Most arguments about retirement systems pick one villain. Pensions disappeared. Or people do not save enough. Or fees are too high. Or markets got harder.
I do not think any single one of those is the story, and I think the reason the problem has been difficult to see clearly is that three changes happened at once and compound each other. Each is documented in its own right. Their interaction is the part that has not been assigned to anyone.
Employment Changed
The first paper established that the worker, rather than the employer, has become the enduring economic unit. Long careers inside one institution gave way to fragmented careers across employers, industries and sometimes countries.
The consequence for preservation is structural. Whatever architecture carries a worker's deferred value now has to survive boundaries that the classical designs never contemplated. A pension designed in 1955 assumed the boundary would be crossed once, at retirement. A modern career crosses it seven or eight times, and each crossing is an opportunity for value and structure to be lost.
Responsibility Changed
The movement from institutional promises toward individually directed accumulation returned investment, behavioral, sequence and longevity responsibility to individuals. That much is well known.
What gets said less often is that it also transferred a professional function. The saver became a required investor.
The nineteenth-century worker was primarily a saver, and a saver's task is abstention and safekeeping. The modern worker has been required to become an investor, and an investor's task is allocation under uncertainty. The migration from the first to the second was largely accomplished between the 1980s and the 2000s, as defined contribution arrangements displaced defined benefit plans, and it handed ordinary people the working content of several licensed occupations: asset allocation, diversification, inflation exposure, sequence risk, fees, rebalancing, withdrawal rates, longevity.
That expectation is a historical anomaly, and naming it as an anomaly rather than as nature is the first step in asking whether compensation architecture has kept pace with the responsibility it now implicitly assigns.
The observation is not mine, and its provenance strengthens it. Merton has argued that defined contribution arrangements ask individuals to perform tasks that defeated professionals. Ambachtsheer has pressed the same point from inside pension governance. What I am adding is placement: the observation belongs inside compensation architecture, where the responsibility is assigned, and not only inside retirement policy, where its consequences are managed.
The capability gap this creates is among the best-measured facts in household finance. Financial literacy research finds that large fractions of adults in every surveyed economy cannot answer elementary questions about compounding, inflation and diversification, which are the entry-level concepts of the function they have been assigned.
One honesty condition on all of this. The pension era is more nostalgically remembered than it deserves. Even at its peak around 1980, the employer defined benefit pension covered fewer than four in ten private-sector workers in the United States. The old system was not a lost universal. It was a partial system whose partial coverage is worth remembering whenever someone proposes returning to it.
Money Changed
The monetary and fiscal environment through which long-duration claims travel has moved through several distinct regimes within living memory.
A worker who entered the labor force in 1980 and retires in 2025 carried their deferred compensation through four distinctly different monetary environments. None was announced in advance. Each changed what preservation required. The environment is not a constant of the problem. It is a variable, and a fast-moving one within a single working life.
Here I have to meet the strongest available objection head-on, because it is a good one.
Monetary change is not new. The worker of 1925 lived through regime change more violent than anything since.
That is true, and it means the interesting claim is not about the amount of change. It is about who absorbs it.
When the 1970s inflation arrived, the archetypal worker's deferred compensation was an institutional promise, and the shock landed first on sponsors, insurers and public plans. Those institutions could pool, negotiate, recapitalize, or fail visibly enough to trigger rescue. Today the equivalent shock lands directly on individual balance sheets, because the responsibility transformation removed the institutional shock absorbers before the monetary environment resumed moving.
Monetary change is old. Unbuffered monetary change, at household scale, is the new condition.
And a buffer, before anyone recruits this observation to a favorite cause, is a function rather than an owner. What absorbed the old shocks was pooling, governance and scale. History has attached those technologies to collectives, and the collective is the proven bearer of the function. Any proposed alternative bearer inherits the burden of proving it can perform the same absorption. That is a standard I intend to hold anything to, including anything I would personally find congenial.
One More Honesty Condition
A list of four regime changes creates a mood, and the realized outcome cuts against that mood, so it has to be said plainly.
A worker who held a diversified default portfolio across those forty-five years retired into one of the best realized preservation outcomes any cohort has enjoyed. The long disinflation delivered capital gains to bonds and equities together, and doing nothing was, for once, richly rewarded. Regime variety is not preservation failure, and the two must not be conflated.
The claim the history supports is narrower and more useful. The architecture did real work in that outcome. Automatic contributions, diversification, tax deferral, employer matches and defaults are what let ordinary workers capture the favorable era at all, and a worker without them captured nothing.
But one favorable realized sequence does not demonstrate that the same architecture would preserve equally well through a different sequence. The era supplied the returns. The architecture helped the worker capture them. Nothing in the realized outcome tells us what either would have delivered had the era declined to cooperate.
That question, about variance across possible sequences rather than the path one fortunate cohort walked, is the question preservation design has to answer.
Why the Interaction Matters More Than Any of the Three
The transformations did not happen in sequence or in isolation. They happened together, and they compound.
The worker became more responsible for preservation at the same time that preservation itself became a more complex problem.
Stated at full length, the modern worker is expected to solve a forty-year preservation problem without institutional investment expertise, across many employers, through potentially multiple economic and monetary regimes, against a liability that cannot be known precisely in advance, while managing investment, inflation, sequence, behavioral, tax, longevity, portability and conversion risks, and while simultaneously living an ordinary life.
The employee did not merely become an investor. The employee became an investor while the environment governing preservation was itself moving.
A word about durability, chosen carefully, because a working paper cannot demonstrate permanence. Nothing here predicts that current conditions will persist indefinitely. The defensible characterization is that these transformations are structural rather than merely cyclical, meaning persistent enough that compensation designers cannot prudently assume a return to the conditions under which the inherited architecture was built.
Design has to proceed on the world as it is.
For an employer, all of this lands on a present decision rather than a historical verdict. The default menu through which the next cohort's deferred pay will travel is being chosen right now, under the same uncertainty about the next four regimes that nobody resolved about the last four.
Next week: what the inherited inventory actually solves, argued as strongly as I can argue it.