After building four infographics out of the 2026 Trustees Reports, the obvious follow-up landed: Social Security and Medicare started small and escalated — what else did?
The answer turned out to be bigger than the thing I had just spent a week documenting, and it is not a tax at all.
Health insurance costs an employer more than both programmes combined
The Bureau of Labor Statistics measures what employers actually spend per hour worked. For private industry in March 2026:
| Component | Per hour | Share of total |
|---|---|---|
| Wages and salaries | $32.60 | 69.9% |
| Health insurance | $3.41 | 7.3% |
| Social Security + Medicare | $2.78 | 6.0% |
| Paid leave | $3.54 | 7.6% |
| Supplemental pay | $1.90 | 4.1% |
| Retirement and savings | $1.57 | 3.4% |
| UI + workers' comp | $0.60 | 1.3% |
| Total | $46.60 | 100% |
30.1 percent of what an employer spends on you is not wages. And health insurance at $3.41 an hour is 1.23 times the entire Social Security and Medicare payroll tax at $2.78.
If you actually carry family coverage, the ratio is far worse. KFF puts the 2025 employer contribution to a family plan at $20,143 a year. The employer's payroll tax on a worker at the 2026 national average wage of $74,046 is $5,664. That is 3.6 times.
The two numbers differ because they measure different things — BLS averages across all workers including part-timers and the roughly one in three private-sector workers whose employer covers nobody, while KFF describes the premium for people who actually have family coverage. Both are correct. Mixing them is how this subject gets argued badly, so the infographic keeps them separate throughout.
The arc: 4.65x nominal, 2.4x real
KFF has surveyed employer premiums every year since 1999. The average family premium then was $5,809. In 2025 it is $26,993.
Adjust both to 2026 dollars and the increase is still 2.4 times — from $11,526 to $27,614. Over the same period the national average wage grew 1.22 times in real terms.
Put the premium against the wage it comes out of and the squeeze is stark: the family premium was 19.1 percent of the national average wage in 1999 and is 37.5 percent today.
The worker's own share is not spared either. The employee contribution to a family plan went from $3,101 to $7,008 in constant 2026 dollars — it 2.26x'd in real terms while real wages grew 22 percent. That is before the deductible, which is a separate and separately growing bill.
The triggering moment, and the part almost everyone gets wrong
Here is where I expected to confirm a familiar story and found something else.
Premiums did not decouple from wages steadily across twenty-five years. They exploded in one window and have been converging ever since. Real compound annual growth, premium minus average wage:
| Period | Premium outgrew wages by |
|---|---|
| 1999–2005 | +7.7 points a year |
| 2005–2010 | +2.4 |
| 2010–2015 | +2.0 |
| 2015–2020 | +0.9 |
| 2020–2025 | −0.6 |
The damage was done between 1999 and 2005. Every period since has narrowed, and over the last five years premiums grew slightly slower than wages — KFF's own summary makes the same point, putting the five-year premium rise at 26 percent against wage growth of 28.6 percent and inflation of 23.5 percent.
So the honest framing is that today's premium is a level problem inherited from an explosion that ended twenty years ago, not evidence of a current sprint. Anyone arguing from the last five years alone will get it wrong in either direction.
As for what caused the level: the structural answer is older than the data. Wartime wage controls in 1942 stopped employers competing on pay, so they competed on benefits instead; a 1943 ruling and the 1954 tax code made those benefits permanently untaxed. That is the subsidy that built the entire employer-based system. What the KFF series can show is only that the modern escalation is already running in 1999, its first year — the inflection sits at or before that, and the data cannot see further back. I am not going to pretend otherwise.
The public-sector gap is real — just not where people point
I went looking for evidence that government employees get dramatically better health subsidies. Part of that held up and part of it did not.
| Employer cost per hour, March 2026 | Private | State & local | Ratio |
|---|---|---|---|
| Retirement and savings | $1.57 | $8.83 | 5.62x |
| Health insurance | $3.41 | $7.52 | 2.21x |
| Total benefits | $14.01 | $25.59 | 1.83x |
| Wages and salaries | $32.60 | $40.82 | 1.25x |
| Social Security + Medicare | $2.78 | $2.81 | 1.01x |
State and local government does spend 2.2 times what private industry spends per hour on health. But the genuinely enormous gap is retirement, at 5.6 times — the pension, not the health plan. And on payroll tax, where the statute is identical for everyone, the two sectors are indistinguishable at 1.01.
A caveat that matters: those are averages per hour worked across all workers in each sector, so the health gap mixes plan generosity with coverage rates, and government coverage rates are much higher. It is not a like-for-like comparison of two identical plans.
Federal was the real surprise. The FEHB government contribution is statutorily the lesser of 72 percent of the programme-wide weighted average or 75 percent of the plan you pick. The 2026 maximum for Self and Family works out to $20,229 a year. The average private employer already pays $20,143 — and 74.6 percent of the premium.
The federal advantage is not the size of the subsidy. It is that the share is written into law and cannot be trimmed at renewal.
The same three mechanisms, everywhere
The programmes in the earlier infographics grew three ways: a rate ratchet, a cap removal, and a frozen threshold. All three recur here.
Frozen threshold. FUTA — and most state unemployment insurance — taxes only the first $7,000 of wages, unchanged since 1983 and never indexed. That $7,000 is worth $22,713 in today's money; the base has lost 69 percent of its real value. It covered 46 percent of the average wage in 1983 and covers 9 percent now. Same mechanism as SSI's $2,000 asset limit, opposite beneficiary.
Cap removal. Medicare's wage ceiling matched Social Security's through 1990, ran at $125,000–$135,000 for 1991–93, and vanished entirely after that — which is why Medicare tax follows your whole income while Social Security stops at $184,500. California did the identical thing thirty years later: SDI's ceiling was removed on 1 January 2024 and the rate went 0.9 to 1.3 percent. Someone earning $300,000 went from about $1,378 to $3,900 with no rate change large enough to explain it.
Rate ratchet. State paid family and medical leave programmes are the current cycle, standing up since 2019 at rates small enough that nobody objects. Social Security also began at 2 percent.
The retirement gap, split properly
That 5.6x retirement figure turned out to be hiding the real finding, so it now has its own infographic.
Split BLS's retirement line into its two halves and the picture inverts. On defined-benefit pensions a government employer spends $8.22 an hour against a private employer's $0.40 — 20.6 times. On 401(k)-type plans the private sector spends nearly twice as much ($1.16 against $0.60). These are not one system funded at two levels; they are two different systems.
Only 14 percent of private-sector workers can even reach a defined-benefit plan, and 53 percent participate in a retirement plan of any kind — against 88 percent of full-time public workers.
The honest counterweight is that most of that $8.22 is not buying today's public employee a pension. Between 2001 and 2024, payments toward normal cost rose 159 percent while payments toward unfunded liabilities rose 2,541 percent. A large share of the public figure is debt service on promises made decades ago.
And private-sector workers sit on both sides of it: they hold the weaker arrangement and help fund the stronger one through state and local taxes. Median property tax went from $1,614 in 2005 to $3,211 in 2024 — 1.99x nominal, but only 1.24x after inflation, against real wage growth of 1.17x. Applying the national average, roughly $166 of that bill goes to public pensions. The grievance is not the size of the cheque; it is the asymmetry behind it.
Read the pension gap infographic → · Download the PDF (6 pages)
Fact-check notes and sources
- BLS, Employer Costs for Employee Compensation, March 2026 — news release and table 1. Private industry: total $46.60/hour, wages $32.60 (69.9%), benefits $14.01 (30.1%), health $3.41, Social Security and Medicare $2.78. State and local government: total $66.41, health $7.52, retirement $8.83. bls.gov
- KFF 2025 Employer Health Benefits Survey — 2025 family premium $26,993, single $9,325, employer share of family $20,143, worker share $6,850, single deductible $1,886. Five-year comparison: premiums +26%, wages +28.6%, inflation +23.5%. kff.org
- KFF premium series 1999–2025 — family premium and worker contribution, exported from KFF's published interactive. 1999 family premium $5,809; 1999 worker contribution $1,563. kff.org
- FUTA — 6.0% rate on the first $7,000, effective 0.6% after the maximum 5.4% state credit (IRS Topic 759). The wage base rose to $7,000 in 1983 and "is not adjusted, or indexed, to increase with inflation and has remained unchanged since 1983" (CRS R44527).
- FEHB — 2026 government contribution is the lesser of 72% of the programme-wide weighted average or 75% of the plan premium; biweekly maximum for Self and Family $778.03, which is $20,229 over 26 pay periods. opm.gov
- California EDD — 2026 SDI rate 1.3%, and "effective January 1, 2024, all wages are subject to SDI contributions." UI wage base $7,000, ETT 0.1%. edd.ca.gov
- Medicare wage ceiling history — SSA contribution-and-benefit-base page: HI matched OASDI for 1966–1990, separate caps of $125,000/$130,200/$135,000 for 1991–93, "after 1993, there has been no limitation on HI-taxable earnings." ssa.gov
- Wages and inflation — SSA national Average Wage Index; CPI series underlying the 2026 OASDI Trustees Report, spliced to CPI-U before 1970.
All real-dollar conversions are mine, computed from the published CPI series, with the nominal figure shown beside them. The 1942–54 origin of the employer health exclusion is standard tax history, not a finding from this data.
Related reading
- The 2026 Trustees Reports in four infographics
- Diversifying AI API spend across providers
- Audit your own web app's security
- ImageMagick for favicon sets and brand images
If the thing you take from this is that most of your compensation is decided before it reaches you — withheld, matched, remitted, and spent on your behalf — that is the argument of The W-2 Trap, and the companion piece is at thew2trap.com.
This post is informational, not financial, tax, or benefits advice. I am not a licensed adviser. It is a reading of public data from BLS, KFF, SSA, IRS, OPM, and California EDD, with sources attached so you can verify any figure yourself.
