> ## Content Index
> Fetch the complete content index at: https://www.thefringetheory.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# No shoes to fill
- URL: https://www.thefringetheory.com/greening-rate-aging-workforce-cost/
- Published: 2026-08-02T19:08:24.000Z
- Updated: 2026-08-09T03:02:02.000Z
- Description: Current labor trends are quietly repricing your benefits, and nobody's escalation assumption has caught up.
- Author: Derek Winn
- Tags: Theory

I've been having a run of conversations with my eight year old about what he wants to do for a living. He has a long list of ambitions, some of which I endorse enthusiastically and some of which I'm still working on. But I've told him one thing plainly. If you want to go into a trade, work for somebody else for four years first, and then if you want to start your own business, I'll help you.

Here's why. I grew up in a manufacturing town where the advice was simple and everybody repeated it: go to college, get a good job. The people giving that advice were watching their own jobs get outsourced, some of them overnight. The advice turned out to have a shelf life nobody mentioned. First the tech bubble burst in 2000, and plenty of people holding IT degrees sat unemployed longer than they had planned for, and now AI looms over the careers we were told to pursue in roughly the same way. I'd rather he own something than be told what to train for.

That's the personal version. The professional version shows up in nearly every strategic conversation I have about cost pressure, and it usually arrives at the same place: the client would like to go back to 2019.

Not out of nostalgia. Ask what specifically they want back and the list is fairly consistent. COVID tail risk, which is still working through claims in ways nobody has fully priced. Inflation that ran white hot and pulled wage scales up with it. A labor market where people asked for raises and generally got them, asked for more leave and generally got that too. Interest rates that made every capital decision harder. And for anyone doing federal work, a stretch of spending volatility that made a signed award feel less signed than it used to.

Here's the part that doesn't get said plainly enough. Most of that wasn't a bad year that passed. It was a step up in direct labor and fringe that never stepped back down. Wage scales reset upward and stayed there. Leave programs that got richer to win talent in 2022 are still on the books in 2026, because you generally don't claw a benefit back without paying for it twice. Whatever the market has done since, the burden those years added to a labor rate mostly stayed.

So most employers are already carrying a heavier base than they planned for. What I want to point out is that one more variable has been moving quietly on top of it, in a direction nobody budgets for. Your covered population has an age structure. That structure is held in place by people leaving and being replaced, and when the leaving slows down, the structure drifts older and your cost per employee climbs on its own, before the carrier does anything at all.

Let's break this down.

## The headlines that got my gears turning

"Early-career workers (ages 22-25) in AI-exposed occupations experienced 16% relative employment declines, controlling for firm-level shocks, while employment for experienced workers remained stable."

Brynjolfsson, Chandar and Chen, Stanford Digital Economy Lab, November 2025

Unemployment for recent college graduates sat at roughly 5.7 percent in the first quarter of 2026, with underemployment near 41.5 percent.

Federal Reserve Bank of New York, The Labor Market for Recent College Graduates

The quits rate has been running at or below 2.0 percent per month, right around its long-run average and well off the late-2021 peak of 3.0 percent. Recent employment gains are being driven by a historic drop in separations rather than by new hiring.

BLS Job Openings and Labor Turnover Survey; Indeed Hiring Lab

![A wooden stepladder with the lowest rung missing](https://storage.ghost.io/c/4a/ff/4aff8612-5ecf-4bf5-ab9b-80e2dfd11821/content/images/2026/08/glyph-ladder-transparent-1200.png)

Read those together and you have a labor market where the front door is narrowing and the back door is closing at the same time. The ladder we were all told to climb is a few rungs short at the bottom, and the people already standing on it have stopped climbing off the top. That is a hard thing to explain to an eight year old and a harder thing to plan a career around. It also does something specific to a benefits plan, and it is not the thing most people would guess.

## Two scenarios that look nothing alike

Scenario one is the quiet company. Low turnover, clean claims, flat headcount, a management team that treats retention as a core competency and is generally right about that. Every year the renewal lands a little worse than the trend assumption and nobody can point to a claim that explains it. There isn't one. The population is simply a year older than it was, and nobody left to make room for anyone younger.

Scenario two surprises people more. A company growing fast, hiring hard, claims experience that looks fine, but the hiring is senior. Directors and above, brought in at 45 rather than 28 because the roles are senior roles. That company is adding headcount and adding age at the same time, and its demographic cost per employee deteriorates faster than the quiet company's does. Bigger isn't better here. It's just bigger and older.

Both are arithmetic rather than accounts of anyone in particular, and both are doing what looks right on every dashboard they own.

## The theory

If people aren't quitting, and can't afford to retire, and there's a lid on entry-level hiring, then the mechanism that keeps a covered population young has stopped running. A population renews itself through vacancies, and right now there aren't many shoes to fill. Call it the greening rate: the amount of turnover a population needs, at its current age and hiring mix, just to hold its demographic cost flat.

Every plan has one. Most employers have never calculated it, and a good number are below it right now.

I built a tool to test this, and you should run your own numbers rather than take mine. [Open the calculator.](https://www.thefringetheory.com/greening-rate/)

Bottom line up front, this is a model, not the gospel. Its sourcing and its limits are stated on the tool itself, so I won't relitigate them here.

What matters for the argument is that healthcare cost rises with age, and that this flows into your fringe costs, meaning the benefits dollars sitting inside your labor rate. For a contractor bidding multi-year work, those dollars are locked at bid time. Hiring practices are not universal either, which is why every input is yours to set.

Two things the model does not do. It does not project medical trend, so whatever the carrier hands you stacks on top of everything here. And it prices average allowed cost by age, generally speaking, not the incidence of large claims. Hold that second one, because it comes back later.

## The arithmetic

Take a thousand covered employees at an average age of 42, roughly the median for the American labor force in the BLS projections. Set annual plan cost at $16,000 per employee, in the neighborhood of published per-employee averages. Assume flat headcount, 13 percent annual turnover, half of hires arriving early in their careers, and people working until 67, the Social Security normal retirement age for everyone born after 1959\. Run it five years, which is about a contract, a plan, or a set of option years.

Greening rate this population needs

17.9%

Turnover it's actually getting

13.0%

Cost per employee, five-year drift

+3.4%

Unbudgeted, per employee per year

$540

Cumulative across the group

$1.9M

Expressed as annual trend

+67 bps

Five-year horizon, demographic effect only, before medical trend. Cost index is the federal default standard age curve at 45 CFR 147.102; separation rates scale to median tenure by age from BLS Employee Tenure.

That last line is the one to carry into a budget meeting. Basis points, meaning hundredths of a percentage point, are how your carrier quotes trend. Sixty-seven of them belong on top of whatever renewal number you were handed, and nobody put them there.

Run the same population out ten years and the annual number falls to 34 basis points. That is not the drift going away, it is the drift front-loading. The population is converging on a new, older equilibrium, and most of the damage lands while it travels. The rate fades. The level shift stays.

Now change one input. Hold everything else and set headcount growth at negative 5 percent a year, a mild contraction by 2026 standards. The greening rate climbs to 25.5 percent, because cutting headcount also cuts the openings that would have brought in younger people. Over five years cost per employee runs up 10 percent, or $1,604 a head, while total plan spend falls 15 percent.

That gap is the part worth sitting with. Total spend down fifteen percent reads as a win in any board deck. Cost per employee up ten percent is what actually prices your renewal, your stop-loss, and your fringe rate. A shrinking workforce almost never gets younger.

Run it the other way and the same lever helps. A company growing 10 percent a year holds its average age at 39.8 by year five and 38.0 by year twenty, because growth creates openings without requiring anyone to leave. That is the cheapest greening available, and it is the one nobody buys on purpose.

Two caveats on that, and they matter. It is conditional on the growth continuing: five years of growth followed by a flat decade gives most of the gain back, drifting from 39.8 back to 41.5, because the cohort you hired young ages together. And retention works against it at the margin, since keeping your junior people lowers turnover, which raises the bar. Growth is doing the work there, not retention.

## Recruiting and retention, both ends

Here's the part that surprised me, and it's the reason the tool has a control asking who is doing the leaving.

Turnover isn't one number. Two employers can both report 13 percent and land in opposite places, depending on whether the people walking out are early in their careers or late. Set the same population to lose mostly early-career people, which is the national pattern since median tenure is shortest for the youngest workers, and the greening rate you need is 17.9 percent. Set the exits to fall evenly across ages and it drops to 10.6\. Set them to fall mostly late-career and it's 6.7.

So the retention program aimed at your most junior people is the one that raises your bar. That's an uncomfortable sentence, and I want to be careful with it, because the answer is not to stop retaining young people. The answer is that retention at one end and renewal at the other are the same problem viewed from two sides, and most organizations only staff one of them.

Two places this shows up concretely.

A business that runs last in, first out, whether by policy or by habit, is choosing its exits from the youngest and least expensive end of the curve. That's a defensible way to handle a reduction, and it is also the version that leaves the age structure most exposed, because it takes the people whose absence the greening rate is measuring. If that's your approach, the tail is worth pricing before the cut rather than after.

The other is succession. When there's no visible path upward, the churn doesn't happen at the bottom, it happens in the middle, among the people at 35 and 40 who conclude they're not going anywhere. Those departures are expensive to replace, they're the ones you were counting on to run the place in ten years, and they don't green anything on the way out, since a 38-year-old leaving and a 38-year-old arriving is a wash. Promoting from within isn't only a culture argument. It changes which end of your distribution the door is on.

Which is really a question about where the effort is best spent, and it splits three ways.

Retaining everyone is the default setting, and it is the one that quietly raises your greening rate, because it removes the churn that was renewing the population. Retaining your key people is a different exercise and almost always the right one, since the cost of losing somebody who runs a program or holds a clearance has nothing to do with the age curve. And being the kind of place a healthy flow of young talent wants to join, with somewhere visible to go once they arrive, is the only one of the three that moves this number in the right direction without anyone having to leave.

Most organizations resource the first, assume the second, and hope for the third.

None of that is an argument about age. It's an argument about structure. The people who pick up a new tool first, or take on the harder assignment, or figure out the thing nobody asked them to figure out, are generally the people who were already good at the job. That describes a 29-year-old and a 57-year-old equally well.

> Adoption tracks performance, not birthday. Culture eats strategy, and performance eats age.

Some of the most adaptable people I work with are the ones with the longest tenure, and some of the least are twenty-five. Young at heart counts, and it doesn't show up on a census.

## On the obvious question

To be explicit about it: the ADEA exists, the EEOC enforces it, and none of this is close to a gray area. Age is a protected characteristic and nothing in this piece is an argument for hiring or not hiring anyone on that basis. This model measures; it doesn't prescribe a roster.

What it does suggest is that being a genuinely good place to work, for people at every stage of a career, carries a cost consequence most employers have never quantified. Strong early-career pipelines, real development paths, and the kind of place people want to join are all defensible business practices that happen to move this number in the right direction.

## The fork in the road

There are three readings of what AI is doing to an age structure, and the model accommodates all of them. That's the point of building it with sliders instead of assumptions.

The first is the one the data currently supports. Entry-level roles are the ones being absorbed, and the Stanford work on ADP payroll data is the strongest evidence for it. Your hiring mix shifts older and the drift accelerates.

The second is that this is temporary, and it deserves more weight than it usually gets. The same tools that let an experienced person absorb a junior's work eventually let fewer experienced people absorb each other's. That appears to be underway already. Employment records reported by the Wall Street Journal show managers at public companies fell 6.1 percent between May 2022 and May 2025, against a 4.6 percent decline in executive roles, and Korn Ferry has 44 percent of U.S. employees reporting their organization eliminated managerial levels.

Here's why that matters. Middle management skews 40s and 50s, the steep part of the age curve, so a flattening that removes that layer greens the population and pushes demographic cost the opposite direction from everything above. Two caveats worth keeping: the most-quoted figure in this space, that a fifth of organizations will cut half their middle management, is a 2024 Gartner forecast rather than a measurement, and span of control has been widening since long before generative AI existed, from roughly 8 reports per manager in 2013 to about 12 by 2025.

The third reading is mine, so I'll mark it plainly as a guess. It's tempting to say the people who don't adopt these tools get replaced by the people who do, sorting your population by adaptability rather than age. I can't support that, and one large poll found higher earners adopting faster than junior staff, which is the opposite of the intuition. It is also the reading I'd most like to be right, for the reason above.

Each reading is a bet on which end of the age distribution thins. You can model all three in an afternoon, and the greening rate doesn't care why the openings went away.

## What this doesn't settle

There's a version of this where greening your population doesn't save you anything, and I wrote most of it already.

In [From snout to tail](https://www.thefringetheory.com/from-snout-to-tail/) I laid out that roughly half of an employer plan's spend comes from about 5 percent of members. What matters here is the age shape of that. Children under 10 generate 39 percent of all stop-loss claims above $1 million, more than triple any other ten-year age band, and most of those come from infants under one.

Sit with what that does to this model. The greening rate prices employees on a smooth curve where cost rises with age. The tail doesn't run through employees at all. It runs through their dependents, and it is loudest at the very bottom of the age range. A young workforce is a workforce in its child-bearing years.

So the two ends of this pull against each other, and the model here only measures one of them.

I'm going to chew on that one for a while before I write it, because the honest version needs a source I don't have yet: claimant incidence by age, rather than the average-cost curves this piece runs on. In the meantime it's worth sitting with. If the tail really does run through the youngest people on your plan, then greening a population buys you something on the average and may cost you something at the extreme, and those two facts have to be held at once. I'll be back with it.

## What to do this quarter

If you want to have a genuinely different conversation at the next executive leadership team meeting, do this.

Pull two numbers from HR. Your average employee age and your total separations rate, voluntary plus involuntary. Both sit in your HRIS today, no carrier and no consultant required, and they are the whole input. Run them and find out which side of the bar you're on.

One trap on that second number, because I nearly fell into it myself. The quits rate everybody cites is monthly. Two percent a month annualizes to roughly 23 percent, and quits exclude layoffs, so the national figure is not the number to compare your own turnover against. Use yours.

Ask whoever builds your escalation assumptions what demographic component is in them. Generally the answer is none, because trend surveys don't carry one. If you're bidding multi-year work at a fixed fringe rate, that omission lands on margin.

Look at your hiring mix by level rather than by headcount. If the roles you've stopped backfilling are disproportionately the junior ones, which is what the national data suggests is happening broadly, then your average hire age is climbing even if your hiring volume looks stable.

The Tail

What I can't tell you is what happens when the job hugging ends. Only that the spring is compressed, and that it may release with the same volatility we watched during the Great Resignation, when the quits rate ran to 3.0 percent a month against a long-run average right around 2.

Every population in this model is running an age structure that assumes today's churn continues. But the surveys describing job hugging aren't describing contentment. They're describing people who feel stuck, working longer hours, watching promotions pass. That's a coiled spring, and if the labor market loosens, some meaningful share of those people leave in a fairly short window.

A plan that has spent five years accumulating demographic cost because nobody left would then face a different problem: a wave of departures concentrated among the people most able to move, which in practice skews toward the younger half of the population. The ones who stay are the ones who were going to stay anyway.

I don't know what that does to a covered population. It might green it. It might age it faster than anything in this piece. I haven't found a clean way to model a release that hasn't happened yet, and I'd be suspicious of anyone who says they have.

Fringe Theory is independent and unaffiliated. Views expressed are my own and do not represent those of my employer. Nothing here is legal, tax, medical, or investment advice.

## Sources

1. Brynjolfsson, Erik, Bharat Chandar and Ruyu Chen. "Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence." Stanford Digital Economy Lab, revised November 13, 2025.
2. Federal Reserve Bank of New York. "The Labor Market for Recent College Graduates," first quarter 2026 data.
3. U.S. Bureau of Labor Statistics. Job Openings and Labor Turnover Survey, 2026 releases; Indeed Hiring Lab analysis of the same data.
4. ResumeBuilder.com. Survey of U.S. workers on job hugging, February 2026.
5. U.S. Bureau of Labor Statistics. Employee Tenure Summary, January 2024.
6. Centers for Medicare & Medicaid Services. Default Standard Age Curve, bulletin of December 16, 2016, under 45 CFR 147.102.
7. Milliman Health Cost Guidelines, allowed-cost spread by age, as reported in The Actuary (Society of Actuaries).
8. Commonwealth Fund, rate-band scaling methodology for age curves.
9. Live Data Technologies, manager and executive headcount at public companies, May 2022 to May 2025, as reported by the Wall Street Journal.
10. Korn Ferry, global employee survey on elimination of managerial levels, reported 2026.
11. Gartner, 2024 forecast on AI-driven organizational flattening through 2026.
12. Financial Times poll of 4,000 U.S. and U.K. workers on AI adoption by seniority, reported April 2026.
13. Tokio Marine HCC, A&H Group, 2026 Annual Market Report (June 2026), stop-loss claim severity by age band.
14. U.S. Bureau of Labor Statistics, Labor Force Projections, Table 3.4, median age of the labor force.
15. KFF, Employer Health Benefits Survey, 2025, for per-employee premium context.
16. Reported span-of-control data, 2013 to 2025, as cited in Forbes coverage of organizational flattening, May 2026.
17. Social Security Administration, normal retirement age by year of birth.