diversification reduces risk when returns are symmetric. Past performance informs the next decision when the mechanism producing it is intact. Indexing captures the market when the index represents it. Vintage discipline protects entry when the allocator controls deployment. Allocation frameworks improve outcomes when the organization can operate them.
The conditions are the content. The rules are what remains after the conditions have been forgotten.
For informational purposes only. Not investment advice, and not an offer to sell or a solicitation of an offer to buy any security.