One word, four very different risks
The previous guide in this rung made the big distinction: health actuarial work is driven by morbidity, not mortality. Death is a single, unambiguous, once-only event; getting sick is a messy journey with an onset, a duration, a cost, and sometimes a recovery. That one shift in raw material now blooms into a whole shelf of products, and the trap for a beginner is to lump them together as "health insurance." They are not one thing. Each insures a *different consequence* of ill health, and the consequence — not the illness — is what the price is really chasing.
Take one event — a 48-year-old has a heart attack — and watch four products respond in four different currencies. Medical expense insurance pays the hospital, the surgeon, and the pharmacy for the *treatment*. Disability income replaces the *paycheck* she loses during the months she cannot work. Critical illness hands her a *lump sum* simply because the diagnosis happened, to spend however she likes. And years later, if the damage leaves her needing help to bathe and dress, long-term care pays for the *daily caregiving*. Same person, same event, four risks: cost of treatment, lost income, the diagnosis itself, and the need for ongoing care.
Medical expense: small claims, but trend never sleeps
Medical expense insurance is the product most people picture first: it reimburses the cost of doctors, hospitals, and drugs, usually net of a deductible and coinsurance you already met in this rung's loss-modelling work. Its claims are *high-frequency and low-to-moderate severity* — most insureds claim something most years, a flu visit here, a prescription there, occasionally a big surgery. Because almost everyone claims, the pooling is excellent and the year-to-year claim count is stable. The danger is not whether claims happen; it is how fast the *cost per claim* climbs.
That climb is medical trend, and it is the thing that keeps health actuaries up at night. It is *not* the same as ordinary price inflation. It runs on two engines: unit price (each MRI, each hospital day, each branded drug gets pricier) and utilization (people use more services, new treatments appear, providers order more tests). A headline trend of, say, 7% might break down as roughly 3% price plus 4% utilization. The reason it is so feared is that it *compounds*, and it compounds on a number that is already large, so a small error in the assumption snowballs across a multi-year rate filing.
Compounding medical trend on the per-member-per-month (PMPM) claim cost Today's claim cost: 1,000 PMPM Assumed trend: 7% per year (~3% price + ~4% utilization) Year 1: 1,000 x 1.07 = 1,070 Year 2: 1,000 x 1.07^2 = 1,145 Year 3: 1,000 x 1.07^3 = 1,225 <- already +22.5% in 3 years Now suppose true trend was 9%, not 7%: Year 3: 1,000 x 1.09^3 = 1,295 Miss of 2 points/year -> 70 PMPM too low by year 3 (~6% of premium). Multiply by a 100,000-member block x 12 months: ~84,000,000 underpriced.
Disability income: insuring your paycheck, not your bill
Disability income insurance (DI) flips the question. It does not care what your treatment costs; it cares whether you can still earn. For most working people their largest asset is not the house — it is the stream of future paychecks, and DI replaces a slice of that stream when illness or injury stops you working. Two dials govern the whole product, and they are the conceptual heart of this guide. The elimination period is how long you must stay disabled before any benefit begins — a *deductible measured in time*, often 30, 90, or 180 days. The benefit period is how long payments may continue once they start — say 2 years, 5 years, or all the way to age 65.
Feel why these two dials move the price so hard. Lengthen the elimination period and a flood of short, self-resolving disabilities — the sprained back that heals in six weeks — never trigger a claim at all, so the premium drops sharply; the insured is self-funding the common, cheap events and pooling only the rare, long ones. Lengthen the benefit period and you take on the long, expensive tail: a claimant disabled to age 65 is an enormous liability, because DI is not one payment but a *stream* of monthly payments whose value the actuary computes as an actuarial present value — a life-table-style calculation, except the table tracks not death but *staying on claim*.
Two more honest subtleties. The benefit is deliberately capped *below* full salary — pay someone 100% to stay home and you have built a quiet incentive never to recover, a cousin of the moral hazard you met in the foundations rung; 60% keeps a reason to return to work. And the whole thing hinges on the policy's *definition* of "disabled." An own-occupation definition (you cannot do your specific job — a surgeon who can no longer operate) is far more generous, and more expensive, than an any-occupation definition (you cannot do any job at all). The definition, not the diagnosis, is what claims actually turn on, which is why DI claim costs swing with both the contract wording and the economy: people claim more, and stay on claim longer, in a recession.
Long-term care: the product that humbled the profession
Long-term care insurance (LTC) covers something neither medical nor disability cover reaches: the cost of *help with everyday living* in old age — bathing, dressing, eating, moving — or supervision for dementia. This is not treatment; it is care, and it can run for years. A benefit typically triggers when you can no longer perform a set number of activities of daily living (commonly 2 of 6) or have severe cognitive impairment, and then pays a daily or monthly amount, often after an elimination period of weeks. Like DI it is a *morbidity-and-duration* product, but with the duration stretched out brutally long and pushed far into the future.
LTC is famous among actuaries as one of the hardest products ever priced — and an honest cautionary tale worth telling in full, because it shows what happens when several gentle assumptions all drift the wrong way at once. Early LTC was badly underpriced because *three* things went against the insurers simultaneously. First, policyholders lapsed far less than assumed — the pricing quietly counted on many people dropping cover before ever claiming, and they did not, so far more policies survived to claim. Second, people lived and needed care much longer than the morbidity tables assumed — longevity and morbidity improvement, the same trend that is good news for humans and a liability for the insurer. Third, the low-interest era gutted the investment income that was supposed to grow the premiums across the decades-long deferral before claims began.
Critical illness: a lump sum on a diagnosis
Critical illness insurance (CI) is the odd one out, and the cleanest to grasp once you see its shape. It pays a single lump sum — say 100,000 — on first diagnosis of one of a listed set of severe conditions: typically cancer, heart attack, stroke, and a handful of others, provided you meet the policy's precise definition and survive any short waiting period. You then spend the money on anything: lost income, travel to a specialist, a mortgage payment, remodelling a home. It does not reimburse a bill and it does not replace a wage on an ongoing basis; it simply converts a *defined event* into a *fixed cheque*.
This structure makes CI a fascinating hybrid: it *looks like life insurance* (pay a sum on a contingency) but is *morbidity-driven* in its assumptions. The actuary's main job is the incidence of each listed condition by age — how many 50-year-olds are first diagnosed with cancer in a year — rather than any continuance or duration, because once the cheque is paid the claim is closed. Hospital indemnity is its simpler sibling: a fixed amount per day in hospital (say 300 a day) regardless of the actual bill, so it too is fixed-benefit and built from frequency-of-event, not cost-of-care.
Be honest about CI's pitfalls, because they are subtle. The payout hinges entirely on the *medical definition* in the contract — "heart attack" must mean a specific clinical event with specific markers, or borderline cases turn into disputes; as diagnostic technology improves and finds disease earlier, the very same words can quietly start paying out more often than when they were priced. And CI is not a substitute for medical-expense cover: a lump sum is wonderful flexibility but it does not scale with an open-ended treatment bill the way reimbursement does. A buyer who treats CI as "health insurance" has misunderstood which risk they bought.
Reading the four as one map
Step back and the four products line up along two questions you can ask of any health risk. *Does the payout follow a cost, or a defined event?* Medical expense and (loosely) disability follow a cost or a wage; critical illness and hospital indemnity follow an event with a pre-set sum. *Is the claim a one-off, or does it have a duration?* Critical illness is one-and-done; disability and long-term care are *durations* the actuary must value as a stream, where elimination and benefit periods are the dials and the claim cost is incidence times continuance times benefit. Place any new health product on these two axes and you already know most of what makes it tick.
Notice, too, which products carry the medical-trend risk and which do not. Reimbursement products (medical expense) are fully exposed to trend, which is why their rates are revisited annually and a wrong trend assumption is the classic way to lose money. Fixed-benefit products (CI, hospital indemnity) are largely insulated, because the sum was set in the contract — though they carry their own version of "trend" through shifting medical definitions and improving diagnosis. Disability and LTC sit in between: the benefit amount is fixed, but the *duration* and the *incidence* drift with the economy, with medicine, and with how people behave — which is exactly why they are the morbidity products that are hardest to price and to reserve.