Key takeaways
The framework, not the portfolio. What traveled from Yale was the decision architecture.
Writing was the strategy. A public framework lets managers, ICs, and successors challenge it.
Misalignment is a documentation problem. Governance debates rarely close cleanly when LP rules stay unwritten.
Legibility is a competitive variable. GPs adapt to LPs who write down what they expect.
The year the framework left Yale
In 2000, 15 years into his tenure as Yale's Chief Investment Officer, David Swensen published Pioneering Portfolio Management. The endowment was already producing. He had been hired in 1985 at age 31, taking an 80% pay cut from a Wall Street career to manage what was then a $1.3 billion portfolio.
Dean Takahashi joined a year later. Together, drawing on the work of Swensen's Yale mentor James Tobin, they had spent 15 years quietly rebuilding how a major endowment thought about risk, manager selection, and time horizon.
The book is the inflection point. Swensen had a working portfolio before he published, but he did not have a framework other allocators could be measured against. After 2000, any GP walking into a Yale meeting knew the diligence standard before the conversation began. Most managers across most LPs still do not have that kind of clarity from the institutions they fundraise with.
The Yale model is the headline, the framework is the part that traveled
By Swensen's death in May 2021, Yale's endowment had grown from $1.3 billion to $42.3 billion, with an annualized return of 13.7% over 36 years and outperformance of more than $50 billion versus the average endowment. Those are Yale's own published figures.
Most coverage credits the asset allocation. Heavy alternatives, equity orientation, illiquidity tolerance, the so-called Yale Model: imitated for two decades, contested through the post-2008 liquidity crunch and the recent denominator-effect debates. The portfolio is the part everyone copies and almost no one can replicate.
The framework around the portfolio is different. Pioneering Portfolio Management contains the asset allocation, but its longer half-life is the decision architecture surrounding it:
A clear principal-agent posture toward GPs, treating manager selection as the primary source of alpha rather than asset weighting
A published view that consultants tend to function as what Swensen called a dysfunctional filter, pulling decisions toward the average
An explicit framing of peer comparison as informed choice management trap rather than a governance tool
A standard for how Yale evaluated active management, including when it rejected the case for active managers entirely
"We're principals for the university, engaging agents, the hedge fund managers or the buyout managers, and trying to find ways to get those agents to act primarily in our interest." - David Swensen
A CIO running a $500 million endowment cannot replicate Yale's portfolio. The decision rules above are size-agnostic. That asymmetry is why the framework outlasted the allocation. Andrew Golden, who joined the Yale Investments Office as a senior associate before going on to run Princeton's PRINCO for nearly 30 years, has been consistently public about how much of his approach to organizing an investment office traces back to what he learned at Yale.
Why the second book is the one most LPs misfile
Unconventional Success, published in 2005, gets shelved as personal-finance advice. That misreads what it actually is.
Swensen, then the most influential institutional allocator of his generation, used the book to publicly call the for-profit mutual fund industry a "colossal failure" defined by "systematic exploitation of investors." He recommended Vanguard and TIAA-CREF by name as the only firms operating in their clients' interest. He accepted the commercial friction that came with publishing those views. His manager access did not contract afterward.
What the book signaled
Swensen demonstrated that an LP could publish a philosophy that imposed real costs on the relationships around it and continue to function. That is rare. Most allocators keep their investment philosophy internal — sometimes because the content is genuinely proprietary, often because publishing imposes a decision framework that running an opaque process avoids.
The practical consequence shows up most clearly during fundraises. A PE manager opening a process often modifies its terms two or three times in response to verbally communicated preferences from different LPs that none of them have written down. The final fund structure resembles something no individual LP wanted but every LP partially shaped.
As Collateral has written about how allocators evaluate first-time managers, much of what feels like negotiation late in a raise is interpretation drift that earlier documentation would have surfaced.
The four jobs the writing did inside Yale
The books were operational. Four functions of the published framework sit inside Yale's investment process:
Directive management. GPs approaching Yale knew the selection standard before the meeting started. The asymmetry that defines most first GP-LP conversations, where a manager pitches into an unspecified standard, was substantially reduced. The result, per Yale Investment Committee chair Charles Ellis: Yale's average manager relationship lasts 17 years.
"Loyalty flows both ways. Investors owe external advisors the opportunity to pursue investment activities within a reasonable time frame. Firing a poorly performing manager due to short-term results is unreasonable." - Pioneering Portfolio Management
IC alignment across personnel turnover. Yale's investment committee saw its chair rotate at least twice during Swensen's tenure, and members turn over throughout. The framework gave incoming members a single dated reference document that did not depend on Swensen being in the room.
Successor training at scale. Swensen's mentees now lead the endowments of Princeton, Stanford, MIT, and the University of Pennsylvania, "among other major institutions," per Yale's own succession announcement. Mentorship at that scale required documentation. Books were what their incoming staff read first.
Authority beyond AUM. Swensen's leverage in the LP community came from the framework being public and defensible, not from the size of the portfolio. By 2021, Yale's $42.3 billion endowment was smaller than Harvard's and an order of magnitude below the major sovereign wealth funds. The framework's reach extended further than the dollars.
Writing converted his practice into something that compounded outside his own working hours, and outside Yale's institutional perimeter. This is the part the Howard Marks memo programme at Oaktree operationalises on the GP side, and that almost no LP operationalises on theirs.
Swensen warned smaller endowments not to copy him
The most credible critique of the "Swensen as replicable model" thesis is documented inside his own work and in his closest colleagues' commentary.
Charles Ellis, who chaired Yale's investment committee for 17 years, has been explicit that most institutions claiming to follow the Yale Model are not, and should not try. Speaking at a Yale School of Management programme for Chinese executives, Ellis used a line from When Harry Met Sally to describe what he sees as the imitation problem: "I'll have what she's having."
Consultant Anna Dunn Tabke, in the same feature, points out that there are "only so many really good alternatives managers, and they are oversubscribed." The Yale Model only works if you can already get inside the door.
Swensen himself worried that under-resourced endowments would chase Yale's returns without the staff or governance to support them. He used Unconventional Success in part to redirect them toward passive index investing.
"Establishing and maintaining an unconventional investment profile requires acceptance of uncomfortably idiosyncratic portfolios, which frequently appear downright imprudent in the eyes of conventional wisdom." - David Swensen
The framework travels, the portfolio doesn't
The framework is replicable. The portfolio is not. Conflating the two has produced exactly the imitation-without-replication problem Ellis describes. Even on documentation, scale matters. A small endowment writing a 30-page framework no one reads is theatre. Replication depends on the analytical process, not page count.
Misalignment is debated as a governance problem and sits one layer deeper
ILPA Principles 3.0, the continuation-vehicle controversies, the NAV-loan disclosure fights, the LPAC-consent debates: all typically treated as governance issues solvable through better terms. The economic causes behind them (carry waterfall design, GP commit levels, fee offsets) will not be resolved by anyone writing a better policy document.
But governance debates rarely close cleanly even when the term-sheet language gets tightened. Part of the reason is interpretive. When LPs do not publish their decision rules, GPs guess. The relationship arrives at the term-sheet stage already drifting.
What the documentation gap looks like in real time
Two patterns from the past two years make the mechanism visible.
The first plays out before the pitch. PE buyout management fees hit a 20-year low of 1.74% in 2024 according to Preqin data, with GPs making concessions on fees and terms across the market in response to slow LP commitments. ILPA's senior managing director Greg Durst told the Financial Times that LPs were being "very slow and judicious about how they're going to be making new commitments."
The headline number is the output of a specific process. GPs run multi-month fundraises, modify terms in response to verbally communicated LP preferences, and end up with fund structures that average out conflicting feedback rather than reflecting what any single LP actually wanted. Documentation rigor on the LP side would not have prevented fee compression. It would have changed which conversations happened, and when.
The second plays out after the commitment. In November 2025, the Abu Dhabi Investment Council sued Energy & Minerals Group in the Delaware Court of Chancery to block a continuation-vehicle transaction involving Ascent Resources. ADIC's complaint alleged the GP ran a compressed timeline, provided different data to different investors, and prevented LPs from conferring with each other.
ADIC's filing notably cited ILPA's continuation-fund guidance as evidence of industry standards, which is the live question — what the LP and GP agreed those standards would be. The allegations remain unproven and will be resolved in confidential arbitration. But the case sits squarely on the gap between what an LPA permits and what an LP believed it had agreed to, and that gap is a documentation problem before it is a governance problem.
Neither pattern is solved by tighter LPAC consent. Both surface the same asymmetry: GPs publish PPMs, LPAs, fundraising decks, performance reports, side-letter templates, and ESG frameworks. Most institutional LPs publish almost nothing about how they actually evaluate managers.
What is replicable from Swensen, and what is performance
Most CIOs will not write books, and they should not pretend they will. What is replicable is the approach: a written, dated framework that articulates an LP's decision rules, the trade-offs the LP accepts, and the standards GPs are evaluated against. It should be circulated internally for IC consistency, shareable with prospective managers under appropriate confidentiality, and updated when necessary.
The book-as-platform, the public-intellectual posture, the named-firm critiques — those are Swensen-specific. They are not the practice the rest of the LP industry needs to copy. The replicable part is whether your IC, your successors, and the GPs you partner with can read your decision rules without sitting in a room with you.
Where opacity remains an advantage
Not every LP should publish a framework. Allocators whose edge is opportunistic, relationship-driven, or built on information asymmetry actively benefit from being illegible. Some family offices fall into this category. Some sovereign-wealth strategies do too.
The Swensen practice is most useful for LPs whose strategy depends on long-horizon GP partnerships, where the cost of repeated misalignment compounds over a decade or more.
Bottom line
What changes if even a meaningful minority of allocators start writing their rules down? GPs tend to adapt to legibility faster than to most other shifts in LP behaviour, and three things follow:
Self-selection improves: managers who know they are not a fit stop pitching, which raises the average quality of the conversations the LP does take.
Fundraising timelines compress, because the rounds of mutual interpretation that pad most processes are no longer necessary.
And the LPs who have done the work develop a quiet bargaining advantage over the LPs who have not, because GPs would rather negotiate against a known standard than guess at an unknown one.
Documentation rigor is currently treated as an internal exercise: useful for the LP, neutral for the GP. That framing leaves out the GP side of the equation. Allocators with explicit, written decision rules give managers something to plan against, which tends to surface better-fit conversations earlier in the process.
Allocators without them stay vulnerable to the same alignment debates every fundraise — not because the debates are unsolvable, but because the upstream conditions for resolving them have not been put in writing.
If you're working through how your investment philosophy reads to the managers you partner with, or how your fund's positioning reads to the LPs evaluating it, that's the kind of conversation we have at Collateral Partners.
