Chapter 18: Health Insurance Underwriting: The ACA Revolution and What Remains of Medical Underwriting

"The whole art of health underwriting used to be deciding who to keep out. Then the law decided that question for us — and we found out how much underwriting was left once we could no longer say no." — constructed line, in the voice of a group-benefits underwriter, the kind of thing you hear at any industry conference where someone who started before 2014 is being honest about how the job changed.

Overview

Most of this book is about the freedom to say no. The underwriter reads the risk, decides whether to accept it, and prices it for what it is. Health insurance is the line where, for the products ordinary people buy on their own, that freedom was taken away on purpose — and studying why teaches you more about adverse selection than any other chapter in the book. For decades, individual health insurance was the most aggressively medically underwritten coverage in America. An applicant filled out a health history, an underwriter combed it for anything chronic, and the company declined the diabetic, rated up the hypertensive, excluded the bad back, and wrote the healthy twenty-five-year-old at a profit. It was, by the narrow logic of risk selection, good underwriting. It was also a system in which the people who most needed coverage were the ones who most reliably could not get it — adverse selection's mirror image, run by the insurer instead of against it. In 2010 the Affordable Care Act declared most of that machinery illegal, and on January 1, 2014, the new rules took hold: in the individual and small-group markets you must now offer coverage to everyone (guaranteed issue) and you may no longer price by health (community rating).

So is there nothing left for a health underwriter to do? That is the question this chapter exists to correct. Underwriting did not vanish; it moved. It moved up-market to large groups, where the law still lets a carrier price an employer's plan on the employer's own claims experience. It moved inside the employer, where any firm large enough to bear its own claims can self-fund and buy stop-loss to cap the tail — and stop-loss is medically underwritten to this day. And it moved into a set of behind-the-scenes mechanisms — risk adjustment, the medical-loss-ratio rule, temporary reinsurance — that do, at the level of the whole market, the job individual underwriting used to do at the level of the single life. The adverse-selection math never went away. The law simply changed who is allowed to manage it, and how.

In this chapter, you will learn to:

• Describe pre-ACA individual medical underwriting and explain why a profitable selection system can still be a social failure.
• Define guaranteed issue and community rating and state exactly what the ACA abolished — and what it left standing.
• Explain why large-group experience rating is where most surviving health underwriting lives.
• Underwrite a self-funded employer's stop-loss program — specific vs. aggregate, attachment points, lasering, run-in/run-out.
• Explain how risk adjustment and the medical-loss-ratio rule replace individual underwriting in a guaranteed-issue pool.
• Articulate the tension between adverse selection and affordable access, and find the line between risk-based pricing and unfair discrimination.

Learning Paths

This chapter is unusual: it is a personal-lines chapter whose most durable underwriting lives on the commercial side, in employee benefits. Read for the regulatory architecture and for the way a market manages adverse selection when individual selection is forbidden.

🏠 Personal Lines: Sections 18.1, 18.2, and 18.5 are your core — what individual underwriting was, what the ACA replaced it with, and the Medicare Supplement market where medical underwriting still operates on individuals. Watch how guaranteed issue turns the adverse-selection problem from the insurer's tool into the market's standing threat. 🏢 Commercial Lines: Sections 18.3 and 18.4 are the heart of the surviving craft — large-group experience rating and self-funded stop-loss. This is real underwriting with real declines, and it is where a benefits underwriter actually spends the day. The Harbor Steel aside lives here. 📊 Analytics: Section 18.6 is your chapter — risk-adjustment models, the medical-loss-ratio constraint, and reinsurance corridors are some of the largest applied-actuarial machinery in insurance, and they exist precisely to do, by formula, what underwriting can no longer do by hand. 📜 Certification: §18.1–§18.3 and §18.6 map to the health/group benefits material in the AINS and CPCU sequences; community rating, guaranteed issue, and the ACA's market reforms recur on exams.

18.1 Before the ACA: individual medical underwriting and its problems

To understand what changed, you have to see clearly what existed before, because the pre-2014 individual health market is the cleanest case study in the entire book of underwriting working exactly as designed and producing a result society decided it could not live with. Hold both halves of that sentence at once; the chapter does not resolve into "the old way was evil" or "the new way is free." It is the genuine collision of two fairnesses, and you cannot underwrite health honestly without feeling the pull of both.

Begin with the mechanics. When a person applied for individual health coverage — coverage they bought themselves, not through an employer — they completed a detailed health questionnaire, often supplemented by attending physician statements (the records request you met in Chapter 17), a prescription-history check, and sometimes a paramedical exam. The underwriter read this evidence with one question in front of them: what will this person cost us in claims, and is that cost something we can price or something we should avoid? Health risk, unlike the mortality risk of Chapter 17, is dominated by frequency as much as severity (Chapter 6): a person with a well-managed chronic condition will not necessarily die early, but they will generate a steady, predictable stream of office visits, prescriptions, and tests, year after year. That predictability is precisely what makes a chronic condition uninsurable in the strict sense of Chapter 1 — when a loss is expected rather than fortuitous, "insurance" collapses into pre-payment with overhead. From the underwriter's chair, a diagnosed diabetic was not a risk to be priced; they were a known future cost.

So the underwriter had four moves, and the pre-ACA individual market used all of them:

1. Decline. Refuse the application outright. Diabetes, a recent cancer history, serious heart disease, HIV — whole categories of applicants were simply uninsurable in the individual market.
2. Rate up. Offer coverage at a surcharged premium reflecting the elevated expected claims — the substandard logic of Chapter 17, applied to morbidity instead of mortality.
3. Exclude the condition. Issue the policy but attach a pre-existing condition exclusion — a rider stating that the policy will not cover claims arising from a named condition the applicant already had. The bad back is permanently carved out; everything else is covered.
4. Impose a waiting period. Cover the pre-existing condition, but only after a defined period (commonly measured in months) during which claims tied to it are excluded — a partial, time-limited version of the exclusion above.

📋 At the Desk The instrument that made all four moves possible was the pre-existing condition exclusion, so define it precisely because the ACA's central reform is its abolition. A pre-existing condition exclusion is a policy provision that denies or limits coverage for a medical condition the insured had (or was treated for, or a prudent person would have sought treatment for) before the coverage took effect. Its honest purpose was anti-selection: without it, a person could wait until they were sick, then buy coverage, and the carrier would pay claims it never collected premium for — the adverse-selection death spiral of Chapter 1 in its purest form. Its dishonest effect was that it punished people for the bad luck of having been sick before, locking them out of coverage at exactly the moment coverage mattered. Both descriptions are true. The exclusion was a real defense against a real problem and a real barrier to real people, and the policy fight of 2009–2010 was a society choosing which of those truths to weight more heavily.

The problems with this system were not subtle, and they compounded. First, job lock: because employer group coverage was guaranteed-issue and individual coverage was not, a person with a chronic condition could not safely leave a job, start a business, or retire early, because the individual market would not take them. Second, the death spiral was always one step away even in the underwritten market — because the people who shopped hardest for individual coverage skewed toward those who expected to use it, carriers underwrote ever more tightly to keep ahead of the selection, which pushed more healthy people to skip coverage, which worsened the pool further. Third, and most visibly, the system produced a large population of the uninsurable — people who, through no fault of their own, simply could not buy coverage at any price. States tried to patch this with high-risk pools (subsidized, separate pools for the uninsurable), but these were chronically underfunded and expensive. The pre-ACA individual market was, in the language of this book, a market where the insurer had won the adverse-selection war so completely that it had underwritten a large share of the population out of the pool entirely.

⚖️ Compliance Corner Note what was already illegal before the ACA, because it sharpens the line the whole chapter walks. Even under aggressive medical underwriting, a health insurer could not lawfully decline or rate an applicant on the basis of a protected class — race, religion, national origin (Chapter 4's unfair discrimination doctrine). Underwriting on health status was legal; underwriting on protected characteristics never was. The ACA did something different and larger: it took a factor that was a legitimate, actuarially sound risk predictor — your actual health — and declared that the social cost of pricing on it outweighed the actuarial benefit. That is a different kind of decision from banning a proxy for race. It is society judging that, for this one product, access matters more than actuarial precision. Chapter 35 takes up that judgment in full; flag it here as the deepest version of the actuarial-fairness vs. social-fairness tension in the entire book.

18.2 The ACA's revolution: guaranteed issue and community rating

In March 2010 the Affordable Care Act (ACA) became law, and its core market reforms took effect on January 1, 2014. For an underwriter, the ACA is best understood not as a sprawling statute but as a single, coherent attack on the pre-ACA individual market's central feature: it took away the carrier's ability to select and price by health. Two paired reforms did the work, and you must be able to state each precisely because the rest of the chapter is about what they left standing.

Guaranteed issue is the requirement that an insurer offer coverage to any eligible applicant, regardless of health status — no declines, no medical underwriting, no condition-based exclusions. The applicant who walks in with diabetes, a cancer history, and a heart condition must be offered the same policies as the marathon runner next to them. The pre-existing condition exclusion — the instrument of §18.1 — is simply abolished for ACA-compliant plans. Guaranteed issue is the part of the law most people know by its consequence: you can no longer be turned down for being sick.

Community rating is the companion that prevents guaranteed issue from being gutted by price. Define it carefully, because it is one of this chapter's owned terms and its variants matter. Community rating is a pricing method in which premiums are set for a whole community of insureds rather than for each individual's risk — everyone in the rating area pays the same base rate, regardless of health. Pure community rating would let premium vary by nothing personal at all. The ACA uses adjusted (or modified) community rating, which permits premiums to vary only on a short, enumerated list of factors and forbids everything else. In the individual and small-group markets, an ACA-compliant premium may vary by — and only by — a handful of allowed factors:

ACA ADJUSTED COMMUNITY RATING — what a premium MAY vary on (individual & small-group)   [Tier 1: real law]

  ALLOWED to affect premium                FORBIDDEN to affect premium
  ─────────────────────────────────        ─────────────────────────────────
  • Age (within a capped ratio,            • Health status / medical history
    older-to-younger)                      • Claims experience
  • Tobacco use (within a capped           • Gender
    surcharge)                             • Pre-existing conditions
  • Geographic rating area                 • Occupation / industry (for individual)
  • Family size / tier (individual vs.     • Duration since last covered
    family)                                • (essentially: anything about THIS
  • Plan metal tier (actuarial value)        person's own health)

The single most important line in that table is the boundary itself. Before the ACA, an underwriter's whole value was on the left-hand side of a very different table — health, history, claims. After the ACA, in these markets, health is on the forbidden side. The age ratio is capped (an older adult may be charged only a bounded multiple of a younger adult's premium, compressing the true age-cost curve), and the tobacco surcharge is capped. Everything that made individual underwriting a craft — reading the APS, weighing the build, excluding the condition — is, for these products, gone.

📋 At the Desk Guaranteed issue and community rating are a package, and seeing why they must travel together is the single best lesson this chapter offers about adverse selection. Guaranteed issue without community rating would be hollow: a carrier forced to offer coverage to a diabetic could simply price the offer at \$40,000 a year and achieve a decline by other means. Community rating without guaranteed issue would be equally hollow: a carrier forced to charge one price could simply decline every sick applicant and community-rate only the healthy. You need both clamps, or the underwriter's instinct to manage selection reasserts itself through whichever lever is left open. The ACA closed both levers at once. The predictable consequence — the whole reason the rest of the law exists — is that with both levers closed, nothing the carrier does keeps healthy people in the pool. That job had to be handed to something else.

That "something else" is the part of the ACA an underwriter must understand, because it is the adverse-selection management that replaced individual selection. If a healthy twenty-six-year-old can buy guaranteed-issue, community-rated coverage the day they get sick and not before, the rational move is to wait — and a pool of people who all wait until they are sick is uninsurable at any community rate. The ACA's original answer had three legs:

• A coverage incentive (originally a tax penalty for going uninsured — the "individual mandate") meant to keep healthy people in the pool rather than waiting to get sick. The federal penalty was later reduced to zero, and the practical strength of this leg has varied; some states have enacted their own.
• Premium subsidies for lower-income buyers, which make the community-rated premium affordable enough that healthy people actually enroll rather than skip — subsidies are, in adverse-selection terms, a way to buy the good risks back into the pool.
• Limited enrollment windows (open enrollment and special enrollment periods tied to life events) so that people cannot buy coverage the afternoon they are diagnosed and drop it the afternoon they recover — the structural descendant of the waiting period, moved from the policy to the calendar.

⚠️ Underwriting Trap The trap here is intellectual, and it catches both critics and defenders of the law. The mistake is to imagine that abolishing underwriting abolishes the adverse-selection problem. It does the opposite: it removes the carrier's primary defense against adverse selection while leaving the incentive to select fully intact on the buyer's side. Everything bolted onto the ACA — the mandate, the subsidies, the enrollment windows, and the §18.6 machinery — is scaffolding erected to hold up a pool that can no longer defend itself through underwriting. Weaken any one strut and the pool tilts: the healthy leave, the remaining pool gets sicker, the community rate rises, and more of the healthy leave. The disciplined way to think about the ACA's individual market is as a permanent, managed standoff against the Chapter 1 death spiral — never a victory over it. Adverse selection was not defeated; it was contained, at the cost of a great deal of public machinery.

18.3 What remains: large-group experience rating

Now the chapter turns, and the turn is the whole point: the ACA reformed the individual and small-group markets, but it left the large-group market — employer plans above a size threshold — largely free to underwrite. This is where most surviving health underwriting lives, and it is real underwriting with real declines, real pricing judgment, and a real combined ratio at stake.

The reason the law treats large groups differently is, at root, an adverse-selection argument turned on its head. Recall from Chapter 1 that adverse selection is a problem of individuals self-selecting into a pool. A large employer group is not a collection of self-selecting individuals; it is a pre-formed pool that exists for reasons having nothing to do with health. People do not take a job at a 2,000-employee manufacturer in order to get health coverage for their diabetes — they take the job for the work, and the group's health profile is, to a first approximation, just the health profile of a working population. A large group is therefore naturally close to a representative pool, which means the carrier can price it on its own experience without the death-spiral dynamics that make individual experience rating so dangerous. The larger the group, the truer this is — which is the credibility logic of Chapter 10, applied to morbidity.

Large-group experience rating is the practice of setting a group's premium primarily from the group's own historical claims experience, blended with a manual (book) rate according to credibility. This is the same experience-rating machinery you met in Chapter 11, now applied to health claims:

GROUP HEALTH EXPERIENCE RATING — the credibility blend            [constructed teaching example]

  group's own claims experience (per member per month)  ████████████████  $480 PMPM
  manual / book rate for the class & area               ██████████        $430 PMPM

  credibility Z assigned to the group's own data:  Z = 0.70  (a ~1,200-life group)

  blended expected claims = Z × own + (1 − Z) × manual
                          = 0.70 × $480  +  0.30 × $430
                          = $336 + $129  =  $465 PMPM

  + administrative load, margin, pooling charge, stop-loss/large-claim pooling → the premium.

Read the example as the credibility statement it is. A 1,200-life group is large enough that its own experience carries substantial weight (here, 70%), but not so large that the carrier ignores the book rate entirely — a single catastrophic claimant in a small year could otherwise swing the price wildly. A 50,000-life group might be assigned credibility at or near 1.0: it is its own pool, and the carrier prices it almost purely on its experience. A 60-life group is on the other side of the small-group line and is, in ACA-compliant terms, community-rated — too small for its own experience to be credible signal rather than noise, exactly the small-sample problem Chapter 10 warned you about. The size thresholds that sort groups into "small" (community-rated) versus "large" (experience-rated) are set by regulation and vary, but the logic is pure credibility: experience-rate a group only when its own data is credible.

🤖 Model vs. Judgment Health is one of the most heavily modeled lines in insurance — predictive models score group renewals, flag emerging high-cost claimants, and forecast trend — yet large-group underwriting remains stubbornly judgment-dependent at the margins, and seeing why sharpens the book's central theme. A model reads a group's claims history and projects forward, but it cannot easily see that last year's \$900,000 claimant has recovered and left the company, that the employer just acquired a 300-person division with a different risk profile, or that a new on-site clinic and wellness program are about to bend the trend. The underwriter prices the group the model sees plus the group the model can't — the plan-design changes, the demographic shifts, the one-time claims that will not recur, the broker's intelligence about the employer's intentions. The model owns the base projection; the underwriter owns the adjustments and the defense of them. This is the same division of labor you will see formalized for property and casualty in Chapter 32 — the model proposes, the underwriter disposes — and health is one of the places it is most mature.

The large-group underwriter's day, then, looks like the underwriting craft of Part II applied to a benefits plan. You gather information (the group's claims experience, its census, its plan design, its industry and demographics); you assess the risk (is the trend sustainable, are there shock claims that will or won't recur, is the demographic aging); you do the math (the credibility blend above, plus trend and a pooling charge for the largest claims); you price it (the experience-rated premium plus loads); and you structure the terms (the plan design, the network, and — critically — whether the group should self-fund, which §18.4 takes up). And you can decline: a large group with a runaway trend, a hostile labor situation, or a claims history that signals worsening morbidity can be quoted at a price that says "no," or declined outright. The adverse-selection war the carrier lost in the individual market, it still fights — and can still win — in the large-group market.

🔍 Check Your Understanding 1. Why can a carrier safely experience-rate a 5,000-life employer group but not a single individual or a 50-life group? Name the Chapter 10 concept that governs the answer. 2. The ACA abolished medical underwriting in the individual market but left large groups largely free to experience-rate. Give the adverse-selection reason the two markets are treated differently. (§18.2, §18.3)

18.4 Self-funding and stop-loss underwriting

If you remember one thing from this chapter as a practicing underwriter, make it this section, because self-funded stop-loss is where medical underwriting is most alive, most technical, and most often the actual job. The structure rests on a single, powerful idea that the ACA left fully intact: a large enough employer does not have to buy insurance for its employees' routine health claims at all. It can pay those claims itself, out of its own cash, and insure only the catastrophic tail.

Start with the mechanism. An insured (or "fully insured") plan is the familiar one: the employer pays a premium to a carrier, the carrier collects the premium, pays the claims, and keeps or loses the difference — the carrier bears the risk. A self-funded (self-insured) plan inverts this: the employer keeps the money, pays its employees' claims directly as they come in (usually through a third-party administrator that processes claims), and bears the risk itself. Why would an employer take on that risk? Because for a large, healthy group, most health claims are predictable and routine — the ordinary office visits and prescriptions of §18.1 — and an employer that funds them itself avoids the carrier's risk margin, premium taxes, and certain ACA market rules that apply to insured plans but not to self-funded ones. The employer keeps the savings of its own good experience instead of handing it to a carrier.

But "most claims are predictable" is not "all claims are predictable," and the gap is exactly the variance problem of Chapter 1. A self-funded employer is fine in an average year and ruined in the year three employees develop million-dollar conditions at once. The employer has, in effect, become a tiny insurer with a tiny pool — and a tiny pool, as the law of large numbers tells you, has terrible relative volatility. The answer is to buy insurance for the tail, and that insurance is stop-loss.

Stop-loss insurance is coverage sold to a self-funded employer (not to the employees) that reimburses the employer when its own claims payments exceed a defined threshold. It is, in the language of this book, insurance for a self-insurer — structurally a cousin of the self-insured retention and large-deductible programs of Chapter 12 and, conceptually, of the excess-of-loss reinsurance you will meet in Chapter 27. The employer retains the routine, predictable claims (its "deductible," in effect) and transfers the catastrophic tail to the stop-loss carrier. And here is the point that makes this section the chapter's center of gravity: stop-loss is medically underwritten. A carrier deciding whether to insure an employer's tail — and at what attachment point and price — does exactly what a health underwriter has always done. It reviews the group's claims history, its census, and a disclosure of known large claimants, and it prices the risk it sees.

Stop-loss comes in two forms, and you must keep them distinct:

TWO KINDS OF STOP-LOSS                                            [constructed teaching example]

  SPECIFIC (individual) stop-loss — protects against any ONE claimant's claims exceeding a threshold
    specific deductible (attachment):  $200,000 per covered person, per year
    → the employer pays the first $200,000 of ANY individual's claims; stop-loss pays the excess.
    → defends against the single catastrophic claimant (the premature infant, the transplant, the
      late-stage cancer).

  AGGREGATE stop-loss — protects against TOTAL claims for the whole group exceeding a threshold
    aggregate attachment:  typically set at ~125% of expected total claims for the year
    → if the group's TOTAL paid claims exceed the aggregate point, stop-loss pays the excess.
    → defends against many moderate claims piling up — a bad flu season, a cluster of surgeries —
      that individually never pierce the specific deductible but collectively blow the budget.

Read the two together and you see the structure protects the employer against both failure modes of a small pool: one giant claim (specific) and an unlucky accumulation of ordinary claims (aggregate). A well-structured self-funded plan typically carries both. The specific attachment is the employer's appetite for single-claim risk; the aggregate attachment, conventionally expressed as a percentage of expected claims, is its appetite for total-budget risk. Set them too low and the stop-loss premium approaches the cost of just buying insured coverage (you have transferred so much that you kept little of the savings); set them too high and the employer is exposed to a tail it cannot actually absorb. The underwriter and the broker work with the employer to find the retention the employer can genuinely bear — the same skin-in-the- game logic from Chapter 1 and the same terms-structuring judgment from Chapter 12.

📋 At the Desk Two structural features separate the stop-loss professional from the amateur, and both turn on the mismatch between the policy period and the claims lifecycle. A medical claim is incurred when the care is delivered but paid weeks or months later, so a twelve-month stop-loss contract has to specify which claims it covers. The contract basis is written as two numbers — the incurred window and the paid window. A "12/12" contract covers claims incurred in the twelve policy months and paid within those same twelve months — the tightest, cheapest basis, and the one that leaves the employer exposed to the claims that were incurred late in the year but paid after it ends. A "12/15" or "12/18" basis extends the paid window three or six months past the policy year, picking up that tail. The two related exposures have names worth knowing: run-in coverage reaches back to pick up claims incurred before the policy started but paid during it (it matters when an employer switches stop-loss carriers mid-lifecycle); run-out coverage reaches forward to pick up claims incurred during the policy but paid after it ends. Mismatch these windows and an employer can fall into a gap where a claim is covered by neither the old carrier nor the new one. Getting the contract basis right is unglamorous and is precisely where stop-loss underwriting earns its keep.

⚠️ Underwriting Trap The sharpest practice in stop-loss is lasering, and it is where the line between risk-based pricing and abandoning the insured gets thin. A laser is a higher specific deductible applied to a named, already-known high-cost claimant — the carrier, seeing that the employer already has an employee mid-treatment for a \$2,000,000 condition, sets that person's specific attachment at, say, \$1,000,000 instead of the \$200,000 that applies to everyone else. The logic is impeccable underwriting: that claim is not fortuitous, it is known, and Chapter 1 told you that you cannot insure a loss that is already happening — you can only pre-fund it with overhead. The trap is for the employer, not the carrier: a self-funded employer who does not read the lasers in the renewal can believe it has \$200,000 protection when in fact its largest known claim is carved back to a million dollars of retained exposure. Some employers buy "no-laser" or "no-new-laser" guarantees (at a price) precisely to cap this. The disciplined underwriter prices the known claim honestly — through a laser or a loaded rate — and documents it in plain language so the employer is not surprised. The amateur buries it, and the surprise becomes a coverage dispute and a lost relationship.

There is one more reason self-funding and stop-loss matter to your education beyond their mechanics: they are the cleanest example in the book of a regulatory line creating an underwriting market. Because self-funded plans are governed differently from insured plans (federal employee-benefit law largely preempts state insurance regulation of self-funded plans, the way Chapter 4's McCarran-Ferguson framing would lead you to expect for a federally-regulated arrangement), an entire industry of stop-loss carriers, third-party administrators, and benefits consultants exists in the space the ACA's individual reforms do not reach. Tell an underwriter "the law abolished medical underwriting" and the stop-loss desk down the hall is the standing refutation.

18.5 Medicare Supplement and supplemental products

The chapter has so far drawn a clean line — individual coverage is community-rated and guaranteed-issue, group coverage is experience-rated — but the real market is messier, and one important corner still medically underwrites individuals in the open. That corner is Medicare Supplement insurance, and a family of supplemental products around it, and you should understand it because it is where the old individual-underwriting craft survives on the personal-lines side.

A brief frame: Medicare is the federal health program for people aged 65 and older (and some younger people with disabilities). It pays a great deal but not everything — it leaves deductibles, coinsurance, and gaps. Medicare Supplement insurance (often called "Medigap") is private coverage sold to individuals to fill those gaps. Crucially, Medigap is not an ACA market-reform product in the way individual major-medical is. It operates under its own federal and state rules, and those rules permit medical underwriting outside of specific protected windows. Here is the structure that matters to an underwriter:

• During a person's Medigap open-enrollment period (a defined window around their enrollment in Medicare Part B) and in certain guaranteed-issue situations (for example, when an employer plan terminates), the insurer must issue and may not medically underwrite. Inside these windows, it looks like the ACA individual market: guaranteed issue, no health rating.
• Outside those windows — a person who did not buy Medigap when first eligible and now wants it, or wants to switch to a richer plan — the insurer generally may medically underwrite: ask the health questions, review the history, and decline or rate up the applicant. Here the old craft of §18.1 is alive and legal.

⚖️ Compliance Corner The Medigap structure is a small, elegant lesson in how regulation manages adverse selection without abolishing underwriting — a middle path between the pre-ACA individual market and the post-ACA one. By guaranteeing issue during a defined enrollment window and permitting underwriting outside it, the rules give people a fair, protected chance to get covered when they first become eligible (the access goal) while preventing them from gaming the system by waiting until they are sick to buy in (the anti-selection goal). It is the enrollment-window idea from §18.2, used as a scalpel instead of a sledgehammer. An underwriter working this market must know exactly which window an applicant is in, because the same applicant is guaranteed-issue on one date and fully underwritten on another — and getting that wrong is both a compliance failure and a pricing failure. The lesson generalizes: when a person is allowed to buy is itself an underwriting control, sometimes a more powerful one than whether.

The same medical-underwriting logic appears in several other supplemental products that sit outside the ACA's major-medical reforms — coverages that pay a fixed benefit on a defined event rather than reimbursing medical costs. Critical-illness policies (a lump sum on diagnosis of a covered condition), hospital- indemnity policies (a fixed daily benefit for a hospital stay), and similar products are generally medically underwritten on the individual, because they are not the guaranteed-issue major-medical coverage the ACA reformed. An underwriter touching these products is doing morbidity selection of the classic kind: reading a health history, estimating the probability and cost of the covered event, and declining or rating accordingly. The craft did not die with the ACA; it retreated to the products the reform did not cover.

🔍 Check Your Understanding 1. A 68-year-old who declined Medigap at first eligibility now applies, in good health, for a rich Medigap plan. May the insurer medically underwrite the application? What single fact determines the answer? 2. Why does permitting underwriting outside a guaranteed-issue window actually protect the people who do enroll during the window? Tie your answer to adverse selection. (§18.2, §18.5)

18.6 Risk adjustment, reinsurance, and the mechanics that replace underwriting

Return now to the guaranteed-issue, community-rated individual and small-group markets of §18.2 and ask the question an underwriter cannot help asking: if a carrier can no longer select or price by health, how does it survive enrolling a worse-than-average share of sick people? In a free market it couldn't — a carrier that happened to attract the diabetics would be crushed by claims it could not price for, and the rational response would be to avoid attracting them, which would defeat guaranteed issue. The ACA's answer is a set of behind-the-scenes mechanisms that do, at the level of the whole market, the job underwriting used to do at the level of the single life. Three are worth your understanding; the most important and most durable is the first.

Risk adjustment is a mechanism that transfers money among carriers in a market based on the relative health risk of the people each one enrolled — taking from plans that enrolled healthier-than-average members and paying plans that enrolled sicker-than-average members. Define it as one of this chapter's owned terms and grasp what it accomplishes: it removes the carrier's financial incentive to avoid sick enrollees. If you will be compensated for enrolling a high-risk member through a transfer from carriers who enrolled low-risk members, you no longer have a reason to design your plan, your network, or your marketing to repel the sick. Risk adjustment, in other words, is the institutional substitute for the underwriting decision: it lets the carrier be indifferent to who enrolls, which is exactly what guaranteed issue requires and exactly what underwriting used to make impossible.

The machinery underneath is a risk-adjustment model — a large actuarial/statistical model that assigns each enrollee a risk score based on their demographics and documented diagnoses, so that a member with several serious chronic conditions carries a higher score (and generates a larger transfer) than a healthy member. This is some of the largest applied-statistical machinery in all of insurance, and it is, in a precise sense, underwriting done by formula after the fact rather than by judgment before the sale:

RISK ADJUSTMENT — underwriting moved from the front door to the back office   [conceptual; illustrative]

  PRE-ACA (individual underwriting):
     applicant → underwriter reads health → DECLINE / RATE / EXCLUDE  → carrier avoids the bad risk

  POST-ACA (guaranteed issue + risk adjustment):
     applicant → MUST be issued, community rate → model scores their diagnoses after enrollment
                                                  → money transfers from "healthy-pool" carriers
                                                    to "sick-pool" carriers
                                                  → carrier is made financially INDIFFERENT to the risk

📊 Model vs. Judgment Risk adjustment is the purest case in the book of a model fully replacing an underwriter, and it earns a careful note about what that costs. The strength is exactly the point: by paying carriers for the risk they enroll, the model lets a guaranteed-issue market function without each carrier trying to dodge the sick. But a model that pays on documented diagnoses creates its own incentive — to document diagnoses thoroughly, since a recorded chronic condition raises the risk score and the transfer. This is not underwriting judgment; it is coding and data completeness, and it has become its own discipline and its own area of regulatory scrutiny. The lesson for you as an underwriter is sober: when you remove human selection and replace it with a formula, you do not remove the incentive to win — you relocate it. The game stops being who you enroll and becomes how completely you document what you enrolled. Every automated replacement for judgment in this book carries a version of this warning, and Chapter 35 returns to it.

Two further mechanisms complete the picture. The medical-loss-ratio (MLR) rule requires health insurers to spend a minimum share of premium dollars on actual medical care and quality improvement (rather than on administration and profit), and to rebate the difference to policyholders if they fall short. In the language of this book, the MLR rule is a cap on the expense-and-profit load — it directly constrains the right-hand side of the premium build-up you learned in Chapter 11, putting a regulatory ceiling on how much of the premium can be anything other than claims. It is, in effect, a public limit on the underwriting margin itself. And in the ACA's first years a transitional reinsurance program operated: a temporary, market-wide reinsurance mechanism that reimbursed individual-market plans for a portion of their highest- cost enrollees' claims, cushioning carriers against the catastrophic claimants they could no longer avoid while the new market found its footing. It functioned exactly like the excess-of-loss reinsurance of Chapter 27, but run for the market as a whole rather than purchased by a single carrier — temporary training wheels for a pool learning to stand without underwriting.

🔍 Check Your Understanding 1. Explain, in one sentence each, how risk adjustment and the medical-loss-ratio rule each substitute for a job that individual underwriting used to do. 2. Risk adjustment pays carriers based on enrollees' documented diagnoses. What incentive does that create, and why is it not the same thing as underwriting judgment? (§18.6)

18.7 The enduring tension: adverse selection vs. affordable access

Every section of this chapter has been circling one question, and it is time to state it directly because it is the deepest version, anywhere in the book, of the tension between actuarial fairness and social fairness that Chapter 35 will take up in full. Here it is in its starkest form: actuarially fair pricing and universal affordable access cannot both be fully satisfied in a voluntary market. You can have one or the other in full, and a great deal of the other, but not both completely — and the history of health insurance is the history of a society moving the dial back and forth between them.

Trace the two poles honestly, because an underwriter who can argue only one side does not understand the problem. Actuarial fairness says the price should reflect the risk: the person who will cost more in claims should pay more, just as the coastal property pays more than the inland one, because to do otherwise is to make the low-risk subsidize the high-risk — a hidden tax that the healthy pay so the sick can be covered. By this logic, pre-ACA medical underwriting was fair: each person paid for their own expected cost. Social fairness says that health is different — that a person's diabetes or cancer history is largely not their doing, that being priced out of health coverage is being priced out of survivable life in a way that being priced out of a beach house is not, and that the social cost of letting the sick go uncovered exceeds the actuarial cost of pooling them with the healthy. By this logic, pre-ACA underwriting was a failure, however actuarially sound. Both arguments are coherent. Neither is obviously wrong. The ACA is what one society's answer looks like when it weights the second more heavily than the first for this one product — while leaving the first to govern auto, home, and life.

⚖️ Compliance Corner Notice what makes health the hardest case rather than just another line, because the distinction is the one to carry into Chapter 35. In most lines, the factors an underwriter prices on are at least partly within the insured's control or are morally neutral — driving record, building construction, security systems — and pricing on them is uncontroversial because it both predicts risk and rewards good behavior. Health status is different on both counts: it is largely not within the insured's control (you did not choose the genetic predisposition, the childhood illness, the cancer), and pricing on it does not reward any behavior the insured could have chosen — it simply charges the unlucky more for being unlucky. That is why society treated health underwriting differently from auto underwriting: not because the actuarial logic was weaker (it was if anything stronger — health is a superb predictor of health cost), but because pricing on an unchosen misfortune felt, to enough people, like punishing the victim. The line between risk-based pricing and unfair discrimination (Chapter 4) usually runs along protected classes. Health insurance is the line in the book where it ran, instead, along a legitimate risk factor that society decided was too cruel to price on. Hold that distinction; it is the subtlest fairness judgment in insurance.

For the underwriter, the practical lesson of this tension is not a position on the policy debate — that is not your job, and reasonable people land in different places. The practical lesson is where, in this market, your judgment is still wanted, and where it is not. In the individual and small-group markets, the society has decided that access outweighs selection, and your selection judgment is, by design, switched off — replaced by the §18.6 machinery. In the large-group, self-funded, stop-loss, Medigap, and supplemental markets, the underwriting judgment of Parts I and II is fully alive, because in those markets the adverse-selection problem is either naturally contained (a large pre-formed group) or managed by a narrower control (an enrollment window) rather than by abolishing selection. Knowing which market you are in — knowing whether the law wants your judgment or has replaced it — is itself the first underwriting decision in health. And it advances the book's sixth theme as sharply as any chapter: insurance serves a social function, and in no other line has society reached so directly into the underwriting decision to adjust the balance between the pool's protection and a person's access to it.

🗂️ The Underwriting File

The group-health adjacency (a teaching aside). The Harbor Steel file does not change this chapter — nothing in our property/casualty program for the metal-fabrication plant turns on the health rules above. But the account is a useful place to locate health underwriting in the real world of a 180-employee employer, because Harbor Steel almost certainly offers its welders, fabricators, drivers, and office staff some form of group health coverage, and which form it chooses is a live underwriting question — just not one your property/casualty desk decides.

Here is the read. Harbor Steel, at roughly 180 employees, sits in an interesting zone: it is well above the small-group line (so its medical coverage, if insured, would be large-group experience-rated, §18.3, not community-rated) and it is large enough that self-funding with stop-loss (§18.4) is a real option a competent benefits broker would put on the table. A self-funded Harbor Steel would pay its employees' routine claims from its own cash, retain (say, illustratively) the first \$150,000–\$200,000 of any single claimant through a specific stop-loss attachment, cap its total exposure with an aggregate attachment around 125% of expected claims, and medically disclose any known large claimants to the stop-loss carrier — which would price them honestly and might laser a known catastrophic claim. The same skin-in-the-game logic that governs the property deductible (Chapter 1, Chapter 12) governs the stop-loss retention: Harbor Steel keeps the predictable claims it can budget and transfers only the ruinous tail.

What this aside settles, and what it doesn't. It settles nothing about our P&C decision — group health is a separate placement, separately underwritten, and not part of the package we are building. What it adds to your education is the recognition that the same employer is simultaneously a risk we underwrite (its property, liability, workers' comp, auto) and an underwriter of its own health risk (through self-funding) — and that the health underwriting it touches is governed by an entirely different body of rules from the McCarran-Ferguson, state-regulated world of its commercial package (Chapter 4). The running disposition on Harbor Steel is unchanged from Chapter 17: a quote-with-conditions in progress on the P&C program, with the health adjacency noted here only as context. The next chapter leaves the personal lines behind and opens commercial property — the deep dive on Harbor Steel's \$20M building and \$10M business-income exposure — where the file does its heaviest work.

Conclusion

Health insurance is the line where this book's central tension is most fully visible, because it is the line where society reached into the underwriting decision and changed it on purpose. Before the ACA, individual health coverage was the most aggressively medically underwritten product in America — declines, rate-ups, and pre-existing condition exclusions, all of it actuarially sound, all of it producing a population the market had underwritten out of coverage entirely. The ACA's paired reforms — guaranteed issue and community rating — abolished individual medical underwriting in the individual and small-group markets, and the mandate, subsidies, enrollment windows, risk adjustment, the medical-loss-ratio rule, and transitional reinsurance are the public machinery erected to hold up a pool that can no longer defend itself through selection. Adverse selection was never defeated; it was contained.

But underwriting did not disappear — it migrated. It lives, fully and legally, in large-group experience rating, where a pre-formed pool can be priced on its own credible claims; in self-funded stop-loss, where the catastrophic tail is medically underwritten exactly as health risk always was; in Medicare Supplement and the supplemental products that sit outside the ACA's reforms; and in the credibility, the trend, and the terms judgment that no model fully owns. The first underwriting decision in health is knowing which market you are in — whether the law wants your judgment or has replaced it with a formula.

We close Part III here. We began it with the auto policy in everyone's glovebox and end it with the health coverage that touches the most lives and stirs the deepest argument about what insurance is for. The next chapter opens Part IV and the heart of commercial underwriting — commercial property — and the Harbor Steel file, which stepped back through the personal lines as a set of asides, returns to center stage with its \$20M building, its \$10M business-income exposure, and the catastrophe peril that started this whole story.

Key Terms

• Community rating — a pricing method that sets premiums for a whole community of insureds rather than for each individual's health risk; under the ACA's adjusted community rating, individual and small-group premiums may vary only on a short enumerated list (age within a cap, tobacco within a cap, area, family tier, plan tier) and not on health.
• Guaranteed issue — the requirement that an insurer offer coverage to any eligible applicant regardless of health status, with no medical underwriting, declines, or condition-based exclusions on ACA-compliant individual and small-group plans.
• Pre-existing condition exclusion — a policy provision (now abolished for ACA-compliant plans) denying or limiting coverage for a medical condition the insured had before the coverage took effect; historically the individual underwriter's primary anti-selection instrument.
• The Affordable Care Act (ACA) in underwriting — the 2010 law whose 2014 market reforms removed the carrier's ability to select and price by health in the individual and small-group markets (guaranteed issue + community rating), replacing individual selection with risk adjustment, an MLR rule, and enrollment controls.
• Stop-loss insurance — coverage sold to a self-funded employer (not its employees) that reimburses the employer when its own claims payments exceed a defined threshold; medically underwritten, and sold in specific (per-claimant) and aggregate (whole-group) forms.
• Risk adjustment — a mechanism that transfers money among carriers in a market based on the relative health risk of their enrollees, removing the financial incentive to avoid sick members and serving as the institutional substitute for individual underwriting under guaranteed issue.

Spaced Review

1. State precisely what the ACA's guaranteed issue and community rating abolished in the individual market, and explain why the two reforms must travel together rather than being enacted one without the other. (§18.2)
2. A self-funded employer buys specific stop-loss at a \$200,000 attachment and aggregate stop-loss at 125% of expected claims. Explain, in plain terms, which failure mode of a small pool each one defends against. (§18.4)
3. (Reaching back to Chapter 10.) Why can a carrier credibly experience-rate a 5,000-life group's health claims but not a 50-life group's? Name the concept and explain what changes as the group grows. (§18.3, Ch. 10)
4. (Reaching back to Chapter 1.) The ACA removed the carrier's ability to underwrite the individual pool. Using adverse selection and the death spiral, explain why the individual mandate, the subsidies, and the enrollment windows are not optional add-ons but structural necessities. (§18.2, Ch. 1)
5. (The recurring pricing-discipline question.) A benefits underwriter is tempted to win a large-group account by setting its experience-rated premium below the credibility-blended indication "to get the relationship." Would that decision help or hurt the carrier's combined ratio, and when would the consequence appear? Tie your answer to rate adequacy (Chapter 11). (§18.3, Ch. 3, Ch. 11)