Product Specification#

Status: Draft, 2026-08-20 (all cited sources accessed 2026-08-20).

Scope note. This is a standardized composite specification assembled for reference liability cash-flow modeling. It does not describe any single insurer’s product. Facts carrying a source tag — [S#] (primary product documents: policy conditions (yakkan, 約款), policy booklet (go-keiyaku no shiori, ご契約のしおり), pre-contract disclosure (keiyaku teiketsu-mae kōfu shomen, 契約締結前交付書面), product brochure (商品パンフレット) and product pages) and [R#] (product-specific regulatory, statistical and actuarial references), both numbered per _research/nursing-care.md and resolved in sources.md (same directory; numbering frozen, never renumbered), and [REG-R#] (the cross-product reference library references/regulatory-and-actuarial-references.md, whose own R-numbering is distinct) — were extracted from the cited document. Values marked std are standardizations introduced for the reference implementation; each std table row carries a numbered footnote giving the rationale and the observed range across insurers. Facts the research file could not verify are flagged unverified. The composite is drawn from seven carriers’ current retail nursing-care and dementia products: one carrier’s three-tier contract, the only one for which both the pre-contract disclosure and the full 約款 were retrieved [S1] [S2] [S3]; a second carrier’s five-型 elective chassis with its rider menu and a published rate table [S4] [S5] [S6]; a third carrier’s annuity-only contract on simplified underwriting [S7]; a fourth carrier’s three separate products — lump sum, 要支援 cover and annuity [S8] [S9] [S10]; a fifth carrier’s 要支援1 contract [S11]; a sixth carrier’s accelerated-benefit whole-life chassis [S12]; and a seventh carrier’s stand-alone dementia product [S13]. Only [S1], [S2] and [S12] are contractual or pre-contractual documents; the other ten are consumer product pages, and where such a page is silent this specification says so rather than inferring.

This document states its deltas against the medical product specification, the jplib third-sector chassis. It inherits that chassis’s contract machinery — non-participating (mu-haitō, 無配当) and 無解約返戻金型, carrying no surrender value (kaiyaku-henreikin, 解約返戻金) at any duration; no automatic premium loan (jidō furikae kashitsuke, 自動振替貸付); the grace/lapse/復活 sequence; the declaration (kokuchi, 告知) duty; the third-sector (dai-san-bun’ya, 第三分野) reserving overlay; and the 第三分野標準生命表2018 valuation mortality — and replaces its benefit structure entirely. Medical cover is frequency × severity × limit: a daily amount multiplied by paid days, capped per hospitalization and again in aggregate. Nursing care is incidence into a persistent state: a lump sum on entering a defined care state, an annuity while the insured survives in it, and a premium waiver from a lower care state than either. There is no daily amount (nichigaku, 日額), no per-event day limit (支払限度日数), no 通算 aggregate day limit and no 181-day one-hospitalization rule anywhere in this product. What replaces them is a decrement into a state whose entry is certified by a municipality, not diagnosed by a physician.

And this is private cover written on top of the public scheme, not the public scheme. The public long-term-care insurance scheme (kōteki kaigo hoken, 公的介護保険) is a compulsory social-insurance programme run under the Long-Term Care Insurance Act (Kaigo Hoken-hō, 介護保険法) by municipalities; it pays for services in kind against a certification of need R1. The product specified here is a private 第三分野 insurance contract that pays cash to the insured, and it borrows the public scheme only as a benefit trigger. Nothing in this document describes an entitlement under 公的介護保険.

Product overview and market role#

Nursing-care insurance (kaigo hoken, 介護保険) in its private form is 第三分野 business, the class 保険業法 第3条 makes writable under either a life or a non-life licence REG-R1, and the 金融庁 groups it with 医療保険 and がん保険 as the field paying “insurance money or benefits on disease or injury and for treatment” R10. The dominant design is public-scheme-linked (kōteki kaigo hoken rendō-gata, 公的介護保険連動型): the contract’s payment trigger (支払事由) is written by reference to a grade of public certification, so the insurer outsources the adjudication of “is this person in a care state?” to a municipal administrative process it does not control.

The public scheme the trigger points at. 介護保険法 第7条第1項 defines a state requiring nursing care (yō-kaigo jōtai, 要介護状態) as a state in which, because of a physical or mental impairment, the person is expected to require constant care for all or part of the basic daily-living actions — bathing, toileting, eating and the like — continuously over a period fixed by ministerial ordinance R1 REG-R42. 要支援状態 is the parallel, milder definition for a state needing support to prevent deterioration R1. Two classes of insured exist: 第一号被保険者, residents aged 65 and over, and 第二号被保険者, residents aged 40 to 64 who belong to a public health-insurance scheme R1. The 40–64 class can be certified only where the care state arises from one of the 16 特定疾病 (tokutei shippei, specified diseases) listed exhaustively in 介護保険法施行令 第2条 — terminal cancer, ALS, early-onset dementia, cerebrovascular disease, Parkinson’s disease and twelve others R3 R1. One carrier states the restriction in plain terms on its own consumer page [S8]. That single statutory fact is why almost every company-basis alternative trigger in this product class is written for lives under 65 (see Contractual mechanics).

Certification runs on seven grades — 要支援1, 要支援2 and 要介護1 through 要介護5 — and the grades are defined quantitatively, in the estimated minutes of care per day computed by a ministry-defined method from a 74-item assessment (yō-kaigo nintei-tō kijun jikan, 要介護認定等基準時間) R2:

Band

要介護認定等基準時間

Additional test

要支援1

25–32 minutes

—

要支援2

32–50 minutes

support would materially reduce or prevent worsening

要介護1

32–50 minutes

(the 要支援2 case excluded)

要介護2

50–70 minutes

—

要介護3

70–90 minutes

—

要介護4

90–110 minutes

—

要介護5

110 minutes and over

—

要支援2 and 要介護1 share the same 32–50 minute band and are separated only by the improvement-potential test R2. The certification scale is therefore not a clean severity ladder, and it is not one at exactly the point where several carriers place their trigger.

The public incidence data is the sharp contrast with uklib, and with medical. Where a UK long-term-care model has no public morbidity series at all, and where medical must construct incidence from 患者調査 prevalence, a Japanese nursing-care model can be calibrated from a national census of certified persons, published annually by 厚生労働省 and split by sex, six age bands and all seven grades. At 31 March 2024 there were about 7.08 million certified persons — 6,952 thousand 第一号被保険者 and 131 thousand 第二号被保険者 — against 35,890 thousand 第一号被保険者, a certification rate (認定率) of 19.4%, up from 19.0% a year earlier and from 13.9% in the scheme’s first year R4 R5 REG-R30. The age gradient is the product: 4.3% of the 65–74 population is certified against 31.1% of the 75-and-over population R4 R5, and among 第一号被保険者 the counts run 200 thousand at 65–70 to 2,028 thousand at 90 and over, with women outnumbering men 3.6 : 1 at 90+ R4. Certified persons have grown roughly 2.8-fold in the 23 years since the scheme began R16.

The grade composition is what converts that prevalence into a benefit-triggering prevalence R4 REG-R30:

Band

Certified (thousands)

Share

要支援1

1,020

14.4%

要支援2

996

14.1%

要介護1

1,464

20.7%

要介護2

1,191

16.8%

要介護3

927

13.1%

要介護4

895

12.6%

要介護5

590

8.3%

Total

7,083

100.0%

From which 要介護2以上 = 50.8%, 要介護3以上 = 34.0% and 要介護1以上 = 71.5% of all certified persons R4 REG-R30. One carrier’s own brochure quotes the same 50.8% for an earlier year [S12]. Two caveats travel with the arithmetic: the source rounds to thousands and warns that column sums need not match printed totals, and the 要介護5 share is derived as the residual rather than read, though it is consistent with the printed male 7.2% and female 8.9% sub-shares R4. Utilisation, as a sanity check rather than an incidence measure, ran to 6,754.0 thousand distinct recipients in 令和6年度, drawn from every claim record in the 介護保険総合データベース R6.

Market position. Private nursing-care cover is a growing minority line, not a saturated one. Among households insured with a private-sector life insurer — the survey’s own base, which excludes the postal insurer (minpo kanyū setai, 民保加入世帯), 20.1% hold a 介護保険 or a 介護特約 (a nursing-care 特約 — tokuyaku, rider), up from 16.7% at the previous survey, and 7.6% hold a 認知症保険 or 認知症特約, up from 6.6% R14. Against that, 医療保険・医療特約 stands at 95.1% and ガン保険・ガン特約 at 68.2% R14. The 20.1% is a floor rather than a point estimate: 31.4% of respondents answered 不明 to the nursing-care question REG-R32. Average enrolled 介護給付金月額 is ¥92,000 for the household head and ¥65,000 for a spouse R14. The benefit-adequacy backdrop is the same survey’s account of care actually given: an average duration of 55.0 months (4 years 7 months), with 27.9% of cases at 4–10 years and 14.8% at ten years or more, at a monthly cost averaging ¥90,000 plus a one-off average of ¥472,000 R14. The most common grade among those being cared for was 要介護3, then 要介護2, then 要介護4 R14.

Two honest limits on that duration figure, because a model will be tempted to use it as a termination basis. It surveys people who provided care, not certified persons; and it is truncated, since respondents still caring are counted at elapsed duration R14. No official statistic for the duration of the certified state was found in any register.

Representative specification#

Product identity and issue rules#

Parameter

Representative value

Basis

Design type

介護保険, 公的介護保険連動型, 無配当 and 無解約返戻金型; stand-alone main contract (shu-keiyaku, 主契約) paying 介護一時金 and 介護年金, with 特約 attached

[S1] [S4] [S7] [S8] [S11]

Regulatory class

第三分野 (保険業法 第3条第4項第2号 / 第5項第2号)

REG-R1 R10

Chassis

Stand-alone third-sector cover, no death benefit and no surrender value; the accelerated-benefit whole-life chassis is out of scope

five of seven carriers [S1] [S4] [S7] [S8] [S11] against [S12]; std (1)

Issue age (契約年齢) range

40–79

observed 15–80; std (2)

Age basis

Attained age at 契約日 with the fraction discarded (man-nenrei, 満年齢), incremented at each 年単位の契約応当日

[S1]

Policy term (保険期間)

Whole of life (shūshin, 終身), to the terminal age of the mortality table

[S1] [S4] [S7] [S8] [S10] [S11]

Premium-paying period (保険料払込期間)

Pay for life (終身払)

[S1] [S4] [S7] [S8] [S10] [S11]

Lives basis

Single life only

[S1] [S4] [S7] [S8] [S11]

Death benefit (死亡保険金)

None; the contract terminates on death with nothing payable

[S11] explicit, [S1] [S7] by absence; std (3)

Lump-sum amount (介護一時金額) menu

¥500,000–¥3,000,000 in ¥100,000 steps; composite default ¥3,000,000

[S1] [S8]; std (4)

Annuity base amount (基準介護年金額) menu

¥200,000–¥1,200,000 in ¥100,000 steps; composite default ¥600,000 per year

[S1] [S6] [S7]; std (5)

Underwriting

Written declaration (告知扱い), no medical examination; anyone who holds, has ever held, or has applied for a 要支援 or 要介護 certification is declined, as is anyone resident in a 高齢者向け施設

[S1] [S7] [S11]

Anchor model cell

Male, 契約年齢 60, 終身 / 終身払, 介護一時金 ¥3,000,000 on 要介護2以上, 介護年金 ¥600,000 per year from 要介護3以上 capped at 10 payments, 保険料払込免除 from 要介護1以上, 認知症一時金特約 off, level 月払 premium ¥11,500

std (6)

Footnotes to std rows:

  1. Two chassis are in the market and they are different products, not variants. Five of the seven carriers write stand-alone third-sector cover with no death benefit and (usually) no surrender value [S1] [S4] [S7] [S8] [S11]; one writes nursing care as an accelerated death benefit on a 低解約返戻金型 whole-life contract — 介護保険金 equal to 50% of the sum insured, after which the sum insured is reduced by the same amount, premiums are waived and death cover continues for life [S12]; one sells dementia cover only [S13]. The composite takes the stand-alone chassis: it is the majority design, it is the one whose benefit trigger is the interesting mechanic, and the accelerated design already has a jplib home — it is whole_life with a benefit acceleration, not a third-sector product.

  2. Observed: 満18歳–満79歳 [S1]; 15–69 on one direct channel, 70 and over face-to-face [S6]; 40–79 across all three products at one carrier, 40–75 on its 定期 variant [S8] [S9] [S10]; a published premium table spanning 30–80 with the envelope itself unstated [S7] unverified; not published at two carriers [S11] [S13] unverified. The composite takes 40–79: 40 is the age at which public cover itself begins R1, so it is the age below which the 公的介護保険連動型 trigger cannot fire at all, and 79 is the modal upper bound. The 15 and 18 lower bounds are outliers on products whose company-basis limb does the work at young ages.

  3. No stand-alone product observed carries a death benefit; one carrier says so in terms — 死亡保障はありません [S11]. Two conditional forms exist and are excluded: a 死亡給付金 equal to the first annuity instalment where the insured dies having never received one [S10], and a 死亡時返戻金 equal to the (5%) surrender value [S4]. The composite pays nothing on death, which makes death a pure decrement with no cash flow — the same position as medical, and the opposite of whole_life.

  4. Observed lump-sum ranges: 要介護1一時金額 and 要介護2一時金額 each ¥100,000–¥1,000,000 in ¥10,000 steps [S1]; ¥500,000–¥3,000,000 on one postal plan [S8]; a fixed ¥3,000,000 in the only published rate table [S6]; up to ¥2,000,000 [S9] [S10]; ¥500,000 maximum where the insured is 65 or over at issue [S11]; 介護一時金特約 and 認知症一時金特約 capped at ¥2,000,000 combined [S7]. ¥3,000,000 is the default because it is the amount at which two independent carriers publish rates on nearly the same specification [S6] [S8], so the anchor cell’s premium and its benefit are consistent with each other.

  5. Observed annuity ranges: 基準介護年金額 ¥300,000–¥1,200,000 in ¥60,000 steps [S1]; 介護年金額 ¥200,000–¥1,000,000 in ¥100,000 steps for issue ages 20–60, ¥200,000–¥500,000 for 61–80 [S7]; up to ¥2,000,000 [S10]; a fixed ¥600,000 in the published rate table [S6]. The composite takes the union range on the ¥100,000 step, which is the majority step, and ¥600,000 as the default for the same reason as footnote 4.

  6. No carrier publishes 予定発生率, assumed interest rate (yotei riritsu, 予定利率) or 予定死亡率 for this product, and the regulator confirms there is nothing standard to publish R10; the 算出方法書 is a 基礎書類 filed with the 金融庁 and is not public REG-R2. The office premium is therefore a model-point input, not a computed quantity. The ¥11,500 anchor is built from two published scales at male 60, 月払, 終身/終身払: ¥5,820 for a ¥3,000,000 要介護2 lump sum [S6] and ¥5,817 for a ¥3,000,000 要介護3 lump sum at a second carrier [S8] — an agreement to three yen that is itself worth recording — plus ¥5,730 for a ¥600,000 要介護2 annuity on the 5年確定 basis [S6]. The lump-sum and annuity rates sum to ¥11,550, rounded to ¥11,500: a modeling value, not a quote. The composite’s annuity sits at the higher 要介護3 trigger (cheaper) but runs to ten payments rather than five (dearer), and the two effects are treated as offsetting. Age 60 is one of the three ages — 40, 50 and 60 — at which all three of those carriers publish a rate [S6] [S7] [S8], and is the one in the middle of the composite’s 40–79 issue range.

Premiums#

Parameter

Representative value

Basis

Premium basis

Level 平準払 for the whole of life, 無配当 — no dividend, no premium review, no renewal repricing

[S1] [S4] [S7] [S8]

Rating factors

Sex and 契約日における満年齢 only — no smoker, occupation or amount band was disclosed by any carrier

[S1]

Frequency (払込回数)

Monthly (月払) default; 半年払 and 年払 available, with a refund of the unexpired whole months where premiums cease mid-period

[S1]; default std (7)

Payment route (払込経路)

口座振替 or クレジットカード扱

[S1] [S6]

Rate structure

Not published by any carrier; the office premium is a model-point input, backed by a std incidence basis constructed from 介護保険事業状況報告 in the technical notes

R10; std (8)

Anchor premium

¥11,500 per month

std (6)

Sex differential

Female rates exceed male at every age on the lump-sum shapes, by about 1.09–1.16×; on the 終身年金 shape the female loading reaches 1.65× at age 40 and 1.79× at 69. The 要支援-level whole-life product runs the other way, female slightly below male

[S6] [S8]; counter-example [S9]

Premium waiver (保険料払込免除)

On 要介護1以上 certification (or the company-basis limb at that level), and on 高度障害状態, and on a listed 身体障害状態 reached within 180 days of an accident; waived premiums are treated as paid on each 払込期月の契約応当日

[S1] [S2] [S8]; threshold std (9)

  1. Monthly is the dominant retail mode and the mode of every published rate table found [S6] [S7] [S8] [S9] [S10] [S12]; the composite standardizes on it, which is also why the model runs on a monthly grid. The unearned-premium refund on 半年払・年払 [S1] is treated as an immaterial modal refinement and is not modeled.

  2. The absence of a published pricing basis is not a research gap, it is the regulatory position: for 第三分野 business 「標準死亡率、参考純率といったスタンダードな指標が存在しておらず、 公的なデータや各社の実績等から給付事由ごとその発生率を見込まざるを得ない」 R10. Each insurer estimates incidence per benefit trigger from public data and its own experience, and must disclose the model it used R10. The reference implementation does the same thing in public: the incidence basis is a std construction from the 認定率 and grade composition of R4 and R5, and every value in it carries that provenance.

  3. Waiver triggering at a lower care state than the main benefit is the market pattern and is the reverse of medical, where the base waiver is disability-triggered and independent of the benefit. Observed: waiver on the 要介護1一時金 trigger against a 要介護3以上 annuity [S1]; waiver at 要介護1以上 against a 要介護3以上 lump sum [S8]; waiver on the first annuity payment, so simultaneous with the benefit [S4]; waiver on the 介護年金 trigger, carrying the riders’ premiums too [S7]; waiver on payment of the 介護保険金 [S12]; not stated at two carriers [S11] [S13] unverified. The composite takes 要介護1以上 — the lowest observed threshold and the one two carriers publish [S1] [S8] — because a waiver that fires strictly before the benefit is the case a model can get wrong, and a waiver simultaneous with the benefit is the trivial special case of it.

Benefit provisions#

Parameter

Representative value

Basis

Care lump sum (kaigo ichijikin, 介護一時金)

介護一時金額 paid once only per contract, on first satisfaction of the 要介護2以上 trigger; does not terminate the contract

four of seven carriers [S1] [S4] [S7] [S12]; threshold std (10)

Care annuity (kaigo nenkin, 介護年金)

基準介護年金額 paid annually in advance, the first instalment on the date the 要介護3以上 trigger was first met (介護年金支払基準日) and later instalments on its annual anniversaries, at most 10 payments

[S1] [S7] [S10]; threshold and cap std (11)

Annuity metering

Survival-tested — each instalment requires only that the insured be alive on the payment date; the state-tested variant, which additionally requires the care state to persist, is a switch

[S4] [S7] [S10] [S12] against [S1]; std (11)

Unpaid lump sum at annuity start

Where the annuity triggers before the lump sum has been paid, the unpaid lump sum is paid together with the first annuity instalment

[S1]

Company-basis limb — physical

Dependency in defined daily-living actions, physician-diagnosed (約款所定の要介護状態) that has continued 180 days or more; available only where the insured is 満65歳未満

[S1] [S2] [S4] [S12]; std (12)

Company-basis limb — dementia

Organic dementia (kishitsusei ninchishō, 器質性認知症) with disorientation (見当識障害) in the absence of clouded consciousness, continued 90 days or more; 満65歳未満 only

[S1] [S2]

Both limbs satisfied

The public-certification limb governs and its amount is paid

[S1]

Waiting period

None on the care benefits; a 180-day 認知症診断責任開始期 applies to the dementia rider only

[S1] against [S4] [S5] [S7] [S13]; std (13)

Pre-inception (責任開始期前) rule

Nothing is paid where the certification, or the company-basis state, results from an illness contracted or an accident occurring before the 責任開始期

[S1] [S2]

Exclusions (免責事由)

故意 or 重大な過失 of the policyholder or the insured; the insured’s 犯罪行為; 戦争その他の変乱; 薬物依存. The war exclusion is qualified — where the number of lives affected would not materially change the insurer’s liability the benefit may still be paid

[S2]

Other refusal grounds

Certification at 要支援1, 要支援2 or 非該当（自立）where the contractual threshold is a 要介護 band; failure to meet the 約款 definition of a company-basis state; lapse; rescission for non-disclosure or for 重大事由

[S1]

Suicide

Nothing is paid where the care state results from an intentional act of the insured; no separate 免責期間 is stated for this product

[S2]; statutory frame REG-R34

Contract termination

On the 10th annuity payment, effective retroactively to the date that payment’s trigger was met; otherwise on death, on lapse, or on rescission

[S1]; std (14)

  1. The observed threshold range is the widest single spread in this product, running from 要支援1 to 要介護3: 要支援1以上 as a main-contract trigger at one carrier [S11] and as a 特則 at a second [S5]; 要支援2以上 at a third [S9]; 要介護1以上 at three [S1] [S4] [S8]; 要介護2以上 at four [S1] [S4] [S7] [S12]; 要介護3以上 at three [S1] [S8] [S10]; and a graded payout across 要介護3/4/5 at two [S1] [S12]. The composite takes 要介護2以上 for the lump sum: it is the modal single threshold, it is what 公的介護保険連動型 usually means in market copy, and it is the threshold whose public base rate is best evidenced — 50.8% of all certified persons R4 REG-R30, a figure a carrier quotes back in its own brochure [S12]. It is also the threshold at which the 要支援2 / 要介護1 band overlap R2 stops mattering, which removes the single worst source of ambiguity in the certification scale.

  2. Annuity metering is the most model-relevant divergence in the product, and the two designs need different machinery. State-tested: instalments fall on the annual anniversaries of the 介護年金支払基準日 and each requires the insured to still be in the state on that date; at most one payment a year and at most 10 over the contract; if the state lapses and the insured later re-qualifies, a new 介護年金支払基準日 is set [S1] [S2]. Survival-tested: 5年確定年金 or 終身年金 elected at issue, tested only on 生存している限り, with the remaining instalments commutable at present value [S4]; 5年有期 / 10年有期 / 終身年金 with the carrier stating explicitly that recovery does not stop payment [S7]; 終身 or 5/10/15年有期 [S10]; 保証金額付 or 10年保証期間付 terminal annuities by conversion [S12]. The composite takes survival-tested, 10 annual payments: survival-testing is four carriers against one [S4] [S7] [S10] [S12], and ten payments is simultaneously the one carrier’s hard cap [S1] and one of another’s published terms [S7], so the cap is evidenced from both sides. The state-tested variant is retained as a switch because it needs a recovery decrement that the base run does not — see technical-notes.md. The composite’s 要介護3以上 annuity threshold is the one written by two of the four carriers selling a main-contract annuity [S1] [S10]; a third writes 要介護2以上 [S7] and the fourth lets the buyer elect 要介護1以上 or 要介護2以上 at issue [S4]. Putting it one grade above the lump sum makes the contract a ladder of thresholds, which three carriers write in some form: one across waiver, two lump-sum tiers and the annuity [S1], a second splitting a 要介護1以上 waiver from a 要介護3以上 lump sum [S8], and a third reserving its top annuity tier for 要介護4 / 要介護5 [S12].

  3. Every carrier writes the trigger as an either/or — the public certification, or a company-defined care state that has persisted for a stated number of days [S1] [S4] [S7] [S11] [S12] [S13] R15. Observed durations: 180日以上 [S1] [S4] [S12]; 180日を超えて, a strictly tighter form of the same test [S7]; 180日間継続, stated by that carrier to correspond to 要介護2以上 [S11]; 90日以上 for the dementia-defined state [S1] [S2]; 180日 continuation for a dementia treatment annuity [S13]. Age restriction: 満65歳未満 at three carriers [S1] [S4] [S12], no age restriction stated at two [S7] [S11]. The composite takes 180 days (90 for dementia) with the 満65歳未満 restriction, the majority position and the one the statute explains: below 65 the public limb can only fire on a 特定疾病 R1 R3, so the company limb is filling a hole that closes at 65. One carrier grants a relief the composite adopts: where the insured qualified under the company limb while under 65, the age condition is not re-applied to instalments 2 onwards once the insured passes 65, provided the state continues [S1].

  4. Two mechanisms both appear in marketing copy as “180日” and they are not the same thing. A 不担保期間 or waiting period means cover has not started yet; a duration test inside the trigger means the care state must have persisted before it counts. Observed waiting periods: none stated on the care benefits at one carrier, which relies on the 責任開始期前 rule instead [S1] [S2]; none on the care benefits but a 認知症診断責任開始期 of 責任開始日から180日を経過した日の翌日（181日目）on the dementia and MCI benefits at a second [S4] [S5]; a 1-year 不担保期間 on every benefit at a third, the explicit price of three-question simplified underwriting [S7]; 90 days from the 契約日 at a fourth, on dementia benefits [S13]; not stated at the remaining three. The composite takes no waiting period on the care benefits — the position of the only carrier whose 約款 was read [S2] — and the 180-day 認知症診断責任開始期 on the dementia rider [S4] [S5]. The 1-year 不担保期間 belongs to a simplified-underwriting product and is carried as a model-point flag, not as a composite default.

  5. One carrier extinguishes the contract on the 10th annuity payment, retroactively to the date that payment’s trigger was met [S1]; one pays once only and extinguishes [S11]; one extinguishes on payment of the 軽度介護保険金, the 死亡保険金 or the 高度障害保険金 [S9]. The composite terminates on the last annuity payment. This gives the model a benefit-driven termination decrement, as medical has on exhausting both aggregate limits — but here it is a counter of ten annual payments, not a counter of days.

Options — 特約 and 特則#

Parameter

Representative value

Basis

Dementia lump-sum rider (認知症一時金特約)

¥1,000,000 on the first diagnosis of 約款所定の器質性認知症 after the 認知症診断責任開始期, once per contract; 10% of that amount on the first diagnosis of MCI (keido ninchi shōgai, 軽度認知障害), once and extinguishing for that benefit line; where dementia is diagnosed first with no prior MCI claim, both are paid together

[S4] [S5] [S7] [S13]; std (15)

Dementia diagnostic standard

器質性認知症 requires an acquired organic brain lesion and a persistent, global loss of already-acquired intelligence caused by it; diagnosis must rest on cognitive testing and imaging, with a reasonable-grounds fallback where those cannot be performed. MCI is 日常生活動作は自立しているものの、認知機能が低下し、認知機能領域の障害が認められる状態

[S2] [S4] [S5]

Mild-care lump sum (軽度介護一時金給付特則)

A lump sum at 要支援1以上 or 要支援2以上; paying it extinguishes the 特則 and stops its premium but leaves the contract in force. Out of the base run

[S5] [S9] [S11]; scope std (16)

新保険料払込免除特約

Extends the waiver to a listed disease range, priced separately. Out of scope

[S5]

健康祝金特則

10% of the 介護年金額 every five years while no claim has been made. Out of scope

[S7]

介護保険金割増年金支払特約

Converts a lump sum into an enhanced annuity, with a 40-or-over commencement age and the top tier reserved for 要介護4 / 要介護5. Out of scope

[S12]

引受基準緩和型 / simplified underwriting

Out of scope — a different morbidity basis, not a rider

[S7]; scope std (16)

  1. Dementia cover is sold in four architectures: as a 主契約の型 elected instead of a care 型, paying 認知症診断一時金 [S4], with MCI on a separate 特則 [S5]; as a rider splitting one amount 10% MCI / 90% dementia, with the 認知症一時金特約 and 介護一時金特約 capped at ¥2,000,000 combined [S7]; as a stand-alone product with three layers — 軽度認知障害保険金 fixed at 10% of the 認知症診断保険金 (¥10,000–¥300,000), 認知症診断保険金 ¥100,000–¥3,000,000, and 認知症治療年金 ¥40,000–¥2,000,000 payable once the 所定の状態 has continued 180 days and annually thereafter on survival [S13]; and as a separate carrier product at issue ages 40–75 [S8]. One carrier sells no dementia benefit at all on this contract, letting dementia in only through the 90-day 認知症による要介護状態 limb of the same three care benefits [S1] [S2]. The composite takes the rider form with a ¥1,000,000 amount and the 10% MCI fraction, which is uniform at both carriers that publish one [S7] [S13] — one of the few genuinely market-wide parameters here. Household penetration of dementia cover is 7.6% against 20.1% for care cover R14, so it is a rider, not the chassis.

  2. Scope exclusions rather than value standardizations. 要支援-level benefits are a real and growing segment — one carrier’s whole main contract triggers at 要支援1以上 [S11] — but they roughly double the triggering population against the composite’s lump-sum threshold (要支援1以上 is 100% of certified persons against 50.8% at 要介護2以上 R4) and treble it against the annuity threshold (34.0% at 要介護3以上 R4), and they change the product from a severe-disability cover into a high-frequency one. They are carried as a threshold parameter, not as a base-run benefit. Simplified underwriting is a different risk pool priced with a 1-year 不担保期間 [S7]; the composite is fully underwritten.

Termination and values#

Parameter

Representative value

Basis

解約返戻金

None at any duration — 「この保険契約の解約払戻金はありません」

[S1] [S2] [S7] [S8]; std (17)

Policyholder dividend (配当金)

None — the chassis is 無配当; none of the four statutory surplus-distribution methods applies

[S1] [S2] [S4]; REG-R9

契約者貸付 / 自動振替貸付

Neither is offered: there is no surrender value to lend against, so a missed premium lapses the contract

[S2]; REG-R14; std (17)

Grace (払込猶予期間), 月払

From the first day of the month following the 払込期月 to the last day of that month

[S2]

払込猶予期間, 半年払・年払

From the first day of the month following the 払込期月 to the 月単位の契約応当日 in the month after that

[S2]

Claims during grace

Unpaid premiums are deducted from the benefit; where the benefit is smaller than the arrears the balance must be paid by the end of grace, failing which the contract lapses and no benefit is paid

[S2]

Lapse (shikkō, 失効)

From the day after the grace period expires

[S2]

Reinstatement (fukkatsu, 復活)

Within 1 year of the lapse date, on fresh 告知, payment of arrears and the insurer’s consent; the 復活日 is the later of the arrears payment date and the 告知日, and the 責任開始期 resets to it

[S1] [S2]; std (18)

Non-disclosure (告知義務違反)

Rescission where a material misstatement is found within 2 years of the 責任開始日; the two-year bar does not protect the policyholder where the claim event fell inside those two years; rescission for fraud is not time-limited

[S1]; ceiling REG-R35

クーリング・オフ

8 days from the later of the 申込日 and the 告知日, full refund; out of scope

[S1]; REG-R36; scope std (18)

Death of the insured

Contract terminates, nothing payable

[S1] [S11]; std (3)

  1. Four of the seven carriers publish an outright nil surrender value [S1] [S2] [S7] [S8], and one of them carries the fact in the formal product name itself — 介護保険〔無解約払戻金 2021〕 [S1]. The two designs that do produce a value are excluded: 5% of the elected benefit amount, available only after a short premium-paying period completes and only before the main benefit has triggered, with a 死亡時返戻金 equal to it [S4]; and the full 低解約返戻金型 whole-life progression, suppressed to 70% of the ordinary value during the premium-paying period and stepping to 100% at 払込満了 [S12], plus a 低解約返戻金型 form of the 要支援 product [S9]. The composite’s nil value is what removes 自動振替貸付 from this product, and with it the whole soft-landing mechanic that whole_life has and this chassis does not REG-R14. The lapse is real and immediate.

  2. Reinstatement within one year is the only window found in a retrieved 約款 [S1] [S2], and it is a caution rather than a market statement: the three-year window commonly quoted for the Japanese market as a whole is unverified and this carrier’s own window is one year. No other carrier’s grace, lapse or reinstatement terms were retrievable, so the composite rests on one contract and says so. クーリング・オフ is the eight-day dispatch-rule right of 保険業法 第309条 REG-R36; jplib models from the point cover is in force and treats the window as a pre-inception matter it has no new-business funnel to represent.

Contractual mechanics#

The trigger, which is the product#

Write g for a certification grade on the ordered scale 要支援1 < 要支援2 ≈ 要介護1 < 要介護2 < 要介護3 < 要介護4 < 要介護5, and G_L, G_N, G_W for the contractual thresholds of the lump sum, the annuity and the waiver — 要介護2, 要介護3 and 要介護1 in the composite. The trigger is a disjunction, evaluated separately for each of the three benefits:

trigger(G, t) = 1  if the insured holds a public certification at grade G or above,
                   effective at or before t
              = 1  if the 約款-defined company-basis care state has persisted 180 days
                   (90 days where dementia-defined) and the insured was under 65 at
                   the date that state was diagnosed
              = 0  otherwise

Where both limbs are satisfied the public limb governs and its amount is paid [S1]. The public limb is an administrative act, not a clinical one, and three of its properties belong in a model rather than in a footnote R1:

  • The decision is due within 30 days of application and takes effect retroactively to the application date. The claim date is therefore the application date, not the decision date, and a projection that dates claims at notification is systematically late by up to a month.

  • The certification has a finite 有効期間 and must be renewed (要介護更新認定). The grade can move up or down at renewal.

  • Below 65 the public limb fires only on one of the 16 特定疾病 R1 R3, which is what the company-basis limb exists to cover, and why three carriers restrict that limb to lives under 65 [S1] [S4] [S12].

The composite treats the care state as absorbing once entered — a std simplification forced by the sources, since no public transition-rate table exists in any register and no carrier publishes one. It is a real simplification and not a harmless one: the statute provides for downgrade at renewal R1, a carrier’s own FAQ addresses what happens 「要介護状態が改善された場合」 [S3], and the state-tested annuity variant [S1] cannot be modeled at all without a recovery decrement. technical-notes.md carries the recovery rate as a named std input set to zero in the base run, not as an omission — and states what that rate does and does not cover, since it models the fall below the annuity grade rather than a return to health.

介護一時金#

lump_sum(t) = A_L * 1{ first t at which trigger(G_L, t) = 1 }

paid once per contract and not terminating it [S1] [S4]. A_L is the 介護一時金額, ¥3,000,000 in the composite. Where a carrier writes two lump-sum tiers — a 要介護1一時金 and a 要介護2一時金, each once only [S1] [S2] — the composite collapses them into the single 要介護2 tier and keeps the 要介護1 threshold for the waiver alone, so that the model has one lump sum and one waiver rather than two of each.

介護年金#

ann_start   = first t at which trigger(G_N, t) = 1        (the 介護年金支払基準日)
pay_date(k) = ann_start + (k - 1) years,     k = 1 .. 10
annuity(k)  = A_N * 1{ insured alive at pay_date(k) }              (survival-tested)
annuity(k)  = A_N * 1{ alive and trigger(G_N, pay_date(k)) = 1 }   (state-tested switch)

with A_N = ¥600,000. Instalments are annual and in advance, at most one a year, at most ten in total [S1] [S7]. The contract terminates on the tenth payment, effective retroactively to the date that payment’s trigger was met [S1]. If the annuity triggers before the lump sum has been paid, the unpaid lump sum is paid together with the first annuity instalment [S1] — which is not merely a timing nicety, because it means a life entering directly at 要介護3以上 receives A_L + A_N on one date.

Under the state-tested switch, a lapse of the state does not merely suspend payment: if the insured later re-qualifies, a new 介護年金支払基準日 is set and the anniversary schedule restarts from it [S1]. A model that keeps the original schedule and skips instalments is wrong for that design.

The two metering bases need different assumption sets, and the difference is first-order. A survival-tested annuity is a life annuity conditional on one-time entry, and after entry it needs only an impaired-life mortality basis. A state-tested annuity needs a recovery (or de-certification) decrement as well, on top of the ten-payment cap. Where the annuity is survival-tested, one carrier says in terms that recovery does not stop it — 「介護年金のお支払事由に該当した後、容体が改善した場合」も年金支払期間中、生存されている限りお受け取りいただけます — and that the premium waiver likewise stays in force [S7].

保険料払込免除#

premium(t) = 0  for every t after the first date at which trigger(G_W, t) = 1

with G_W = 要介護1, plus the two disability limbs — 高度障害状態 from any cause, and a listed 身体障害状態 arising from an 不慮の事故 within 180 days of it [S1] [S2]. Waived premiums are treated as though paid on each 払込期月の契約応当日, so the contract continues in full force; while the waiver is running the payment mode cannot be changed and the benefit amount cannot be reduced [S1] [S2].

Because G_W sits below G_L, the waiver is not a by-product of the claim: there is a band of the population — 要介護1 lives, 20.7% of all certified persons R4 — for whom the contract has stopped collecting premium and has not yet paid anything. That interval is the single most easily mis-modeled item in this product.

Waiting periods against duration tests#

The composite has no waiting period on the care benefits; protection runs instead through the 責任開始期前 rule, which refuses any claim where the certification or the company-basis state results from an illness contracted, or an accident occurring, before the 責任開始期 [S1] [S2]. The dementia rider carries a 認知症診断責任開始期 of 責任開始日から180日を経過した日の翌日（181日目）[S4] [S5].

These are different mechanisms from the 180-day (or 90-day) persistence test inside the company-basis trigger, which is a property of the care state, not of the contract’s cover period. Both are written “180日” in marketing copy. A model that implements one of them twice, or the wrong one, will be wrong at inception and wrong again in the tail.

Underwriting, exclusions, grace, lapse and reinstatement#

Underwriting is 告知扱い and the front door is tight: one carrier declines anyone currently hospitalised or advised to be hospitalised or operated on, or who has ever been certified for, or has ever applied for, 要支援 or 要介護 [S1]; another declines applicants already applying for or actively considering certification, or living in a 高齢者向け施設 [S11]; the simplified-underwriting carrier asks three questions, one of them whether the applicant has ever applied for certification, and prices that leniency with a 1-year 不担保期間 [S7]. The anti-selection this guards against is unusually direct, because the thing being selected against is an application to a public body that leaves a record.

免責事由 are 故意 or 重大な過失 of the policyholder or the insured, the insured’s 犯罪行為, 戦争その他の変乱 and 薬物依存, with the war exclusion qualified where the number of lives affected would not materially change the insurer’s liability [S2]. Note what is not on the list: no period-limited suicide exclusion is stated for this product, which is consistent with 保険法 第51条 imposing no statutory time limit and leaving any 免責期間 to the contract REG-R34.

For a 月払 contract, grace runs from the first day of the month following the 払込期月 to the last day of that month; non-payment produces 失効 from the day after grace expires [S2]. A claim arising in the grace window is paid net of the arrears, and where the benefit is smaller than the arrears the balance must be paid by the end of grace, failing which the contract lapses and no benefit is paid at all [S2]. There is nothing to break the fall: with no surrender value there is no 自動振替貸付 [S2] REG-R14. 復活 is available within one year on fresh 告知 and payment of arrears, and it is not a rewind — cover restarts from the 復活日 and the 責任開始期 resets to it, which matters precisely because the whole benefit definition is anchored to 責任開始期以後の傷害または疾病 [S1] [S2]. 告知義務違反 supports rescission for two years from the 責任開始日, inside the five-year statutory ceiling REG-R35.

Riders and options#

In scope (modeled or parameterized):

  • 介護一時金 — the main-contract lump sum at 要介護2以上, once only [S1] [S4] [S7] [S12].

  • 介護年金 — the main-contract annuity from 要介護3以上, ten payments, survival-tested [S1] [S7] [S10]. Switchable to state-tested, which switches on the recovery decrement.

  • 保険料払込免除 — from 要介護1以上, plus the 高度障害 and accident-disability limbs [S1] [S2] [S8]. Always on; the threshold is a parameter.

  • Company-basis trigger limb — 180-day persistence, 90 days where dementia-defined, lives under 65, with the post-65 continuation relief [S1] [S2] [S4] [S12]. Switchable off, which reduces the product to pure 公的介護保険連動型.

  • 認知症一時金特約 — ¥1,000,000 on first 器質性認知症 diagnosis with 10% of it on first MCI diagnosis, behind a 180-day 認知症診断責任開始期 [S4] [S5] [S7] [S13]. Off in the base run.

  • Trigger-threshold parameter — G_L, G_N and G_W are model-point fields spanning 要支援1 to 要介護3, so the whole observed market spread is reachable without a second chassis [S1] [S4] [S5] [S7] [S8] [S9] [S10] [S11] [S12].

  • 1-year 不担保期間 — a model-point flag for the simplified-underwriting design [S7]. Off in the base run.

Out of scope: the accelerated-benefit whole-life chassis and its 低解約返戻金型 surrender progression, 自動振替貸付, 延長定期保険, 払済保険 and 減額 [S12] — that machinery belongs to the whole life product specification (終身保険) and its model WholeLife_JP_A; 介護保険金割増年金支払特約 and the 介護年金移行特約, which convert a lump sum into an enhanced annuity with 要介護4 / 要介護5 top tiers [S12]; 軽度介護一時金給付特則 and 軽度認知障害診断一時金給付特則 as separate extinguishing benefit lines [S5]; 新保険料払込免除特約 [S5]; 健康祝金特則 [S7]; the 5%-of-benefit post-払込満了 surrender value and the matching 死亡時返戻金 [S4]; the 死亡給付金 equal to one annuity instalment [S10]; the 認知症治療年金 as a third dementia layer [S13]; commutation of remaining annuity instalments at present value [S4]; 引受基準緩和型 chassis [S7]; and クーリング・オフ [S1] REG-R36.

Variations across insurers#

  1. Public-certification threshold. 要支援1以上 [S11] [S5]; 要支援2以上 [S9]; 要介護1以上 [S1] [S4] [S8]; 要介護2以上 [S1] [S4] [S7] [S12]; 要介護3以上 [S1] [S8] [S10]; graded across 要介護3/4/5 [S1] [S12]. Composite: 要介護2以上 for the lump sum, 要介護3以上 for the annuity, 要介護1以上 for the waiver — the modal tiering, and the one with the best public base rates (footnote 10).

  2. Company-basis limb. 180日以上 at three carriers [S1] [S4] [S12]; 180日を超えて at a fourth, strictly tighter [S7]; 180日間継続 at a fifth [S11]; 90日以上 for the dementia-defined state [S1] [S2]; not stated on the pages fetched for a sixth [S8] unverified. Age restriction 満65歳未満 at three [S1] [S4] [S12], none stated at two [S7] [S11]. Composite: 180 days, 90 for dementia, 満65歳未満 (footnote 12).

  3. Benefit form. Lump sum only at three [S8] [S9] [S11]; annuity only at two [S7] [S10]; both, tiered by grade, at one [S1]; one or the other elected at issue from five mutually exclusive 保険契約の型 that cannot be changed mid-term at one [S4] [S6]; lump sum convertible to an annuity at one [S12]. Composite: both, tiered — the design that exercises every mechanic the others use in isolation.

  4. Annuity metering. State-tested annually with a hard 10-payment cap and contract extinction at one carrier [S1]; survival-tested at four [S4] [S7] [S10] [S12], with payout shapes 5年確定 [S4], 5/10年有期 or 終身 [S7], 終身 or 5/10/15年有期 [S10], and 保証金額付 or 10年保証期間付終身 [S12]. Composite: survival-tested, ten payments (footnote 11). This is the widest genuinely structural divergence in the product — the two designs do not differ by a parameter, they differ by whether a recovery decrement exists.

  5. Amount graded by certification band. Two carriers grade the payout: 1型 pays 基準介護年金額 at 要介護5, ×5/6 at 要介護4 and ×4/6 at 要介護3, while 2型 pays ×1 / ×2/3 / ×1/3, with the company-basis limb paid at the 要介護3 fraction [S1]; a second reserves its top enhanced annuity tier for 要介護4 or 要介護5 [S12]. Five carriers pay a flat amount above the threshold [S4] [S7] [S8] [S10] [S11]. Composite: flat — the majority, and the graded form is reachable by running the model at each fraction.

  6. Premium waiver threshold. 要介護1以上 at two [S1] [S8]; simultaneous with the first annuity payment at one [S4]; on the 介護年金 trigger, carrying rider premiums too, at one [S7]; on payment of the 介護保険金 at one [S12]; not stated at two [S11] [S13] unverified. Composite: 要介護1以上, strictly below the benefit threshold (footnote 9).

  7. Waiting period. None stated on the care benefits at one [S1]; none on care benefits but 180 days on dementia and MCI at a second [S4] [S5]; one year on every benefit at a third, the price of three-question underwriting [S7]; 90 days from the 契約日 on dementia at a fourth [S13]; not stated at three. Composite: none on care, 180 days on the dementia rider (footnote 13).

  8. Dementia and MCI. A main-contract 型 with MCI on a 特則 [S4] [S5]; a rider splitting one amount 10% MCI / 90% dementia [S7]; a three-layer stand-alone product [S13]; a separate product at issue ages 40–75 [S8]; and no dementia benefit at all, dementia entering only through the 90-day company-basis limb [S1] [S2]. Composite: a rider at 10% / 90% — the uniform fraction wherever one is published [S7] [S13] (footnote 15).

  9. Surrender value. None at four [S1] [S2] [S7] [S8]; 5% of the elected benefit after a completed short premium term, and only before the benefit triggers, at one [S4]; a full 低解約返戻金型 progression at 70% during the premium-paying period at one [S12], with a 低解約返戻金型 form of the 要支援 product at another [S9]; not stated at two [S11] [S13] unverified. Composite: none (footnote 17).

  10. Issue ages and amount limits. 18–79 [S1]; 15–69 on a direct channel [S6]; 40–79, and 40–75 on a 定期 variant [S8] [S9] [S10]; a rate table spanning 30–80 with the envelope unstated [S7] unverified; a ¥500,000 cap where the insured is 65 or over at issue [S11]; a rate table published only at issue ages 20/30/40/50 [S12]. Composite: 40–79 (footnote 2).

  11. Chassis. Stand-alone third-sector cover at five carriers [S1] [S4] [S7] [S8] [S11]; an accelerated 介護保険金 of 50% of the sum insured on a whole-life contract at one, after which the sum insured drops by the same amount, premiums are waived and death cover continues for life [S12]; a dementia-only product at one [S13]. Composite: stand-alone (footnote 1).

  12. What does not vary. Every product ties its primary trigger to the public 要介護 or 要支援 certification and names 介護保険法 as the scheme it means, one carrier citing the certification ordinance by name inside its own contract wording [S1] [S12] R2; every one carries a company-basis alternative expressed as a defined dependency state that must have persisted for a stated number of days [S1] [S4] [S7] [S11] [S12] R15; every stand-alone product is written 終身 with 終身払 premiums [S1] [S4] [S7] [S8] [S10] [S11]; every one has a premium-waiver mechanic tied to the care state [S1] [S4] [S7] [S8] [S12]; every 一時金 benefit is once-only per contract [S1] [S4] [S5] [S9] [S11]; premiums are rated on sex and 満年齢 alone, with female rates above male at every age on the lump-sum shapes at the two carriers publishing a rate at this severity [S6] [S8] — though the 要支援-level product runs the other way [S9], which is what the invariant is worth; and no carrier publishes an incidence basis, a 予定発生率 or a 予定利率, which the regulator confirms is the expected state of affairs for 第三分野 business R10. These are the invariant core of the composite.

Regulatory context#

Classification. Private 介護保険 is 第三分野 business under 保険業法 第3条, writable under either a life or a non-life licence REG-R1, and the 金融庁 groups it with 医療保険 and がん保険 in the field paying on disease, injury and treatment R10. The classification, not the product’s shape, is what pulls in everything below. The public 公的介護保険 scheme it links to is not insurance business at all: it is social insurance under 介護保険法, administered by municipalities R1 REG-R42.

Prudential — the third-sector reserving overlay. 保険業法 第116条 requires a policy reserve (sekinin-junbikin, 責任準備金) at each 決算期 and delegates the accumulation method for long-term contracts REG-R4; 施行規則 第68条 sets the scope of the resulting 標準責任準備金 REG-R7; 平成8年大蔵省告示第48号 fixes the method as net level premium (平準純保険料式) with the table vintages and the standard valuation interest rate (hyōjun riritsu, 標準利率) reset machinery REG-R10; and for contracts concluded on or after 1 April 2018 the third-sector valuation mortality is 第三分野標準生命表2018 REG-R11. The 標準利率 applicable to this product class could not be established from a retrieved document and is unverified.

On top of that sits the overlay that is specific to this class and decisive for this product. 施行規則 第69条 divides the reserve into 保険料積立金, 未経過保険料, 払戻積立金 and 危険準備金, and requires a separately identified 第三分野保険の保険リスクに備える危険準備金 REG-R8. The 金融庁 framework paper states plainly that for 第三分野 there is no standard incidence indicator — 「標準死亡率、参考純率といった スタンダードな指標が存在しておらず、公的なデータや各社の実績等から給付事由ごとその発生率を見込まざるを得ない」 — and imposes, on top of 標準責任準備金, an annual ストレステスト checking that the 予定事故発生率 covers the 99th percentile of incidence risk over a ten-year horizon, with 危険準備金 topped up where it does not; a 負債十分性テスト by future cash flow analysis where the 予定発生率 fails to cover risk defined at 97.7%; disclosure of the incidence model used; and a transparent numeric 基礎率変更権 exercise standard disclosed at point of sale R10. The stress test itself must be computed under 平成10年6月8日大蔵省告示第231号, with organisational separation between the calculating unit and internal audit REG-R14 REG-R13; the notification’s own text was not retrieved, so its stress magnitudes are unverified and no numeric stress level is asserted anywhere in these documents.

This is the exact boundary at which the reference model marks its incidence basis std. The 日本アクチュアリー会, in its 指定法人 role REG-R23, publishes a mortality table for this class and no morbidity table: 第三分野標準生命表2018 contains qx and nothing else R8. Its own construction summary records that it is graduated from 第21回生命表（2010年）, the national life table rather than insured experience — a deliberate change from the 2007 edition — and that it excludes 高度障害, which the 2007 edition included R9 REG-R20. It is also deliberately lighter than the death-insurance table at every age: male q65 is 0.00845 against 0.01015, and q75 0.02242 against 0.02637 R8. That is the correct direction of conservatism for a living-benefit product, where longer survival means more annuity instalments — and it means a best-estimate nursing-care model must not adopt the valuation table as its mortality decrement without an explicit std adjustment, and must say which of the two it is using at each point. The tables are published free and in full at stable public URLs REG-R18, but the 日本アクチュアリー会 site terms restrict reproduction and transmission REG-R21, so this library cites them and ships a documented proxy, never a copy.

What fills the morbidity hole is public administrative data rather than an actuarial table: the 認定率 and grade composition of 介護保険事業状況報告 R4 R5 REG-R30, with the five-year-band 認定率 that a finer basis wants available in the e-Stat release of the same statistic, which was not fetched REG-R33. Post-onset mortality is a harder gap: no impaired life table for the care state exists in any retrieved source, so the population anchor is the 完全生命表 — 平均余命 19.97 years for a male at 65 and 24.88 for a female, 12.54 and 16.22 at 75 R13 — and since more than 90% of certified 第一号被保険者 are 75 or over R4, the 75 figures are the relevant ones. Both the post-onset mortality basis and the recovery rate are std.

Prudential — solvency. From 31 March 2026 insurers are supervised on the economic-value ESR (経済価値ベースのソルベンシー規制), in which liabilities are 現在推計 plus MOCE, re-measured at each 基準日 on assumptions re-set then, with required capital calibrated at 99.5% and early corrective action triggered below an ESR of 100% REG-R15. It supersedes the ソルベンシー・マージン比率 trigger at 200% REG-R17, and the two are not comparable. jplib computes neither. What it owes the regime is that its projections are re-runnable on a re-set assumption basis at a stated 基準日 — for a third-sector product, precisely the capability the ストレステスト already demands R10. J-GAAP reserving, ESR and IFRS 17 (voluntary in Japan) are three separate measurement bases fed by one projection REG-R47.

Conduct. 監督指針 fixes what the 契約締結前交付書面 must contain — 商品の仕組み, the main 支払事由 and 免責事由, the attachable 特約, 保険期間, 引受条件, 保険料, 配当金に関する事項 and 解約返戻金等の水準 in the 契約概要; クーリング・オフ, 告知義務, 責任開始期, the main non-payment cases and 保険料の払込猶予期間・失効・復活 in the 注意喚起情報 REG-R14 — which is, item for item, the source document this specification is largely built from [S1]. The point-of-sale disclosure of a numeric 基礎率変更権 standard R10 rides on the 説明義務 of 金融サービス提供法 第4条 REG-R39; for a product whose benefit trigger is an administrative grade the customer does not control, that duty is doing real work. クーリング・オフ is the eight-day dispatch-rule right of 保険業法 第309条, scoped out here REG-R36. Non-disclosure remedies run under 保険法 第55条, whose ceiling is five years from inception with a one-month clock from discovery REG-R35; the two-year contractual window observed [S1] is a permitted narrowing in the policyholder’s favour. The suicide exclusion of 保険法 第51条 carries no statutory time limit, so the absence of a stated 免責期間 in this product’s 免責事由 list [S2] is a contractual fact, not an oversight REG-R34. On insurer failure, contracts are compensated up to 90% of the 責任準備金 at the failure date through the 生命保険契約者保護機構 REG-R40, the rate set by ordinance under 保険業法 第270条の3 REG-R41.

Tax. Premiums for this product fall in the 介護医療保険料 basket of the post-2012 生命保険料控除 regime — one of three baskets (一般 / 介護医療 / 個人年金), each computed on the same schedule (up to ¥20,000 in full; ¥20,000–40,000 half plus ¥10,000; ¥40,000–80,000 a quarter plus ¥20,000; above ¥80,000 a flat ¥40,000), with an overall cap of ¥120,000 R11 REG-R43. The basket covers contracts written on or after 2012-01-01 paying on 疾病または身体の傷害等 where payment is triggered by a 医療費支払事由, excludes terms under five years and savings-type contracts, and requires every beneficiary to be the premium payer, a spouse or a relative R12. A contract mixing coverages is assigned to the basket of its 主たる保障内容 and rider premiums are allocated by the rider’s own coverage R11 — which is why the 主契約 + 特約 architecture of Japanese products has a tax consequence, and why a dementia rider on a care contract does not change the basket. The anchor cell’s ¥11,500 monthly premium is ¥138,000 a year, comfortably past the ¥80,000 point at which the basket deduction flattens to ¥40,000, so at the composite’s benefit level the marginal premium yen carries no relief at all.

Professional standards and 三利源. The 保険計理人 must give an 意見書 on, among other things, the soundness of the reserve REG-R6, supported by the 1号収支分析 of the profession’s 実務基準 — a projection of at least ten future years, per segment REG-R22. The ten-year horizon is the same horizon the third-sector ストレステスト uses R10, which is convenient rather than coincidental. Because this chassis is 無配当, none of the four surplus-distribution methods of 施行規則 第30条の2 applies REG-R9, and the 三利源 (死差 / 利差 / 費差) framing that drives dividends on 有配当 products has no dividend to drive here — it survives only as an analysis of where the profit came from.