Healthcare Cases 2026: The Per Member Per Month Trap

Three parties in a healthcare case, and only one of them pays

Start with the size of the thing, because a public figure bounds every estimate you will make. US health spending "grew 7.2 percent in 2024, reaching $5.3 trillion or $15,474 per person", and accounted for "18.0 percent" of gross domestic product, checked 30 September 2026 at cms.gov. Divide the two published figures and the implied population is 342.5 million, which tells you the series is internally consistent and gives you a per person anchor for any sizing.

Now the equation. A health plan collects a premium per member per month and pays claims out of it. Revenue equals members times premium per member per month times twelve. Medical cost is the same shape. What remains after administration is the margin.

The industry reports in per member per month for a reason worth understanding. Membership churns through the year, so annual totals mix a changing denominator. A per member per month figure is comparable across a plan of 20,000 lives and a plan of two million.

Utilization arrives in a different unit again. Admissions, visits and procedures are counted per thousand members per year. So a case hands you a rate per thousand per year and asks for a cost per member per month, and the conversion between them is where the arithmetic goes wrong.

One term has no analogue in other sectors. The loss ratio, which any other insurer wants low, has a legal floor here. That single rule inverts the usual instinct about cost reduction, and it is the subject of a section of its own below.

Rampart Health Plan, a healthcare case counted per member per month

Rampart runs a commercial health plan in two states. The board wants to know whether a care management program that lowers admissions is worth funding. Here is the position.

1. Membership. 180,000 members, stable across the year.
2. Premium. $512 per member per month.
3. Inpatient admissions. 58 per 1,000 members per year, at $18,400 each.
4. Total medical loss ratio. 86 percent of premium.
5. Administrative cost. 10.5 percent of premium.
6. Market. Small group, so the applicable statutory standard is 85 percent.

Premium first. 180,000 members at $512 is $92.16 million a month, and twelve months of that is $1,105.9 million a year. Member months are 180,000 times 12, which is 2,160,000.

Admissions second, and this is the conversion that matters. A rate of 58 per 1,000 members per year applied to 180,000 members gives 180 thousands times 58, which is 10,440 admissions. Then 10,440 admissions at $18,400 is $192.1 million.

Convert that to the reporting unit. $192.1 million over 2,160,000 member months is $88.93 per member per month. So inpatient care alone is $88.93 of the $512 premium, or 17.4 percent of it.

Total medical cost next. 86% on $512 is $440.32 per member per month. Administration follows: 10.5% on $512 is $53.76. The margin is 512 less 440.32 less 53.76, which is $17.92 per member per month, and 17.92 over 512 is 3.5 percent of premium. Annually that is 17.92 times 2,160,000 member months, which is $38.7 million.

The floor bites downward, and it bites hard

Here is the program the board wants to fund. Care management cuts admissions from 58 per 1,000 to 47 per 1,000. Work it through and then look at what the number means.

The new admission count is 180 times 47, which is 8,460. Then 8,460 admissions at $18,400 is $155.7 million, down from $192.1 million. The saving is $36.4 million a year, or $16.87 per member month. That takes medical cost from $440.32 to $423.45 per member per month, and the loss ratio from 86 percent to 82.7 percent.

A generic profit tree stops here and reports a $36.4 million saving on a $38.7 million margin, close to a doubling of profit. That conclusion is wrong, and the reason is in the rulebook.

Federal rules require insurers "to spend at least 80% or 85% of premium dollars on medical care", and an issuer that misses the standard "is required to provide a rebate to its customers", checked 30 September 2026 at cms.gov. Rampart sits in small group, where the applicable standard is 85 percent. At a loss ratio of 82.7 percent it is 2.3 points under the floor, and 2.3% on $1,105.9 million is $25.4 million owed back.

So the saving is $36.4 million and the rebate is $25.4 million. The program nets $11.0 million rather than $36.4 million, and 70 percent of the gain flows straight back to customers.

Push the thought experiment further to see the shape of the rule. Suppose the program were twice as effective and took the loss ratio to 78 percent. The shortfall against the 85 percent standard is 7 points, and 7% on $1,105.9 million is $77.4 million owed. Against the $38.7 million of margin the plan started with, the rebate is twice the entire profit. Below the floor, saving money on care is a liability.

That is the sentence to carry into the recommendation. Fund the program, and size it so the loss ratio lands above the floor. The optimum is not the lowest achievable medical cost, and a candidate who maximizes savings has optimized against the rule.

On the provider side the price is a published multiplier

Turn the case around and look at the physician group Rampart pays. Its revenue is not negotiated in the ordinary sense. Medicare publishes a dollar figure and multiplies it by a relative weight for each service, and commercial contracts are commonly struck as a percentage of that.

CMS states that the CY 2023 conversion factor was $33.07, that the CY 2024 factor from 1 January was $32.74, and that it arrived at a factor of $33.29 for dates of service from 9 March "by adjusting the CY 2023 CF of $33.07 by 2.93 percent and the budget neutrality adjustment", checked 30 September 2026 at cms.gov.

Do the multiplication the sentence implies. 33.07 times 1.0293 is $34.04. The published factor is $33.29, which sits $0.75 lower. So the budget neutrality adjustment removed 0.75 over 34.04, which is 2.2 percent, and the announced 2.93 percent update delivered rather less than its headline. CMS names the adjustment and does not quantify it, so that 2.2 percent is our arithmetic on its figures.

Then price a service. A visit carrying 4.8 relative value units is paid 4.8 times 33.29, which is $159.79. A commercial contract at 140 percent of Medicare pays 1.40 times 159.79, which is $223.71 for the same work.

Two consequences follow for any provider case. The volume lever and the price lever are not symmetric, because the price is published and changes once a year by an amount the group does not influence. And a group whose payer mix shifts three points toward Medicare loses 0.03 times the 40 percent premium on its commercial rate, which is worth computing before anyone proposes a growth plan.

Where a healthcare answer goes wrong

Our own grader puts four weights on a structure. They are our weights, read from the rubric code in this repository, and CMS endorses no scoring of case answers. What sits behind each weight is on how an answer is assessed.

Drilling the per member arithmetic

Rework Rampart with one driver changed at a time, and say the unit out loud before every product.

Membership first, because it is the lever that looks free. Take membership to 210,000 with everything else held. Member months become 2,520,000, premium becomes 512 times 2,520,000, which is $1,290.2 million, and the margin at $17.92 per member month becomes $45.2 million. Growth of 16.7 percent in lives gave exactly 16.7 percent more profit, because nothing in the per member statement changed. That is the test of whether a plan has operating leverage, and here it has almost none.

Severity second. Hold admissions at 58 per 1,000 and take the cost per admission from $18,400 to $21,000. Inpatient cost becomes 10,440 admissions at $21,000, which is $219.2 million, or $101.50 per member per month. That is $12.57 more than before, against a margin of $17.92. So a 14 percent rise in the price of one service consumes 70 percent of the plan's entire margin.

Then check the assumption holding the whole case up. Membership was stable and every member was enrolled for twelve months. Real books churn, and a plan that loses and replaces a fifth of its members has a different risk profile at the same headcount, because a new member's first year of claims rarely matches a renewing member's.

Our breakeven calculator will solve the threshold algebra when a case asks what membership covers a fixed cost, and the math drill times unit conversions like these and grades Units on its own line. Both are free and need no account.

For the version of this economics where the buyer is a laboratory and the price comes from a different fee schedule, see life sciences cases. For net price and the patent cliff, see pharma cases. For a loss ratio with no floor under it, see insurance cases.