The Actuarial Problem Below 10 Lives: Why the Math Breaks at Small Group Sizes

Series 02: The Risk Layer | Article 02.08 | Sharp Analysis

Claims Variance and Group Size

#

Stop loss underwriting assumes a distribution of outcomes across a population. As group size shrinks, the gap between expected and actual claims widens until the concept of “expected claims” loses predictive value for any single plan year. The micro-employer coverage problem is fundamentally actuarial before it is product, regulatory, or market. This article establishes the math. Series 04 and LFP-MS.03 address what the market does about it.

Health care claims follow a highly skewed distribution. Most individuals in a given year generate modest costs: preventive care, routine office visits, generic prescriptions. A small percentage generate very large costs: cancer treatment, organ transplants, NICU stays, hemophilia, severe trauma with extended rehabilitation. AHRQ’s Statistical Brief #556, published in March 2024 using 2021 MEPS data, quantifies this concentration. The top 1% of the population ranked by health expenditures accounted for 24% of total spending, with average per-person costs of $166,980. The top 5% accounted for 51.2% of all expenditures. The bottom 50% accounted for less than 3%. The Peterson-KFF Health System Tracker’s analysis of 2023 MEPS data found the top 5% spending an average of $72,918 annually, while the top 1% averaged $150,467.

This skewness means that total claims for any population are dominated by a small number of high-cost events. In a large group, those events are statistically expected and their frequency is predictable. In a small group, they are rare enough that their occurrence in any given year is a matter of chance.

The variance relationship with group size is not linear. For a 500-person group, the law of large numbers applies. The group will contain some high-cost claimants, but total claims will fall within a relatively narrow band of expected in most years. The actuarial prediction is reliable. For a 50-person group, the variance widens. Actual claims could deviate meaningfully from expected in a given year, a range that is manageable with adequate stop loss protection but that requires careful attachment point selection. For a 10-person group, the variance is wider still. Actual claims could easily run 50% above or below expected. The claims fund is either substantially underspent or substantially overrun, and the midpoint (expected claims) is a theoretical construct that actual experience rarely matches.

For a 5-person group, the math breaks. A single high-cost event, one cancer diagnosis, one premature birth, one severe accident, can push actual claims to 200% or 300% of expected. The claims fund designed around $150,000 in expected annual claims absorbs a $400,000 cancer treatment. The variance is so wide that “expected claims” describes the center of a distribution whose tails extend to multiples of the funded amount. The coefficient of variation (standard deviation divided by mean) for total claims increases as group size decreases, with the increase accelerating below 25 lives and becoming severe below 10. Published actuarial research from the Society of Actuaries and the American Academy of Actuaries documents this relationship: at very small group sizes, the standard deviation of total claims approaches or exceeds the mean, meaning that a plan year with claims double the expected amount is within one standard deviation of the prediction.

Stop Loss Pricing at Very Small Group Sizes

#

The variance problem translates directly into stop loss premium that eliminates the economic advantage level funded is supposed to provide.

Stop loss premium is a function of expected claims above the attachment point, plus a risk charge for variance, plus expenses and margin. At larger group sizes, the expected claims component is relatively predictable and the risk charge is moderate. The carrier can estimate with reasonable confidence how many members will exceed the specific attachment point and what the aggregate claims will be. The risk charge covers the residual uncertainty.

At very small group sizes, the expected claims component is volatile and the risk charge must be large enough to cover extreme outcomes. The risk charge for a 5-person group at a $30,000 specific attachment point is proportionally much larger than for a 50-person group at the same attachment point. The probability of extreme deviation is higher because each member represents 20% of the population. One member exceeding the attachment point is not a statistical expectation (as it would be in a 200-person group where two or three specific claims per year are actuarially anticipated); it is a binary event that either happens or does not, and its financial impact on the plan is enormous relative to the funded amount.

The crossover point is the group size at which the total cost of level funded (claims fund plus stop loss premium plus TPA administrative fee) equals or exceeds the cost of comparable fully insured coverage. Below this threshold, the employer pays more for level funded and receives less administrative simplicity. The economic rationale disappears.

Fully insured carriers spread variance across community-rated risk pools containing thousands of lives. The individual small employer’s risk is absorbed into a pool large enough for the law of large numbers to operate. The fully insured premium reflects the pool’s expected claims, not the individual group’s risk profile (in ACA-compliant markets where community rating applies). This pooling advantage is precisely what level funded forgoes in exchange for claims data, potential surplus, and ERISA preemption.

The crossover point varies by market, demographics, and geographic area, but industry estimates place it between 5 and 15 lives for most markets. Below 5 lives, level funded is almost never economically justified. Between 5 and 15 lives, it works only for groups with favorable demographics and clean health histories. Above 15 lives, the economic case for level funded strengthens progressively as group size increases and variance decreases.

Attachment point constraints compound the problem. At very small group sizes, carriers raise minimum attachment points. A 5-person group may not be offered a specific attachment point below $50,000 or $75,000 because the carrier will not accept the frequency of claims above lower thresholds at that group size. Higher attachment points increase the employer’s retention per member. Combined with the small number of members, the employer’s total unprotected exposure becomes a large percentage of expected claims. A 5-person group with a $75,000 specific attachment point and $150,000 in expected annual claims could see two members generate $60,000 each in claims (neither triggering the stop loss) and consume 80% of the claims fund with no stop loss reimbursement.

The aggregate corridor at small sizes presents the same proportional problem. A 125% aggregate attachment point on $100,000 expected claims means a $25,000 corridor. For a 5-person employer whose annual revenue might be $500,000, that $25,000 represents real financial exposure, not a rounding error.

Adverse Selection at the Micro-Employer Level

#

The groups seeking level funded at very small sizes are disproportionately likely to be the worst risks. This adverse selection dynamic further erodes the viability of micro-employer level funded.

A 5-person employer seeking level funded is often motivated by a high fully insured renewal. A high renewal may signal unfavorable claims experience, adverse demographics, or both. The employer with favorable demographics and clean health history may also seek level funded for savings, but their motivation is less urgent. The employer facing a 20% fully insured renewal increase is more actively shopping for alternatives. This asymmetric motivation creates an adverse selection dynamic in the micro-employer level funded market: the groups most eager to enter are disproportionately the groups most likely to generate high claims.

Stop loss carriers recognize this pattern. Their underwriting for very small groups reflects adverse selection assumptions through higher risk charges, more aggressive health screening, and earlier application of lasers. These defensive measures push stop loss pricing higher, further narrowing the economic gap between level funded and fully insured.

The information problem is acute at micro-employer sizes. At 5 lives, the carrier has almost no credible historical data on which to base its underwriting. A health questionnaire completed by 5 people provides limited actuarial information. The questionnaire may identify known conditions, but it cannot predict the undiagnosed cancer, the unplanned pregnancy, or the automobile accident that transforms a healthy group into a high-cost one. One undisclosed condition (a member who does not accurately complete the health questionnaire, whether through omission or ignorance of their own health status) can produce claims that exceed the underwriting assumptions by a factor of two or more.

The renewal spiral compounds the instability. If a micro-employer’s first year of level funded produces favorable claims, the renewal may be competitive and the arrangement persists. If the first year produces unfavorable claims, the renewal may be dramatically higher, with lasers applied to the identified high-cost member. The employer then faces a decision: accept the higher renewal (which may now exceed fully insured), find alternative stop loss coverage (difficult at micro group sizes because every carrier in the market sees the same claims history), or return to fully insured. This cycle creates churn in the micro-employer level funded market. Groups move between level funded and fully insured based on annual claims experience rather than structural fit, generating administrative cost with each transition and producing instability for the employer, the TPA, and the carrier.

The academic literature on adverse selection in health insurance markets supports this analysis. Handel’s 2013 research in the American Economic Review documented the mechanisms through which adverse selection operates in employer health plan choice, finding that selection effects are amplified when plan switching is easy and cost differences between options are visible to the consumer. The micro-employer level funded market exhibits exactly these conditions: switching between level funded and fully insured is simple, and premium differences are the primary decision variable.

The Actuarial Floor

#

There is a group size below which level funded is actuarially unviable. The stop loss premium required to adequately protect the employer exceeds the cost savings from exiting fully insured. Below this floor, the employer pays more for less coverage with more administrative complexity. The floor is not a fixed number. It varies by market, by the demographics of the specific group, by plan design, and by stop loss market conditions. In a soft stop loss market with abundant capacity, the floor may drop to 5 lives for groups with favorable demographics. In a hard market with restrictive carrier appetite, the floor may rise to 15 lives.

In most markets and most conditions, the floor sits between 5 and 15 lives. Below 5 lives, level funded is almost never economically justified. Between 5 and 15 lives, it works selectively: groups with young demographics, no known high-cost conditions, and an employer with the financial capacity to absorb the aggregate corridor. Above 15 lives, the economics improve consistently as group size increases.

The micro-employer market (1 to 10 lives) is the fastest-growing segment of U.S. small business formation. Independent contractors, consultants, small professional practices, retail operators, and service businesses are forming at rates that have accelerated since 2020. If level funded cannot serve this segment because of actuarial constraints, the market has a structural gap between the employers who need affordable group coverage and the risk transfer mechanisms available to provide it.

The gap is not a product design problem that a better TPA platform or a more creative benefits package can solve. It is not a regulatory problem that ERISA preemption or ACA reform can address. It is an actuarial problem. The variance at very small group sizes is a mathematical reality. No amount of product innovation changes the fact that one member out of five can generate claims that consume the entire plan.

Solving the micro-employer coverage problem requires either pooling mechanisms that aggregate micro-employers into larger risk pools (captives as analyzed in LFP-02.07, association health plans, PEOs) or alternative coverage models that do not depend on group-level risk pooling at all (ICHRA, which moves employers to the individual market). Each alternative has structural limitations addressed in their respective series. Captives require governance infrastructure and time horizons that most micro-employers cannot sustain. Association health plans face regulatory uncertainty. PEOs require surrendering co-employment control. ICHRA depends on a functioning individual market in the employer’s geography.

The actuarial floor is the boundary condition for the level funded market. Every series that follows, from employer segmentation (Series 04) to population analysis (Series 06) to product architecture (Series 15), operates above this floor. Below it, the math does not support the architecture, and the market must find other structures to serve the employers the math excludes.