The conventional wisdom about long-term care insurance is wrong at the boundary. Self-insure if you’re wealthy sounds clean — and it’s the rule of thumb most fee-only advisors will offer in a fifteen-minute call. The math is actually more interesting than that, and the verdict flips in places where rules of thumb don’t.
What follows is a worked analysis at three net-worth tiers using the same hypothetical policyholder: a 65-year-old male, California resident, currently paying $3,500/year for a policy with a $200/day benefit, 5-year benefit period, and 3% compound inflation rider. Premiums assumed to grow 8% annually (consistent with recent industry rate-filing patterns). Real return on invested wealth: 4.5%. General inflation: 3%. Cost-of-care inflation: 4.5%. Median care duration if needed: 1.6 years (sex-adjusted, per HHS).
Every assumption is disclosed. Every result is reproducible in our scenario calculator. The deep-links below open the calculator with these exact inputs pre-filled — you can verify the math yourself in five seconds, then change any assumption to test sensitivity.
What we’re actually comparing
Two paths over a 17-year remaining life expectancy:
Path A — keep the policy. Pay premiums (rising 8%/yr). Collect benefits if a covered care need arises, capped at $200/day × 5 years (with inflation rider). Net result: cumulative premium PV minus expected benefit PV, in today’s dollars.
Path B — drop the policy and self-insure. Invest the would-be premiums at the real return. Pay any care costs out of accumulated wealth at full state cost (no insurance cap). Net result: PV of premium savings (taxed) minus expected out-of-pocket care cost.
Both calculated in nominal-frame present value, discounted at 7.5% nominal (4.5% real + 3% inflation). The math is the math. What changes by net-worth tier is the asymmetric protection wedge: the residual cost in a stressed-bad-case scenario that you’d have to bear from wealth if you self-insure.
Scenario 1: $1M Net Worth — KEEP
At $1M of investable wealth (excluding primary home), the calculator returns a verdict of KEEP: both expected value AND catastrophic protection favor the policy.
The numbers:
- Expected difference (Path A vs B): approximately +$53,000 favoring keeping the policy
- Catastrophic-protection gap if you self-insure: approximately $340,000 — not covered by your $1M wealth without consuming a third of it
Two things are happening. First, on expected value: the policy is roughly break-even-to-slightly-positive in PV because the 70% probability of needing care (HHS) combined with median 1.6-year duration produces an expected benefit payout that approximates cumulative premiums after discounting. The math doesn’t scream “this is a great investment.’’ It says the policy roughly pays for itself in expectation.
Second, on catastrophic protection: the stressed-bad-case scenario (5-year duration at the 75th-percentile state cost, with care starting around year 3) costs roughly $400K in PV today’s dollars. The policy covers about $60K of that with its $200/day × 5-year cap. The residual $340K shortfall is what your wealth has to absorb if you go bare. At $1M, that’s a 34% wealth event — the kind of catastrophic exposure insurance is genuinely designed to cover.
RUN THIS $1M SCENARIO IN THE CALCULATOR →The verdict here is unambiguous. The expected value math is mildly positive, the catastrophic protection is meaningful, and dropping the policy creates a real capital risk you can’t shrug off. At $1M of investable wealth, self-insurance is mathematically inferior on both dimensions.
Scenario 2: $2M Net Worth — KEEP, narrowing
At $2M, the expected-value math is identical. The policy still has the same approximately +$53K PV margin in your favor. What changes is the catastrophic wedge.
- Expected difference: approximately +$53,000 favoring keep (unchanged from $1M tier)
- Catastrophic-protection gap: still approximately $340,000 in PV stressed cost — but now covered by your wealth without strain
This is the wealth tier where the rule of thumb starts to fail. The policy still wins on expected value, but the catastrophic protection has lost most of its marginal utility. You can absorb the stressed-bad-case from your portfolio without re-architecting your retirement plan. The $340K hit at $2M is 17% of wealth — uncomfortable, not catastrophic.
The question becomes: is +$53K of expected value worth the ongoing premium drag and the inflation-rider erosion risk over 17 years?
For most people the answer is still yes. $53K is real money. And the calculator’s expected-value calculation uses median care duration; if your family history suggests longer than median (women with maternal-line dementia, for instance), the policy’s expected value rises further because your personal duration distribution has heavier right tails than the population median.
RUN THIS $2M SCENARIO IN THE CALCULATOR →For people whose family history suggests shorter-than-median duration (cancer-prone families, abrupt cardiac mortality patterns), the calculator output and the personal hazard distribution diverge in the other direction. Override the median-duration assumption in the advanced settings and re-run; you may see the verdict tip toward DROP at this wealth tier.
At $2M, the policy is still mathematically defensible — but the borderline begins here.
Scenario 3: $5M Net Worth — The conventional wisdom is right, mostly
At $5M, the expected-value math is unchanged: approximately +$53K favoring keep. The catastrophic wedge is irrelevant — a $340K bad-case scenario at $5M is a 7% wealth event, in the range of normal portfolio volatility, not a catastrophic loss.
- Expected difference: approximately +$53,000 favoring keep
- Catastrophic-protection gap: $340K, fully covered by wealth, marginal utility near zero
The calculator outputs a verdict of BORDERLINE: the math marginally favors keeping, but only by an amount that’s noise relative to your wealth. $53K is roughly 1% of $5M.
This is where the standard advice (“self-insure if you can’’) starts to be right — and where most policyholders should also factor in considerations the calculator can’t model:
- Capital efficiency. The premiums you’re paying have an opportunity cost. At $5M with growth-oriented allocation, $3,500/yr compounding to retirement-end is non-trivial in absolute dollars even if small as a percentage.
- Premium-increase tail risk. The 8%/yr premium-increase assumption is the recent industry pattern, not a contractual guarantee. Genworth and others have filed for 30%+ single-year increases on certain in-force books. If your carrier files aggressively, the math gets worse for keeping the policy. Track your carrier on our .
- Behavioral arguments for keeping. Some high-net-worth households keep LTC policies primarily for the structure — the policy creates a defined funding mechanism for care that doesn’t require liquidating other assets at potentially poor moments. That has real planning value beyond the EV calculation.
At $5M, the math is borderline and the decision is increasingly about preferences other than expected value. Most fiduciary-only advisors will recommend dropping at this tier; an honest assessment of the math doesn’t strongly disagree.
The asymmetric protection insight
Here’s what the three-tier analysis makes visible: the expected-value calculation is invariant to your net worth. Probability of need, duration distribution, premium PV, benefit PV — none of those depend on how wealthy you are. They depend on age, sex, state, policy structure, and assumed inflation.
What scales with net worth is the catastrophic protection gap: the residual cost in a stressed-bad-case scenario that the policy doesn’t cover. At $1M that gap is meaningful. At $2M it’s uncomfortable but absorbable. At $5M it’s noise.
Insurance, framed properly, is asymmetric protection against tail outcomes. At low and middle wealth, that protection is the policy’s main value. At high wealth, the protection is redundant with your own balance sheet, and the residual question is whether the expected-value math justifies the ongoing premium cost on its own terms.
What shifts the answer for your specific situation
The default scenario above uses median assumptions. Your actual situation may differ in ways that move the verdict:
- Sex-adjusted duration. Women have a sex-adjusted median care duration of 2.7 years (vs 1.6 for men). Re-run with female sex selected; the policy’s expected value rises ~30-40%.
- Aggressive premium-increase pattern. If your carrier’s historical pattern exceeds 8%/yr, override the assumption upward. The policy’s expected value falls; verdict can flip toward DROP at lower wealth tiers.
- Inflation rider matters more than you think. A 3% rider partially erodes against 4.5% medical inflation. A 5% compound rider keeps pace better but typically costs 30-50% more in premiums. Toggle the rider field and re-run.
- State cost of care. California is an expensive state ($145K/yr nursing facility median). If you’re in Mississippi ($90K) the catastrophic gap shrinks; if in Alaska ($459K) it dwarfs everything else.
- Tax-qualified vs non-TQ. Most modern policies are TQ; benefits up to the per-diem cap ($420/day for 2026 per IRS Rev. Proc. 2025-25) are tax-free. If your $200/day policy is well below the cap, tax drag is zero. If you have a $500/day benefit, the portion above $420 is taxable as ordinary income.
How to actually decide
A defensible framework, given the analysis above:
- Run the calculator with your actual numbers. The verdict is sensitive to your specific premium, benefit, state, and rider. Don’t generalize from the hypothetical.
- Stress the assumptions. Bump premium-increase to 12%, sex to F, duration to p90 (5 years). See whether the verdict survives. If it does, the result is robust.
- Separate the EV question from the catastrophic question. The calculator surfaces both. At low wealth, the catastrophic answer dominates. At high wealth, the EV answer dominates.
- Talk to a fee-only fiduciary. The math is one input. A qualified advisor with full visibility into your tax situation, account types, and estate goals will catch things this calculator can’t model.
Our calculator outputs scenario analysis, not personal recommendations. We don’t sell insurance, we don’t collect commissions, and we don’t accept paid placement. Read the methodology for every formula and source. Email the editor if you find a math error — verified corrections get posted publicly with full transparency.
The cost of running the math is fifteen minutes. The cost of getting this wrong is a five-figure-plus mistake. Run the math.