On a Friday afternoon, a person who has carried responsibility for decades hands in a badge, eats a slice of cake, and becomes, by Monday, someone with no colleagues, no schedule, and no claim on the problems she understood better than anyone. Nothing about human capability changes over that weekend. What changes is a status flag in a payroll system.
This is the cliff-edge retirement: one hundred percent of a working life on Friday, zero percent on Monday. It survives not because anyone defends it on the merits but because it is administratively convenient. Pension systems, headcount budgets, and org charts all prefer binary states. People do not.
The demographic bill has arrived
For most of the twentieth century, the cliff was affordable. There were many young workers for every retiree, and the knowledge economy was small enough that what walked out the door on Friday could be rebuilt by Tuesday. Neither condition holds anymore.
The OECD's Employment Outlook 2025 puts the shift in three numbers. The old-age dependency ratio across member countries, people over 65 relative to the working-age population, rose from 19% in 1980 to 31% in 2023, and is projected to reach 52% by 2060. Without productivity gains or policy change, the OECD projects that per capita GDP growth would slow by roughly 40% because of aging: from about 1% a year in 2006-2019 to 0.6% a year over 2024-2060. And the same report estimates that if countries reduced the exit rate of older workers to the level of the best-performing decile, about half of OECD members would gain at least 0.2 percentage points of annual per capita GDP growth.
Read those together and the cliff stops being a private lifestyle question. Societies that push capable people out of work abruptly, at a fixed age, are choosing slower growth. The design failure has a price tag.
Organizations are pushing, not planning
The comfortable assumption is that older workers leave because they want to. Some do. But the evidence suggests organizations often engineer the exit, sometimes without noticing.
A 2024 study by Ding and Ma applied causal methods (distributional synthetic controls) to 260 million résumés to examine return-to-office mandates at Microsoft, SpaceX, and Apple. The mandates caused disproportionate departures of senior employees, and the effect scaled with the rigidity of the rule: Apple's one-day-a-week requirement shifted seniority distribution by under 4%, Microsoft's 50% attendance rule by more than 4%, and SpaceX's full-time mandate displaced up to 15% of the distribution downward. The seniors who left did not retire to the beach; they disproportionately moved to large direct competitors.
A blanket policy, written with no particular age in mind, functioned as a selective ejection seat for the most experienced people. That is what design failure looks like in practice: nobody decided to lose the veterans, but the system was shaped so that losing them was the path of least resistance.
What a designed exit would look like
The alternative to the cliff is not "working forever." It is a slope: an explicit, negotiated reduction of load, scope, and formal authority over years rather than one weekend. A slope requires three things most organizations lack.
First, roles that can shrink. If the only way to hold expertise is a full-time position with a title, then the only way to reduce is to leave. Work needs to be decomposable: advisory days, defined problems, apprentice pairings.
Second, honest conversations about money and status, held early. The cliff persists partly because the descent is undiscussable; raising it reads as announcing decline. An organization that treats the last decade of a career as a designable phase, with the same seriousness it applies to onboarding, removes the stigma by making the conversation routine.
Third, a successor pipeline that starts while the senior person still has energy to teach, not in the final ninety days.
None of this is exotic. It is the same design discipline companies apply to every other transition they care about: product launches, market exits, leadership changes. The end of a long working life is the one transition still run on a payroll flag and a cake.
The cliff was built for an economy of interchangeable labor and short lifespans. We now have neither. Any structure that discards its most experienced members at a stroke, at a fixed date, regardless of capability or desire, is not honoring tradition. It is running obsolete code.
Sources
- OECD (2025). OECD Employment Outlook 2025. Old-age dependency ratio rose from 19% (1980) to 31% (2023), projected 52% by 2060; aging would slow OECD per capita GDP growth ~40% absent policy change; cutting older-worker exit rates to best-decile levels is worth at least 0.2pp annual growth in about half of member countries. https://www.oecd.org/en/publications/oecd-employment-outlook-2025_194a947b-en.html
- Ding & Ma (2024). Return-to-Office Mandates and Brain Drain (arXiv 2405.04352). RTO mandates at Microsoft, SpaceX and Apple caused disproportionate senior departures, scaling with mandate rigidity (up to 15% of the seniority distribution displaced at SpaceX); leavers moved to large competitors. https://arxiv.org/pdf/2405.04352