Resources

2027 Self-Funded Health Plan Renewals: A Broker’s Framework for Allocating Risk

Health In Tech · November 6, 2026

A 10% health care cost trend is useful context for a renewal. It is not a strategy.

The International Foundation of Employee Benefit Plans projects a median 10% increase in employer medical plan costs for 2027; the second consecutive year its survey has produced that expectation. The more important detail for brokers is what sits underneath the number. Catastrophic claims were cited by 32% of respondents as a primary driver of rising costs, ahead of specialty and costly prescription drugs at 21%, chronic-condition and mental-health utilization at 14%, and provider costs at 13%. Among employers identifying specialty drugs as a primary driver, GLP-1s, autoimmune and inflammatory therapies, cancer drugs, and cell and gene therapies all figured prominently.

At the same time, stop-loss pricing is moving even faster. Segal's 2026 national dataset of 225 plans found an average 12.7% increase in stop-loss premium for groups that maintained similar specific stop-loss benefit levels, up from 9.7% the year before. Segal also reports that the number of claimants generating seven-figure claims has grown an average of 25% annually over the past four years in its medical claims data.

For an experienced broker, the interesting story is not that both numbers are high. It is that they are measuring different parts of the risk.

Medical trends reflect what the plan expects to spend across a broad population. Stop-loss pricing is much more sensitive to the tail of the distribution, where frequency is lower; severity can be extreme, and a relatively small number of claimants can materially alter an employer's financial outcome. When stop-loss reprices faster than the underlying medical trend, the market is telling us something about volatility.

That should change the renewal conversation.

Rather than treating medical, pharmacy, stop-loss, plan design and administration as separate negotiations, brokers should increasingly ask how those decisions work together to determine where the client's risk ultimately lands.

Expected cost is only one dimension of the renewal

Most renewal comparisons are very good at showing expected dollars. They are less consistent at showing the economics of uncertainty.

Consider two stop-loss proposals with relatively similar total expected costs. One offers a lower premium but increases the specific deductible, adds an aggregating specific deductible, or applies a claimant-specific laser. Another carries a higher fixed cost, but transfers more of the employer's downside exposure and provides stronger continuity around known claimants.

A conventional spreadsheet may show which option costs less in the base case.

That is not enough information to decide which option is economically better.

The more complete comparison has at least four dimensions.

The first is expected cost, including claims, fixed expenses and stop-loss premium.

The second is volatility, or how much the employer's result can move away from that expectation when claims develop differently than projected.

The third is liquidity, because a theoretically reimbursable claim can still create a meaningful cash requirement depending on when the employer funds it and when reimbursement occurs.

The fourth is contract optionality, meaning the value of provisions that affect what happens when risk changes later. A no-new-laser provision, renewal protection or another form of contractual certainty may have little apparent value in the expected-cost column today, but considerably more value if an emerging claimant deteriorates before the next renewal.

This is where a lower stop-loss premium can become misleading. If the savings were created by moving more variance, liquidity risk or future repricing exposure onto the employer, the broker did not eliminate cost. The broker changed its form.

Segal's data illustrates the point. Plans maintaining similar specific coverage saw average premium increases of 12.7%, while the broader population, including groups that increased deductibles or introduced aggregating specific deductibles, saw an average increase of 11.5%. The lower rate action among the broader group cannot be interpreted independently of the additional risk those plans elected to retain.

For 2027, that distinction should become explicit in the client recommendation.

A sophisticated stop-loss comparison should not answer simply, "What did we save in premium?" but "What additional balance sheet exposure did we accept to obtain that savings, and was the exchange economically rational for this client?"

"High-cost claims" is becoming too broad a category

The same problem appears in the way the industry discusses large claims.

A high-cost claimant report may place a recurring specialty-drug user, an employee undergoing oncology treatment and a candidate for a cell or gene therapy under the same heading. They may all be expensive, but from a risk-financing perspective they behave very differently.

The International Foundation's 2026 GLP-1 survey found that weight-loss GLP-1 drugs represented an average 11.4% of total annual claims for responding corporate plans in 2025. At that level of penetration, GLP-1 exposure is increasingly a recurring expenditure influenced by coverage criteria, eligible populations, utilization patterns, and purchasing economics.

Cell and gene therapies sit at the other end of the spectrum. EBRI found use remained rare, at 9.2 users per 100,000 enrollees in 2022, while costs were concentrated among a very small number of claimants.

Those are two different financial problems.

Recurring pharmacy exposure is largely an expected cost problem. It calls for scrutiny of utilization, clinical criteria, net cost, and plan design.

Rare multimillion-dollar therapies create a severity problem. For many employers, the principal issue is not how accurately next year's utilization can be forecasted. It is how much of a low frequency but financially disruptive event the organization is willing and able to retain.

Oncology often sits between the two. There may be a known claimant, an identifiable treatment pathway and substantial uncertainty around duration, site of care, therapy progression and total cost.

For brokers, the practical implication is that "high cost" should not determine the strategy. Predictability should.

The first question is whether an exposure is recurring enough to manage through the plan's operating economics, known enough to incorporate into underwriting and disclosure, or sufficiently volatile that risk transfer becomes the dominant consideration.

That framing can also prevent a common mistake: using insurance to solve a purchasing problem or using a cost-management program to solve a tail-risk problem.

They are not interchangeable.

The risk that matters may be created by the contract, not the claimant

There is another source of volatility that receives less attention because it does not appear in the claims file.

Contract language can change the employer's financial exposure even when the medical event is exactly the same.

A claimant may be covered differently because of a laser. A specialty drug may receive different treatment depending on the stop-loss policy's definitions. A claim near the end of the contract period may produce a different financial result because of a contract basis or run-out provisions. A reimbursement that ultimately qualifies may still create a meaningful short-term funding requirement.

These are not clinical risks. They are contractual risks.

That distinction becomes increasingly important as self-funded plans assemble multiple specialized arrangements around the same claim.

The widening spread between average cost and tail risk matters

This approach is particularly relevant because the self-funded market has become so large.

KFF reports that 67% of covered U.S. workers were enrolled in self-funded health plans in 2025, including 80% of workers at firms with at least 200 employees. Among workers enrolled in self-funded plans at firms with 200 or more employees, 75% were in plans with stop-loss protection. KFF also found that 37% of covered workers at firms with 10 to 199 employees were enrolled in level-funded arrangements. When I read that, my takeaway was that the majority of employers on Level-Funded programs were on a BUCA Level-Funded Health Plan, which often means the TPA is owned by one of the BUCA’s, the PBM is owned by one of the BUCA’s, and the same company is also underwriting the stop loss. What difference does that make?

It makes a huge difference.

That last figure deserves attention from brokers.

Level funding can make self-funded economics accessible to smaller groups precisely because substantial stop-loss protection limits retained volatility. As self-funded structures reach employers with less balance-sheet capacity to absorb an adverse claims year, the distinction between expected cost and downside exposure becomes even more consequential.

A 5,000-life employer and a 100-life employer can experience the same individual claim. The financial significance is obviously not the same.

The appropriate question is therefore not whether a deductible or retention level is "competitive" in the abstract. It is whether the distribution of possible outcomes fits the employer's cash resources, appetite for volatility, and tolerance for budget variance.

That is the analysis CFOs ultimately care about.

Transparency will give brokers more data, but also raise the standard for using it

The regulatory environment is reinforcing this shift from obtaining information to interpreting it.

In January 2026, the Department of Labor proposed new PBM compensation disclosure requirements under ERISA section 408(b)(2) for providers serving self-insured group health plans. The proposal would require disclosure of direct and indirect compensation, including manufacturer payments and other remuneration, and would give plan fiduciaries audit rights to test the accuracy of the information provided. The rule remains proposed, not final.

The significance for brokers extends beyond PBMs.

Disclosure by itself does not establish whether compensation is reasonable, whether an economic relationship creates a material conflict, or whether a different arrangement would produce a better result for the plan.

Someone still must interpret the information.

The useful advisory question is not, "Did we receive the disclosure?"

It is, "Which facts in this disclosure should change our evaluation of the arrangement?"

The same principle applies to health care price transparency.

CMS's October 2026 amendments to the Transparency in Coverage rules explicitly acknowledge problems with the existing machine-readable data, including file size, ambiguity, and difficulty comparing information across disclosures. The new rules are intended to improve standardization, accuracy and accessibility, while amendments to the self-service tool provisions begin applying for plan years starting January 1, 2027. CMS will also require additional steps to make machine-readable files easier to locate.

For sophisticated brokers, this creates a distinction worth emphasizing: available data is not necessarily decision-grade data.

A negotiated rate is valuable only in context. How much utilization is attached to it? Is the service clinically substitutable? Does the network actually direct members there? Is quality comparable? Is the difference large enough to justify a plan-design intervention?

The industry is entering a period where the scarcity problem is slowly becoming an interpretation problem.

That should elevate the broker's role, but only if the advisory process can distinguish between data that is merely available and information that should actually change the recommendation.

AI should be judged by whether it preserves underwriting meaning

The same discipline should apply to technology.

NAIC's 2025 survey found that 84% of 93 responding health insurers were already using AI or machine learning in some capacity. At that adoption level, the relevant question is no longer whether an organization uses AI.

The better question is what the technology improves.

In stop-loss placement, automation can reduce duplicate data entry, identify incomplete submissions, structure documents, and normalize proposal information. Those are legitimate gains, particularly in workflows that still depend heavily on spreadsheets, PDFs, and manual handoffs. Milliman's survey of employer stop-loss market participants found predictive modeling was becoming more prevalent, although still not used by most participating carriers.

But automation also introduces a less obvious risk: false comparability.

Two proposals can look normalized in a database while containing materially different contractual meanings. A system may successfully extract a deductible, rate or laser and still fail to convey the significance of a definition, exception, or reimbursement provision.

That is why the standard for broker technology should not be speed alone.

The practical shift for 2027: build a risk budget before shopping the market

The practical takeaway is not that brokers need another report or another category on the renewal spreadsheet.

It is that 2027 renewals should begin with a risk budget before they begin with a market exercise.

Start by identifying which costs are sufficiently predictable that the employer should expect to finance them. Separate known elevated exposures that need active management and underwriting attention. Identify the true tail risks that could create an unacceptable financial outcome. Then examine the contracts that determine whether those risks transfer as intended.

The goal is to be able to explain, in plain language, what the employer is choosing to finance, what it is choosing to insure, what could cause that allocation to change, and why the economics justify the risk being retained. If the recommendation cannot answer those questions clearly, the comparison is not finished.