A paper proposing a new framework owes the incumbents a fair hearing first. When I gave the existing preservation institutions that hearing, the finding was inconvenient for anyone hoping to announce a crisis.
The existing inventory solves more of the preservation problem than reform literature tends to admit.
The useful exercise is to ask, for each institution, five questions. What preservation problem was it built to solve? Which risks does it absorb? Which does it hand to the worker? What does the worker actually own? And how does it fail?
Do that across the whole inventory and the institutions sort into three families by their deep structure. Before them stands an older one that a compensation series has to name even though it sits outside the boundary: for most of recorded history, and for a majority of the world's workers now, the preservation architecture is the family itself. Value transferred to the young in rearing and education, returned to the old in support, with pooling enforced by obligation rather than contract. Remittance flows, which in much of the world dwarf formal pension flows, are that architecture operating at global scale. It sits outside this paper's boundary because a compensation series audits the architecture that an employer's increment can carry, and the family is not funded by an increment. That is a boundary of the paper, not a judgment of the institution.
The Promise Family
The public pension pillar, defined benefit plans, annuities and insurance all preserve by promising.
Their genius is pooling and guarantee. The public pillar delivers indexed, portable, mandatory basket-preservation at social scale, immune to the participation failures that afflict every voluntary system. The defined benefit plan, at its best, named a liability and delivered lifetime income against it. And the annuity converts a stock into an income that cannot be outlived, which no individual holding any asset can replicate, because mortality credits require a pool.
What the worker owns in this family is a claim, and the failure modes follow from that. Sponsor insolvency before protective regulation, of which the 1963 Studebaker termination is the canonical case. Underfunding. The inflation vulnerability of nominal promises, demonstrated at scale in the 1970s. The backloading that taxed mobile workers without ever saying so. And the simple structural fact that a promise's horizon is the promisor's horizon, not the worker's.
Bodie described pensions as retirement income insurance, and the description captures both the power and the dependency. Insurance is only as good as the insurer and the terms.
The Property Family
Defined contribution accounts, diversified portfolios, real estate, gold, cash and employer equity preserve by owning.
Their genius is independence and continuity. What the worker holds does not terminate with a sponsor, travels with varying friction across employers, and remains theirs through institutional change.
Diversified equity ownership carries the strongest long-horizon record of real accumulation in the modern data, and it is a claim on future production itself, which is, in the deepest sense, what a future liability is a claim against. Inflation-linked bonds solve the basket target about as completely as anything can. Real estate hedges the housing component of the liability while being illiquid, indivisible, politically encumbered and, for most households, leveraged, which magnifies both the hedge and every risk attached to it. Cash solves short horizons and fails long ones. Gold has held value across centuries while paying nothing and lurching across decades. Employer equity concentrates the worker's financial capital on the same balance sheet as their human capital, which is an assignment only alignment objectives could love.
What the family hands the worker along with ownership is the full weight of decision, timing and discipline. And its characteristic failure is worth naming precisely, because it is not the one people expect. The property family's failure is not confiscation. It is abandonment: accounts unopened, allocations never revisited, savings cashed out at job changes, and the steady leakage of value through fees and inattention.
The Hybrid Family
Target-date defaults, endowment-style pooled vehicles, sovereign wealth funds and modern portable accounts are engineering attempts to bolt the promise family's virtues onto the property family's chassis.
The target-date fund embeds a crude liability logic, de-risking as the horizon shortens, into a property instrument that requires no decisions. Automatic enrollment supplies institutionally the participation that voluntary systems never achieved. Sovereign funds and endowments demonstrate what patient governance can do with pooled property across generations.
The family's lesson is architectural rather than asset-based, and it is the most useful sentence in the whole inventory: defaults, pooling and governance are technologies, and they can be attached to property.
Three Conclusions
First, every family dominates somewhere. Promises dominate pooling and certainty. Property dominates continuity and independence. Hybrids dominate usability. That is the trade-off literature's own conclusion, and it is why serious pension scholarship converged decades ago on multi-pillar designs that deliberately mix the families rather than crowning one.
Second, and this is the finding that reorganized how I read the whole inventory, the strongest performers share one feature that is not an asset property at all. They were matched to an explicitly named liability. The defined benefit plan and the annuity knew exactly what they were for. So does an endowment. The drifting performers are the ones asked to preserve in the abstract.
Third, the worker of the fragmented modern career is typically holding a little of each family, assembled by accident across employers, matched to nothing.
The problem is not an empty inventory. It is an unassembled one.
The Confrontation I Owed the Reader
There is one place where the paper speaks with two voices, and I decided to stage the collision rather than let a reviewer find it.
The inventory praises the public pension pillar as indexed, portable, mandatory basket-preservation at social scale, which it is. Elsewhere the paper examines sovereign balance sheets and their preservation failure modes with some care: inflation, negative real rates, financial repression, duration losses, taxation, policy-regime change.
Both passages describe the same balance sheet.
The public pillar's genius and its exposure are one fact viewed from two sides. A single obligor, whose benefits are adjustable by legislation and eroded by the same inflation channels. And a worker whose floor rests on the public pillar while their so-called safe portfolio assets rest on the same sovereign's bonds has concentrated more on one balance sheet than any statement discloses.
That is not an argument against the public pillar, which remains the best-performing preservation institution most workers will ever touch. It is the discipline of naming concentration wherever it occurs, including inside the institutions the paper admires. A framework that only finds problems in the arrangements its author dislikes is not a framework.
What This Means If You Design Plans
The practical translation is short.
You are almost certainly not looking at a menu of bad instruments. You are looking at a set of decent instruments that were never assembled against a stated purpose, in a workforce that carries fragments from three previous employers you will never see.
The improvement available is not a better fund. It is an assembly decision: what is the floor, what covers it, what pools, what travels, and what happens at the end. Those are questions your plan can answer without changing a single line item, and most plans have never been asked them.
Next week: once you accept that risk is assigned rather than eliminated, the question becomes who is carrying what, and whether anybody chose.