The Employment Relationship Is Fracturing: What It Means for Employer-Sponsored Health Coverage

The employer-sponsored insurance system in the United States was designed for a specific kind of worker: full-time, single employer, multi-year tenure. That worker is not disappearing, but the share of the workforce that fits the description is shrinking, and the shrinkage is not a pandemic artifact or a cyclical adjustment. It is demographic, technological, and economic in origin. The data is clear enough that anyone running a TPA, investing in benefits technology, or advising employers on coverage strategy should understand the scale, the trajectory, and the specific populations driving the change.

The Numbers Behind the Fracture

#

The Bureau of Labor Statistics published its most recent Contingent Worker Supplement in November 2024, based on a July 2023 survey of approximately 60,000 households. The headline figures: 7.4 percent of all employed workers were independent contractors on their sole or main job, and 4.3 percent held contingent jobs they did not expect to last. Among workers aged 55 and over, 11.5 percent were independent contractors, compared with 6.9 percent of those aged 25 to 54 and 2.2 percent of those aged 16 to 24. Thirty-six percent of all independent contractors were aged 55 or older, a share far higher than their representation in the overall workforce (Bureau of Labor Statistics, “Contingent and Alternative Employment Arrangements”).

Those figures undercount the professional independent workforce. The BLS survey captures a single week. A fractional CFO who worked for one of her four clients that week shows up as a traditional worker. MBO Partners, which has tracked the independent workforce annually for 15 years, reports 72.9 million Americans working independently in 2025, including 27.6 million full-time. A record 5.6 million independent workers earned more than $100,000 annually, up 19 percent from 2024 and 86 percent from 2020 (MBO Partners, “State of Independence 2025”). This is not the gig economy narrative of ride-share drivers scraping by. The fastest-growing segment of the independent workforce is high-income professionals choosing independence, not being forced into it.

The organizational forms these workers create are visible in the Census Bureau data. The United States had 30.4 million nonemployer businesses in 2023, generating $1.8 trillion in receipts. Nonemployer business formation has outpaced employer business formation in almost every year since 2012, growing an average of 2.7 percent annually versus 1.1 percent for employer firms. The share of all U.S. businesses that are nonemployers expanded from 75.4 percent in 2012 to 78.4 percent in 2023 (Census Bureau, “Nonemployer Statistics”). New business applications hit a record 478,800 per month in 2025, a pace more than four times the pre-2020 average (Census Bureau, “Business Formation Statistics”). Many of these applications will become the 1 to 10 employee firms that are the hardest segment of the small group benefits market (FWD.03).

The Kauffman Foundation’s longitudinal data on entrepreneurship provides the age dimension that the Census aggregates obscure. The share of new entrepreneurs between the ages of 55 and 64 rose from 14.8 percent in 1996 to approximately 25 percent by 2019 and has held near that level since. The monthly rate of new entrepreneurship for the 55 to 64 cohort is 0.38 percent of the adult population, compared with 0.22 percent for the 20 to 34 cohort. Adults aged 55 to 64 have started businesses at a higher rate than adults aged 20 to 34 in every year the Kauffman Foundation has tracked the data, from 1996 through the most recent report (Kauffman Foundation, “Early-Stage Entrepreneurship National Report”).

The coverage gap this creates is measurable. An analysis by the HHS Office of the Assistant Secretary for Planning and Evaluation, using 2022 American Community Survey data, found 16.3 million self-employed workers between the ages of 21 and 64, of whom 2.9 million were uninsured, a rate of 17.9 percent. That rate is down from 30.2 percent in 2011, before the ACA’s main coverage provisions took effect, but it remains roughly 6 percentage points higher than the uninsured rate for all adults in the same age range. Approximately 4 million self-employed workers were enrolled in ACA marketplace coverage in 2022, representing 28 percent of total marketplace enrollment among 21 to 64 year olds, a share nearly three times their proportion of the overall workforce (ASPE, “Marketplace Coverage of Small Business Owners and Self-Employed Workers”).

The BLS data on health coverage among independent contractors confirms the disparity from the worker side: 74.2 percent of independent contractors had health insurance coverage in July 2023, compared with 84.9 percent of workers in traditional arrangements. That 10.7 percentage point gap, applied to the 12 million independent contractors the BLS identified, represents over a million workers without coverage specifically because of their work arrangement, not their income or eligibility (Bureau of Labor Statistics, “Contingent and Alternative Employment Arrangements”).

Why the 55 to 64 Cohort Is the Strategic Inflection

#

The coverage gap for young gig workers draws political attention. The coverage gap for professionals aged 55 to 64 who have left corporate employment draws almost none, despite being larger in dollar terms and more relevant to the level funded and self-funded markets.

This cohort is distinct in four ways that make them strategically important rather than merely sympathetic. They leave corporate employment with accumulated savings and therefore real capacity to pay for coverage. A 58-year-old former VP who starts a consulting practice is not looking for free coverage. She is looking for coverage comparable to what she had, and she has the income to pay for it if a product exists. The Kauffman data shows that the opportunity share of new entrepreneurs in this age group, those starting businesses by choice rather than necessity, has risen from 15 percent in 1996 to over 25 percent, meaning the majority of new business formation in this cohort is voluntary, not a last resort (Kauffman Foundation, “Trends in Entrepreneurship”).

They are forming businesses with actual employees, not just sole proprietorships. The distinction matters for the benefits market. A sole proprietor may buy marketplace coverage. A 3 to 8 person S corp needs a group product. The employer firms being formed by this cohort are disproportionately in the 1 to 10 employee range, exactly the segment where the product gap is most acute (FWD.03).

They are a decade from Medicare. COBRA lasts 18 months. Short-term medical plans are inadequate for a population with the health complexity that comes with age. The coverage problem for this cohort is durable enough to support a multi-year customer relationship, not a bridge product.

They are sophisticated benefits consumers. Three decades of employer-sponsored coverage has taught them what good coverage looks like: broad networks, reasonable cost-sharing, pharmacy benefits that cover their medications, and a service model that does not require a graduate degree in insurance to navigate. The ACA marketplace does not look like what they are accustomed to, and for those with incomes above the premium tax credit thresholds, the cost comparison is brutal. The 2025 KFF Employer Health Benefits Survey reports average family coverage premiums of $26,993 in employer-sponsored plans, with the employer paying roughly 75 percent. The same family buying unsubsidized marketplace coverage pays the full premium, post-tax, and often gets a narrower network (KFF, “Employer Health Benefits Survey 2025”).

The displacement dimension adds urgency. Not everyone in this cohort left corporate employment by choice. AI-driven organizational restructuring is eliminating the middle management, project coordination, and knowledge synthesis roles that were the career home for much of this population. A McKinsey Global Institute report published in November 2025 found that current AI technologies could automate approximately 40 percent of tasks within existing roles, with the impact concentrated in exactly the non-routine cognitive work, including legal analysis, financial modeling, project management, and strategic planning, that was previously considered resistant to automation (McKinsey Global Institute, “Agents, Robots, and Us”). The World Economic Forum’s Future of Jobs Report 2025 projected 92 million jobs displaced by 2030, with clerical, administrative, and middle-management roles among the most affected categories. The workers who hold those roles are disproportionately in the 45 to 64 age range.

What AI Is Actually Doing to the Employment Unit

#

The popular narrative has AI taking jobs, displaced workers becoming gig workers, and gig workers lacking benefits. The narrative is too simple and pessimistic about the wrong thing.

AI is not primarily eliminating jobs in the professional workforce. It is eliminating the organizational layers that justified full-time employment. The distinction matters. A company that employed three project managers, two analysts, and an operations director to run a function can now accomplish the same output with one senior operator and a set of AI tools. The work has not disappeared. The employment relationships that bundled it together have. The five people whose roles were absorbed do not become unemployed. They become independent: consulting, fractional, project-based, or entrepreneurial. They still work. They do not have group health coverage.

The concrete version of this: a fractional COO working with AI tools can serve four companies simultaneously. She performs the same work a full-time COO would perform at any one of them. None of the four companies will offer her group health benefits because none of them employs her full-time. The employment unit, the one-employer, full-time relationship that the ESI system assumes, has fragmented into four client relationships that the coverage infrastructure has no mechanism to serve.

MBO Partners reports that 74 percent of independent workers used generative AI in 2025, up from 65 percent in 2024, and that AI saves independents an average of nine hours per week (MBO Partners, “State of Independence 2025”). AI is not pushing these workers out of the labor market. It is making them more productive in the independent labor market, which accelerates the fragmentation.

There is a tension in this that the rest of the series traces. AI is simultaneously fragmenting the workforce that needs coverage and enabling the administrative automation that could make covering that fragmented workforce economically viable. The technology that is breaking the employment unit apart is also the technology that could bring the cost of administering coverage for very small groups down to a level where the product math works (FWD.06, FWD.07). Whether those two forces resolve in favor of better coverage or wider gaps depends on who builds the infrastructure and how quickly.

The Coverage Infrastructure Is Not Built for This

#

The employer-sponsored insurance system rests on three assumptions: one employer per worker, that employer has enough employees to form a viable risk pool, and the employment relationship is stable enough to support a plan year. None of these assumptions hold for the emerging workforce pattern described above. The mismatch is not a policy failure or an oversight. It is a structural lag. Coverage infrastructure tends to follow economic reality by a decade.

The ACA marketplace is available to this population. It is not designed for them. For workers with income above the premium tax credit threshold, the marketplace means full-price individual coverage with post-tax dollars. The Inflation Reduction Act’s enhanced premium tax credits expired on January 1, 2026 without Congressional renewal, returning marketplace economics to their pre-2021 structure for above-threshold earners. ASPE data shows that 82 percent of self-employed marketplace enrollees claimed a premium tax credit in 2022. The other 18 percent, the above-threshold earners paying full price, are disproportionately the older, higher-income workers who are the most relevant population for level funded and self-funded coverage models.

ICHRA provides a mechanism for an employer to contribute toward an employee’s individual market coverage, but it is only useful where the individual market is adequate. In rural areas with one carrier and two plan options, a $500 monthly ICHRA reimbursement is a tax-advantaged transfer payment, not a coverage solution. FWD.02 examines where ICHRA wins and where it fails, and the conditions under which it is an adequate response to the coverage gap versus a nominal one.

Association health plans were intended to address the pooling problem by aggregating small employers into a risk pool large enough to make underwriting viable. The Department of Labor’s 2018 rule expansion would have enabled associations formed solely for offering coverage. That expansion was largely struck down in 2019 (New York v. United States Department of Labor). What remains is limited to bona fide associations with purposes beyond insurance.

Level funded coverage through an owned S corp is theoretically available to any business with at least one employee. In practice, stop loss underwriting for groups of 1 to 5 lives is individual health underwriting in a group wrapper, priced accordingly. The actuarial math, the administrative cost, and the adverse selection dynamics at micro-group sizes are the subject of FWD.03, which also examines the reinsurance-at-pool-level mechanism that may change the calculation.

The gap, measured as precisely as the data allows: there are approximately 16.3 million self-employed workers aged 21 to 64, of whom 2.9 million are uninsured and approximately 3 million more are in marketplace coverage they describe as inadequate or unaffordable relative to the employer-sponsored coverage they previously had. The subset of that population aged 55 to 64 who have formed or are forming businesses with 1 to 10 employees, have income above ACA subsidy thresholds, and have no group coverage product designed for their situation is smaller but growing faster than any other segment of the benefits market. The Kauffman data on formation rates, the Census data on nonemployer growth, and the MBO Partners data on high-income independent workers all point in the same direction. This population is not a rounding error. It is a market.

The question is not whether the gap is real. The data settles that. The question is whether the coverage products, the administrative infrastructure, and the technology exist to serve this population. The next seven articles in this series address that question from different angles: the three coverage model architectures and which serves whom (FWD.02), the micro-employer product problem and the economics of solving it (FWD.03), the fractional worker coverage gap and the regulatory constraints on closing it (FWD.04), the strategic choices facing TPAs (FWD.05), the technology architecture that a purpose-built system would require (FWD.06), the AI capabilities that are deployable now versus later (FWD.07), and the competitive landscape that determines who builds the infrastructure first (FWD.08).