Sources#

Source ids [S#] and [R#] are carried verbatim from _research/long-term-care.md, the citation ground truth for this product, and are frozen — never renumber. Numbering is per product: S1 in this file is a different document from S1 in every other product’s sources.md, and the cross-product references/regulatory-and-actuarial-references.md runs its own frozen R1–R62 scheme, cited here as [REG-R#]. The two R schemes must never be read across. The sharpest instance in this product is R4, the 2024 노인장기요양보험 통계연보, against REG-R4, which is 보험업법 제176조 (보험요율 산출기관); and R12, 보험업법 제4조, against REG-R12, which is 감독규정 제6-11조의7 (계약자배당). Access date for every entry: 2026-09-03.

Three tag families do three different jobs in product-spec.md, technical-notes.md and model.md:

  • [S#] — a primary product document: a 상품요약서 (the statutory summary of the 기초서류), a 보험안내자료, a 약관 (yakkan, policy conditions) or a carrier release. These are what make a contractual mechanic sourced rather than assumed. On this product they carry the whole contractual specification and, unusually, two quantitative anchors as well: [S1]’s 예정이율 and its 예정위험률 grid, and [S2]’s premium rate card and 환급률 progression.

  • [R#] — a regulatory, statutory, statistical or market reference that only this product needs. The load-bearing one is R4, the 2024 노인장기요양보험 통계연보, from which the entire morbidity basis is built; R7 and R11 carry the dementia rider and the post-onset survival evidence. Four of the seventeen — R15 to R18 — are news or trade press, one more (R13) is a vendor’s restatement of a decree, and every entry says so.

  • [REG-R#] — the cross-product reference library, reprinted in short form at the end so that a reader of this page alone can resolve every tag on it.

Every tag used in the three documents has an entry here, and every entry here is used. No source was added at drafting. One research entry is omitted as uncited: R5, the 국민건강보험공단 경영공시 「장기요양 등급 판정 현황」, which was retrieved only in part and whose three recovered figures agree exactly with R4. The same disclosure is a cross-product entry, REG-R42, and the documents cite that instead, which is why the gap exists rather than a duplicate.

Where the quantitative basis comes from, in one paragraph. Nothing in this product’s decrement basis is a published insurance table, because none exists: 경험생명표 is not released in full REG-R33 REG-R34, and 보험개발원 publishes neither a 장기요양 incidence table nor a post-onset mortality table — the 참조순보험요율 display that gives Cancer_KR_S its incidence grid REG-R61 carries nothing for this trigger. So the mortality table is a std construction on a published 기대여명, the morbidity basis is converted in the open from the public prevalence of R4, the post-onset mortality multiple rests on R11 and a duration bracket derived from R4 and R18, and the one Korean long-term-care incidence rate anyone publishes — [S1]’s 예정위험률 — is used for its gradient and as a cross-check, never as a level. model.md lists every standardization with its rationale and whatever observed range the documents bound.

Company and branded product names appear in this file and in _research/long-term-care.md and nowhere else in the library: in product-spec.md, technical-notes.md, model.md and the model docstrings a carrier is referred to by its tag alone, so a reader can always resolve who said what — here — and never has to.

Primary product sources#

Five documents from five carriers, on both sides of the market: two life-side documents [S1] [S2], two non-life 약관 [S3] [S4] and one life-side press release [S5]. That spread is a product fact rather than a sampling choice — 간병보험 is a 제3보험 종목 that both licences may write R12 REG-R1, and the non-life shelf is the larger one. Note where the documents actually came from: [S3] from the carrier’s own publication server and [S4] from a bank’s insurance-document mirror, not from the 손해보험협회 portal; no document in this set was reached through an industry association.

Retrieval method, common to the set. Plain curl is blocked for every Korean host tried, so everything was fetched with a summarising fetcher, which renders HTML but returns Korean PDFs as undecodable binary; in every such case the saved binary was extracted locally with pypdf and read directly. That is how the four PDFs below were obtained.

S1 — 우정사업본부(우체국보험), 「무배당 우체국간병비보험 2309 상품요약서」 (statutory product summary)#

  • Publisher: 과학기술정보통신부 우정사업본부 (Korea Post Insurance), sold through 체신관서; product code family P400059, P400060, 36 pp.

  • Doc type: 상품요약서 — a 기초서류 extract, not marketing copy

  • URL: https://www.epostlife.go.kr/resources/js/biz/ip/gs/pdf/yoyak_P400059,P400060.pdf

  • Accessed: 2026-09-03, Retrieved: yes (323.7 kB PDF, 36 pp., text extracted with pypdf and read in full)

  • The richest single document in the set, and the only one disclosing a pricing basis. What rests on it: the 예정이율 of 2.0% 연단위 복리, which is prem_int_rate and the one retrieved pricing interest rate anywhere in this product; the 예정위험률 table — 요양(1등급) and 요양(2등급) 발생률 at 만나이 40, 50 and 60 by sex — which is incidence_table.csv, the sub-65 gradient sub65_factor_at, the sex ratio and the calibration target of disclosed_inc_ratio_at; the 간병자금 annuity shape — monthly instalments over ten years, first twelve months guaranteed, 120-month cap, the first instalment in the month of certification, the amount fixed by the grade at first certification and never re-rated, and the annual 주민등록등본 proof of life that makes ann_tests an annual count; the freeze of the 감액 decision at first certification, quoted verbatim in the notes; the 180-day 보장개시일 with its 재해 carve-back and a two-year 50% 감액 — the combination model point 9 carries; the bar on surrender once the annuity has started, which is why pols_care carries no lapse; the 가입나이 / 보험기간 / 납입기간 grids; the 표준형 surrender-value form; the 12% 세액공제 statement; and the 간병인사용일당 rate that model.md names as a frequency, not a probability.

S2 — ABL생명, 「(무)ABL우리가족THE케어간병보험(해약환급금 미지급형)2504」 상품안내장#

  • Publisher: ABL생명보험주식회사; 16 pp., 제작 2025-09-01, 준법감시인 심의필 제2025-PA276호

  • Doc type: 보험안내자료 — the regulated pre-sale disclosure document under 보험업법 제95조

  • URL: https://abllife.co.kr/cms/prdt/hlthInjry/__icsFiles/afieldfile/2025/08/28/ + (무)ABL우리가족THE케어간병보험(해약환급금_미지급형)2504_20250901.pdf (served percent-encoded)

  • Accessed: 2026-09-03, Retrieved: yes (1.76 MB PDF, 16 pp., extracted with pypdf and read in full)

  • The most modern product in the set and the source of both published numeric series the model consumes. What rests on it: the 월납 premium rate card at ages 40 / 50 / 60 by sex for 25 covers, from which the anchor cell’s ₩5,600 and the female ₩8,400 are built, and from which the market’s own ≈ 4.5 : 1 threshold ladder and its opposite sex pricing on 장기요양 and 치매 are derived; the 해약환급금 미지급형 환급률 progression — 0.0% to year 15, 48.7% at 20, 54.4% at 30, 50.5% at 40, 0.0% at 50 — which is av_table.csv and therefore net_prem_ratio(); the statement that the 기본형 comparator 「가입이 불가능하며 … 해지율을 적용하지 않고 계산합니다」, which is the named weakness in that reconstruction; the 90-day 장기요양상태 보장개시일 and the one-year 치매 보장개시일 with the 90-day persistence test inside each CDR definition, together dementia_wait_mths = 15; the utilisation-conditioned 재가/시설/주야간보호/복지용구 지원금 riders that the absorbing-state simplification would not survive; and the 간편심사 question set behind uw_loading.

S3 — 삼성화재해상보험, 「무배당 삼성화재 간병보험 1808.2 (새시대 간병파트너)」 보험약관#

  • Publisher: 삼성화재해상보험주식회사, 장기상품개발2파트, printed 2018-08; 234 pp.

  • Doc type: policy conditions (약관) — a full primary contract document, with the 가입자 유의사항, the 주요내용 요약서, the 보통약관 and the 특별약관 집

  • URL: https://www.samsungfire.com/publication/pdf/ZPB214010_0_20180802_file1.pdf

  • Accessed: 2026-09-03, Retrieved: yes (8.4 MB PDF, 234 pp.; body text extracts cleanly, cover art and some marketing pages come out as mojibake and were skipped)

  • A non-life carrier’s grade-triggered design and the only document with a genuine multi-tier trigger menu — 기본계약 장기요양지원금(1~2등급) plus selectable 1등급 / 1~3등급 / 1~4등급 tiers and a 5년간 매월 생활자금. What rests on it: the 납입면제 firing on the same event as the benefit and waiving every attached rider, which is why premiums rides on pols_act and the waiver is not an independent decrement; 만기환급금 : 없음, which is why claims_maturity is a column of zeros; the absence of any general death benefit; the statutory-amendment rewriting clause — the contract-continuity answer to a trigger the insurer does not control; and the 두 번째 장기요양지원금 rider with its five-year 면책, the market’s own acknowledgement that grades move.

S4 — 한화손해보험, 「무배당 한화 골드클래스 간병보험」 약관 (policy conditions)#

  • Publisher: 한화손해보험주식회사; document revision date 2023-07-11 (PDF metadata); 173 pp., four plans including 간병보장플랜(납입중50%해약환급금지급형)

  • Doc type: policy conditions (약관)

  • URL: https://image.kebhana.com/cont/download/insdocument/provide/N02C14145_agree.pdf — a bank’s (하나은행) insurance-document mirror, not the carrier’s own site

  • Accessed: 2026-09-03, Retrieved: yes (6.6 MB PDF, 173 pp.; the 유의사항, the 감액/면책 tables and the 장기요양진단비 and 치매진단비 articles read verbatim)

  • What rests on it: the verbatim 지급사유 articles at each threshold from 1등급 to 1~5등급 with the one-year 50% 감액 table set out inside the article, which fixes red_fraction and the fact that the test is on the cause — 질병 halved, 상해/재해 in full — behind disease_share; the CDR definitions of 경증이상 / 중등도이상 / 중증치매 each with a 90-day persistence test, and the one-year full 면책 on the dementia riders against no 면책 on the 장기요양 riders; the 「납입중50%해약환급금지급형」 mechanics, which are cv_form = half_during; and the exclusion referring to 노인장기요양보험법 제29조.

S5 — 교보생명, 「교보더안심치매·간병보험」 출시 보도자료 (carrier press release)#

  • Publisher: 교보생명보험주식회사 (own newsroom); reported launch 2026-02-25

  • Doc type: carrier press release — not a contract document; every fact from it is a carrier’s summary of its own product and is weaker evidence than [S1] – [S4]

  • URL: https://news.kyobo.com/교보생명-교보더안심치매간병보험-출시/ (percent-encoded live)

  • Accessed: 2026-09-03, Retrieved: yes (HTML rendered)

  • Used for market spread only, never for a modelled parameter: a current whole-of-life 치매·간병 design with 진단자금 at three CDR tiers plus a 매월 생활자금 for life with a 36-instalment guarantee, 1~5등급 and 인지지원등급 cover with separate 재가 / 시설 / 방문요양 이용수당 riders, a premium refund on a 1~4등급 진단, 가입나이 30~75세 and 보험기간 종신. It is the evidence that the 120-month cap the composite carries is a choice among observed shapes rather than the only one on sale.

Regulatory and actuarial references#

Seventeen entries: three statutory or official restatements R1 R2 R3, four official statistical or fiscal sources R4 R6 R7 R11, three research-institute or supervisory documents R8 R9 R10, two further legal entries R12 R13, one supervisory guideline that was not retrieved in original R14, and four news or trade-press items R15 – R18 whose figures inherit that weakness and are labelled at every point of use.

R1 — 노인장기요양보험법 (Act on Long-Term Care Insurance for the Aged)#

  • Publisher: 국가법령정보센터 (법제처). Doc type: statute

  • URL: https://www.law.go.kr/LSW/lsInfoP.do?lsId=010436&ancYnChk=0 (index); article text through the print form https://www.law.go.kr/LSW//lsBdyPrint.do?…&joNo=<NNNN>:00&…

  • Accessed: 2026-09-03, Retrieved: in part — the index page returns only the site chrome and the currency line (현행 법률 제21690호, 시행 2026-05-26). The per-article print form works: 제2조, 제15조 and 제23조 were retrieved in full, but at the 2018-09-14 consolidation, so every article text quoted from it is the 2018 text and is flagged as such at the point of use

  • Establishes the scheme the benefit trigger belongs to: 제2조’s 「노인등」 and 장기요양급여, 제15조’s 등급판정 and the delegation of the grade boundaries to 대통령령 — the basis-change risk the insurer does not control — and 제23조’s exhaustive 급여 list. The 시행령 detail is cited from REG-R55, which was retrieved at current consolidation.

R2 — 노인장기요양보험법 시행령 [별표 1] 「노인성 질병의 종류」 (제2조 관련)#

  • Publisher: 국가법령정보센터; 개정 2022-12-20. Doc type: annexe to a presidential decree

  • URL: https://www.law.go.kr/LSW/flDownload.do?gubun=&flSeq=135370071&bylClsCd=110201

  • Accessed: 2026-09-03, Retrieved: yes (1-page PDF, extracted cleanly, full table read)

  • The closed list of 25 diseases with KCD codes that alone let a person under 65 be certified — four dementia codes, one Alzheimer, fourteen cerebrovascular, four Parkinson-family, plus 척수성 근위축, 다발경화증, 중풍후유증 and 진전. It is why the model has no prevalence data at all below 65 and carries the two entry rates down on sub65_factor_at instead, and why no cancer, musculoskeletal or frailty route exists below that age.

R3 — 찾기쉬운 생활법령정보, 「노인장기요양보험 > 등급판정 절차 및 기준」#

  • Publisher: 법제처 (easylaw.go.kr). Doc type: official plain-language restatement

  • URL: https://easylaw.go.kr/CSP/CnpClsMain.laf?popMenu=ov&csmSeq=2038&ccfNo=2&cciNo=2&cnpClsNo=1

  • Accessed: 2026-09-03, Retrieved: yes (HTML rendered)

  • The 장기요양인정점수 band table for 1등급 through 인지지원등급 with each grade’s 심신 기능상태, cited to 법 제15조 and 시행령 제7조제1항, and the 30-day determination deadline with a 30-day extension (법 제16조제1항). The deadline is what fixes the model’s claim-date convention at the 판정일; the bands corroborate REG-R55, read at first hand.

R4 — 국민건강보험공단, 「2024 노인장기요양보험 통계연보」#

  • Publisher: 국민건강보험공단 빅데이터사업실, 발간 2025-06-30. Doc type: official national statistics (국가승인통계)

  • URL (landing): https://www.nhis.or.kr/nhis/together/wbhaec07200m01.do; the attachment actually used: the same path with ?mode=download&articleNo=11003958&attachNo=368127

  • Accessed: 2026-09-03, Retrieved: yes, in the form that matters. The yearbook PDF (attachNo=361507) failed — the fetcher refused it at maxContentLength size of 10485760 exceeded — and the 해설서 (attachNo=364866) is an HWP 5.0 compound document that was not decoded. The third attachment (5.6 MB) is a ZIP of five .xlsx workbooks holding the yearbook’s complete numerical tables, which were read with openpyxl. Every figure is cited by table number, never by page, because the paginated PDF was never opened

  • The single most load-bearing source in this product. 표2-9 (등급별 × 연령별 판정 현황) and 표1-2 (65세 이상 적용인구 by five-year band and sex) give the 연령별 인정률 that is prevalence_table.csv and the age-banded grade composition that is grade_share_table.csv; 표2-5 (인정신청 구분별) gives the 13.3% / 69.8% first-application split behind direct_entry_share and the 107,365 등급변경신청 that bounds the absorbing-state simplification; 표2-1 and the 인정자 series give the stock growth and the duration bracket behind care_mort_mult. Everything in the morbidity basis traces here.

R6 — 국회예산정책처, 「2018~2027년 노인장기요양보험 재정전망」 (2018-12)#

  • Publisher: 국회예산정책처 추계세제분석실 사회비용추계과. Doc type: legislature budget-office fiscal projection, 94 pp.

  • URL: https://www.nabo.go.kr/board/file/down.do?fid=33315305

  • Accessed: 2026-09-03, Retrieved: yes (1.5 MB PDF, extracted and read)

  • The clearest published description of the assessment instrument — 12개 영역 90개 항목 surveyed, 52개 항목 entering the 장기요양인정점수 across ADL, IADL, 인지기능, 행동변화, 간호처치 and 재활 — and of the creation of the 인지지원등급 on 2018-01-01, which is the library’s standing example of a trigger enlarged by decree at no additional premium, and the level shift that contaminates any Korean experience series crossing that date.

R7 — 보건복지부, 「2023년 치매역학조사 및 실태조사 결과 발표」 보도자료#

  • Publisher: 보건복지부 (배포 2025-03-13); survey by 중앙치매센터, 한국보건사회연구원 and 한국갤럽. Doc type: government press release announcing an official epidemiological survey

  • URL: https://www.mohw.go.kr/board.es?mid=a10503010100&bid=0027&act=view&list_no=1484959

  • Accessed: 2026-09-03, Retrieved: yes for the release body (HTML). The two 별첨 attachments and the attachPreview.es viewer returned chrome only, so the survey’s diagnostic instrument is unverified — which matters because the private trigger is the CDR and a prevalence measured on another instrument is not directly transferable

  • The dementia rider’s whole basis: 65세 이상 유병률 9.25%, the five band prevalences (65-69 4.99%, 70-74 5.03%, 75-79 10.70%, 80-84 15.57%, 85+ 21.18%) that are dementia_table.csv, and the sex split (남 8.85%, 여 9.57%) that is dem_factor_m / dem_factor_f — including the 65-79 / 80+ crossover the model’s flat-in-age factors do not reproduce, which model.md names as a weakness.

R8 — 보험연구원, 「최근 치매보험시장의 이슈와 과제」 (KIRI 이슈분석, 2019-05-13)#

  • Publisher: 보험연구원 (정성희, 문혜정). Doc type: research-institute analysis, 4 pp.

  • URL: https://www.kiri.or.kr/pdf/전문자료/KIRI_20190510_9467.pdf (percent-encoded live)

  • Accessed: 2026-09-03, Retrieved: yes (272 kB PDF, read in full)

  • The definitive account of the 2018–2019 치매보험 boom — 초회보험료 약 233억 원 in 2018, 3.5× the prior year — the CDR scale with its seven points, the 약관 defect that produced the supervisory intervention, the industry’s self-imposed ₩30,000,000 aggregate limit, and the epidemiology behind the tier ladder: 경증치매 is 67% of all dementia cases, which is what the product-spec checks the market’s own ≈ 3 : 1 tier pricing against.

R9 — 보험연구원, 연구보고서 2019-11, Ⅱ장 「우리나라 장기요양서비스/보험의 현황 및 평가」#

  • Publisher: 보험연구원. Doc type: research report chapter, 30 pp. in the retrieved extract

  • URL: https://www.kiri.or.kr/pdf/연구자료/연구보고서/nre2020-0129_2.pdf

  • Accessed: 2026-09-03, Retrieved: yes (651 kB PDF, extracted and read)

  • The three benefit-trigger archetypes taken from a 2012 금융감독원 release — ① 회사기준, ② 공적기준, ③ 혼합형 — with the finding that 「일반적으로 간병보험에서는 공적기준을 적용하는 경향이 있다」, which is the taxonomy this product’s design sits in and the reason the composite is a type ②. Also the public/private comparison (근거법, 도입시기 2003-08 against 2008-07, 현물 against 정액), the 2019 product count (99 products, 생보 46 / 손보 53), the 간병보험 가입률 of 2.5%, and the statement that 보험개발원 publishes no long-term-care incidence table — the negative result the whole basis construction follows from.

R10 — 금융감독원, 「치매보험금 분쟁을 선제적으로 예방할 수 있도록 보험약관을 개선하겠습니다」#

  • Publisher: 금융감독원, 2019-07-02. Doc type: supervisory press release announcing a 약관 변경권고

  • URL (as retrieved): https://eiec.kdi.re.kr/policy/materialView.do?num=190159 — the KDI 경제교육·정보센터 mirror; the FSS original was not located

  • Accessed: 2026-09-03, Retrieved: yes, through the mirror

  • The two reforms that fixed the current dementia design: a 치매진단 must rest on a comprehensive clinical assessment rather than on imaging as a gate, and 특정 치매질병코드 and 약제투약 conditions were deleted from 약관. Revised products went on sale from October 2019, which is why dementia_wait_mths is a market answer and not a carrier choice — and why a Korean dementia experience series spanning 2019 is contaminated at the level of the benefit definition.

R11 — 한은정·황라일·이정석, 「장기요양 인정자의 사망 전 의료 및 요양서비스 이용 양상 분석」#

  • Publisher: 한국사회정책 제25권 제1호 (2018), pp. 99–123; authors at 국민건강보험공단 건강보험정책연구원 and 신한대학교. Doc type: peer-reviewed journal article on the NHIS claims census

  • URL: https://journal.kci.go.kr/kasp/archive/articlePdf?artiId=ART002331210

  • Accessed: 2026-09-03, Retrieved: yes (1.8 MB PDF; abstract, methods and results read)

  • The only post-onset survival evidence retrieved anywhere. 271,474 people certified 2008-07 to 2012-12 who also died inside that window: mean time from 등급인정 to death 516.2 days, 8.7% inside one month, 45.6% inside one year; 74.7% aged 75 or over; mean 장기요양인정점수 at death 82.1, inside 2등급. The sample is truncated by construction — only people who died inside a 4.5-year window are in it — so 516.2 days is a lower bound on duration, which is why care_mort_mult = 3.0 takes its shape from this study and its level from the yearbook duration bracket, and why light_mort_mult is bounded below the care multiple by the 82.1 mean score.

R12 — 보험업법 제4조 (보험업의 허가)#

  • Publisher: 국가법령정보센터. Doc type: statute

  • URL: https://www.law.go.kr/LSW/lsSideInfoP.do?lsiSeq=265389&joNo=0004&docCls=jo&…

  • Accessed: 2026-09-03, Retrieved: in part — the article’s structure and 제1항제3호 were returned, but not the full sub-paragraph 목 lettering; the identical article is retrieved from a mirror in _research/ci-insurance.md and as REG-R1 here

  • 제3호 제3보험업 comprises 상해보험, 질병보험 and 간병보험: long-term care is a statutory 보험종목 in its own right and not a sub-species of 질병보험, writable by both licences — which is why the two life-side and two non-life documents in this file describe the same benefit.

R13 — 노인장기요양보험법 시행령 제8조 (장기요양인정 유효기간), 2025-07-01 개정#

  • Publisher: 보건복지부 / 국민건강보험공단, retrieved through a long-term-care software vendor’s customer notice restating the amendment. Doc type: secondary — a vendor notice, not the decree

  • URL: https://www.carefor.co.kr/cs/view_notice.php?calmgno=45794&…

  • Accessed: 2026-09-03, Retrieved: yes, but as a restatement. The decree text itself was not retrieved, so the values are unverified

  • The base 유효기간 of two years, extended on renewal at the same grade — 1등급 4 → 5 years, 2~4등급 3 → 4, 5등급 and 인지지원 unchanged. Used in technical-notes.md only inside the duration estimator’s sensitivity, never as a load-bearing input, precisely because it is a restatement.

R14 — 금융위원회·금융감독원, 「IFRS17 주요 계리가정 가이드라인」 (2024-11-07)#

  • Publisher: 금융위원회 / 금융감독원. Doc type: supervisory guideline — not retrieved in original

  • URL (release index): https://www.fsc.go.kr/no010101/83351

  • Accessed: 2026-09-03, Retrieved: no. The content comes from contemporaneous press accounts surfaced in search and is secondary; the same guideline is cited from a retrieved 보도자료 as REG-R27

  • The log-linear lapse model converging to 0% at the premium-completion point as the 원칙모형 for 무·저해지 business, with 선형-로그 and 로그-로그 permitted by exception subject to quarterly disclosure of the difference. The two endpoint values are verified from REG-R27; the functional form is unverified at instrument level, and lapse_rate(t) says so.

R15 — 헤럴드경제, 「금융당국, 간병인보험 구조 진단 착수」#

  • Publisher: 헤럴드경제 (biz.heraldcorp.com). Doc type: news article

  • URL: https://biz.heraldcorp.com/article/10797305

  • Accessed: 2026-09-03, Retrieved: yes

  • The 간병인사용일당 loss-ratio record — life sector near 100% at August 2024 against 18.7% two years earlier, non-life 83.1%, premium at the five largest non-life carriers up roughly twenty-fold in four years. Cited only to scope that product out: it is a hospital-days frequency-severity cover that shares nothing with this one but the word 간병, and a reader comparing Korean market commentary against this library would otherwise conflate the two.

R16 — 보험신보, 「이슈 — 장기요양등급 인증자 증가하는데」#

  • Publisher: 보험신보 (insweek.co.kr). Doc type: news / trade-press feature

  • URL: https://www.insweek.co.kr/news/articleView.html?idxno=64798

  • Accessed: 2026-09-03, Retrieved: yes

  • The 인정자 and 노인인구 대비 인정률 series 2018–2023, whose 2020-onward values agree with R4 and which give the 71.8% six-year stock growth behind the stationarity caveat; private-market penetration at 2023; the 생명보험협회 2022 survey in which 40.8% named 간병보험 as the cover they would buy next; and a survey of 2024–2025 product launches. Every figure inherits the source’s weakness and is labelled a news figure at each use.

R17 — 생생비즈, 「치매 환자 100만명 시대 커지는 치매·간병보험 시장…작년 70% 급성장」#

  • Publisher: 생생비즈 (livebiz.today). Doc type: news article

  • URL: https://www.livebiz.today/news/articleView.html?idxno=6054

  • Accessed: 2026-09-03, Retrieved: yes

  • 치매·간병보험 초회보험료 of ₩88.4bn for January–November 2024, +70.2% year on year, the 계속보험료 series, and a product survey naming the CDR 검사지원비 and lecanemab covers that product-spec.md records as out-of-scope market features. Market context only; no modelled parameter rests on it.

R18 — 헬스코리아뉴스 and 백세시대, 2024 통계연보 launch coverage#

  • Publisher: 헬스코리아뉴스 (hkn24.com); 백세시대 (100ssd.co.kr). Doc type: news articles reporting the yearbook’s release

  • URLs: https://www.hkn24.com/news/articleView.html?idxno=345333 ; https://www.100ssd.co.kr/news/articleView.html?idxno=123152

  • Accessed: 2026-09-03, Retrieved: yes (both)

  • Used as a cross-check on R4 and for the five-year 인정자 series 857,984 (2020) → 1,165,030 (2024) that gives the roll-forward estimator its +67,117 net increase and hence the 4-to-5.5-year duration bracket behind care_mort_mult. Every figure that also appears in R4 agrees with it, which is the point of carrying a news source here at all.

Cross-product references ([REG-R#])#

[REG-R#] tags resolve against references/regulatory-and-actuarial-references.md, whose own R1–R62 numbering is distinct from this file’s and is likewise frozen. Within that page plain [R#] refers to its own entries, so the two schemes must never be read across. The forty-one entries this product’s documents cite, all accessed 2026-09-03:

  • REG-R1 — 보험업법 제2조·제4조: the 생명보험 / 손해보험 / 제3보험 split and the 제4조제3항 deeming provision that lets both licences write 간병보험. Retrieved: yes. Cited beside this file’s R12.

  • REG-R2 — 보험업법 제5조·제127조 등 (기초서류): the 산출방법서 is filed with the FSC and never published, which is why the office premium is a model-point input and every pricing parameter not in [S1] is std. Retrieved: yes.

  • REG-R3 — 보험업법 제120조: the statutory duty to accumulate 책임준비금, cited as a layer this model does not compute. Retrieved: yes.

  • REG-R4 — 보험업법 제176조 (보험요율 산출기관): 참조순보험요율 are filed, not published — the statutory reason a Korean morbidity basis is not simply looked up. Retrieved: yes. Not this file’s R4.

  • REG-R5 — 보험업법 제181조·제184조: the 선임계리사 who owns the basis and its verification. Retrieved: yes.

  • REG-R7 — 보험업법 시행령 제1조의2 (보험상품): the closed product scope, cited in the boundary paragraph that separates this private contract from the public scheme. Retrieved: yes.

  • REG-R9 — 보험업감독규정 (고시 제2026-16호): the 기준연령 요건 「남자가 만 40세」 that makes the anchor cell the prescribed disclosure cell, and the finding that a full-text search of the 고시 returns zero occurrences of 예정이율. Retrieved: yes (226,083 characters).

  • REG-R10 — 감독규정 제6-11조 등 (책임준비금, 보증준비금): the reserve taxonomy, cited and not computed. Retrieved: yes.

  • REG-R11 — 감독규정 제6-11조의6 (해약환급금준비금): the Korea-only appropriation that measures against a 제7-66조제1항 surrender value even for 제4항 products that contractually pay less — which is exactly this product’s 미지급형 form. Retrieved: yes.

  • REG-R12 — 감독규정 제6-11조의7·제6-13조 (계약자배당): what a 무배당 contract switches off. Retrieved: yes. Not this file’s R12.

  • REG-R13 — 감독규정 제7-1조 등 (K-ICS): in force from 2023-01-01, cited in the measurement-basis paragraph; no 요구자본 is computed. Retrieved: yes (the 별표 22 detail was not).

  • REG-R14 — 감독규정 제7-17조~제7-19조 (적기시정조치) and the 고시 transition: the thresholds behind the “both regimes commenced together” statement. Retrieved: yes.

  • REG-R17 — 감독규정 제7-63조 (제3보험의 보험상품설계): 제1항제1호 requires payment of the 계약자적립액 on death from a cause the policy does not cover, and termination. It is the whole of claims_death — a third of premium income on the anchor cell — and it is the single most load-bearing cross-product entry for this product. Retrieved: yes.

  • REG-R18 — 감독규정 제7-64조·제7-65조 (산출방법서, 계약자적립액): the 필수기재사항, the 현금흐름방식 requirement, and the accrual of the 계약자적립액 monthly before 납입완료 and daily after, which is the timing av_pp implements. Retrieved: yes (the accrual formulas did not extract).

  • REG-R19 — 감독규정 제7-66조 등 (해약환급금): the 계약자적립액 − 해약공제액 floor at zero, the 해약공제기간 capped at seven years that bounds surr_chg_pp, and 제4항, which permits the 미지급형’s reduced value precisely because premiums were computed on a 최적해지율 — so the lapse vector is a condition of the form’s legality. Retrieved: yes.

  • REG-R20 — 감독규정 [별표 14] 표준해약공제액: the statutory cap the 13-months rule of thumb stands in for. Retrieved: yes (1-page PDF, full text).

  • REG-R21 — 감독규정 [별표 15] 보험가입금액의 산정: 제9호 excludes 「치매 또는 일상생활장해 등 타인의 간병을 필요로 하는 상태」 risk premium from the ratio that gives a contract with no death benefit its notional 보험가입금액 — read literally, the schedule cannot be applied to a care-only contract, which is why surr_chg_ratio uses REG-R29’s rule of thumb instead. Retrieved: yes (1-page PDF).

  • REG-R22 — 감독규정 제4-32조·제7-45조·제7-51조: the commission cap behind comm_init_mths and the 보험가격지수, the ratio that exists because a Korean consumer cannot see the pricing basis. Retrieved: yes.

  • REG-R25 — 시행세칙 [별표 15] 표준약관: the supervisor’s model conditions, and why the four carrier documents here agree almost word for word. Cited by article — 제21조 (보험나이, against which the model’s 만나이 basis is stated), 제22조 (the 계약자적립액 on non-covered death), 제26조 (부활), 제33조 (보험계약대출, which this form has nothing to lend against), 제17조 (청약철회). Retrieved: yes (a 492-page PDF read in full).

  • REG-R26 — 시행세칙 [별표 22] (K-ICS) and [별표 24] (보증준비금): Retrieved: no. The 대량해지 shock, including the 고환급형 test, is known only at second hand and anything resting on it is unverified — which matters here because the representative form is a 무해지 one.

  • REG-R27 — 제4차 보험개혁회의 보도자료 (2024-11-07, 계리가정·할인율): the source of lapse_completion = 0.001 and lapse_ultimate = 0.008, of the log-linear 원칙모형 shape, of the disclosure requirement that makes shipping the pyojun comparison vector the right thing to do, and of the 63.8% 무·저해지 share of 2024 H1 보장성 초회보험료. Retrieved: yes (the 보도자료 and its 별첨; the guideline attachment itself is R14, not retrieved).

  • REG-R28 — 무(저)해지환급금 상품구조 개선 (2020) and the FSS 소비자경보 (2019): the 환급률 cap that shapes the published progression, and the no-policy-loan consequence that is a stated product limitation here. Retrieved: yes.

  • REG-R29 — 「불합리한 보험 사업비와 모집수수료를 개편하여…」 (2019-08-01): the 13-months-of-premium rule of thumb for a 보장성보험’s 표준해약공제액 and the 60% cap on annual commission — together expense_acq_mths = 5.2 and comm_init_mths = 7.8. Retrieved: yes.

  • REG-R32 — 예금자보호한도 1억원 (2025-09-01): the protection sentence 표준약관 제43조 requires the 약관 to carry. Retrieved: in part.

  • REG-R33 — 제10회 경험생명표, as reported: the 65세 기대여명 23.7 / 27.1 that calibrates mort_table.csv and the 평균수명 86.3 / 90.7 the construction reproduces without being asked to. Retrieved: yes — a news article, and the only retrieved source for these figures.

  • REG-R34 — 보험개발원 public channels: the evidence that the 경험생명표 is not published in full, which is why the mortality table is a construction rather than a copy. Retrieved: in part.

  • REG-R35 — K-ICS 경과조치 (KIRI): background for why a ratio quoted after 경과조치 is not comparable with one quoted before. Retrieved: yes.

  • REG-R38 — 「2024년 생명표 작성 결과」 / 「2023년 생명표」: the public 기대여명 anchor the insured-lives construction is compared against, and the 만나이 basis every public series here shares. Retrieved: yes.

  • REG-R39 — KOSIS 완전생명표 (single-year qx): Retrieved: no — distributed through KOSIS, which redirected to an SSO endpoint. The named to-do in the table-build note.

  • REG-R42 — 국민건강보험공단 「장기요양 등급 판정 현황」 (경영공시): the 2026-06-30 cross-check on the grade composition — 1등급 3.8% / 2등급 7.1% of all assessed — at a later as-of date and a wider denominator than R4, with the same shape. Retrieved: yes.

  • REG-R43 — the 2024 통계연보 as reported by 메디칼월드뉴스: Retrieved: no (search summary only). Cited once, for a figure used nowhere in the model, and tagged unverified where it appears.

  • REG-R48 — 평균공시이율 / 공시기준이율 carrier disclosure: the evidence that the 예정이율 of a specific Korean product is not a published number, which is what makes [S1]’s disclosure of 2.0% unusual enough to be worth naming. Retrieved: in part.

  • REG-R49 — 상법 제4편 제1장 통칙: the statutory floor under the contract — 제638조의3 (약관교부·설명), 제651조 (고지의무), the three-year claim prescription. Quoted, not modelled. Retrieved: yes.

  • REG-R50 — 상법 제4편 제3장 인보험: 제736조 (보험적립금반환의무), the statutory floor beneath 감독규정 제7-63조제1항제1호’s payment on a non-covered death. Retrieved: yes.

  • REG-R51 — 금융소비자보호법 제46조 (청약의 철회): the 15/30-day withdrawal right, out of scope because the model starts from the point cover is in force. Retrieved: yes.

  • REG-R52 — 예금자보호법 시행령 제18조: the calculation basis behind the ₩100,000,000 limit. Retrieved: yes.

  • REG-R54 — 노인장기요양보험법 제2조·제15조·제23조·제39조·제40조, at current consolidation: the statutory definition of 「노인등」 — the 만 65세 test that is inside the benefit definition and one of the two reasons this model runs on 만나이 — the 등급판정 machinery and the 급여 list. Load-bearing, and the entry to prefer over this file’s R1, which is the 2018 text. Retrieved: yes.

  • REG-R55 — 노인장기요양보험법 시행령 제7조 (등급판정기준) and [별표 1]: the 점수 bands that are the statutory definition of this product’s trigger and the source of its grade-severity split, plus the 노인성 질병 list. Retrieved: yes (Decree text and the 별표 as a 1-page PDF).

  • REG-R57 — 소득세법 제59조의4 (보장성보험료 세액공제): the 12% credit on up to ₩1,000,000 of premium — a credit, not a deduction — inside which the anchor’s ₩67,200 annual premium sits comfortably. Not modelled. Retrieved: yes.

  • REG-R60 — K-IFRS 제1117호 제정 의결: IFRS 17 mandatory in Korea since 2023-01-01, cited in the measurement-basis paragraph. No krlib model computes a CSM or a risk adjustment. Retrieved: in part (the standard text itself was not).

  • REG-R61 — 보험개발원 「장기손해보험 참조순보험요율」 공시: cited here for the contrast it draws. The display carries a published 암 발생률 grid, which is why Cancer_KR_S sources its incidence rather than standardizing it; it carries nothing for 장기요양, which is why this product’s basis has to be converted from a prevalence in the open. Retrieved: yes.

Provenance note#

Every entry above traces to _research/long-term-care.md, which is the citation ground truth for this product: the S# and R# numbering used here is that file’s numbering, unchanged, and it is never renumbered because these documents cite against it. The research file’s numbering is not this one’s — it runs to S5 and R18 and carries R5, which this file omits as uncited for the reason given at the head of this page.

What lives there and not here: the per-document extraction record, table by named table for the yearbook workbooks; the full arithmetic of the prevalence-to-incidence conversion and of the two independent duration estimators; the carrier-by-carrier variation tables behind the threshold, waiting-period, 감액 and surrender-value spreads; and the register of fetch failures and their consequences. The last is worth summarising, because several of its entries are why figures in these documents carry unverified:

  • The 2024 통계연보 PDF itself exceeded the fetcher’s size limit and the 해설서 is an undecoded HWP, so the yearbook’s own definitions of 인정자, 판정 and 수급자 are inferred from the tables’ totals rather than quoted from the compiler — sound, because the totals reconcile, and stated rather than hidden. Citations are by table number, never by page.

  • The 중앙치매센터 「대한민국 치매현황」 report was blocked by a web application firewall in every vintage, and the 보건복지부 별첨 viewer returned chrome only. So the dementia basis rests on the press release R7 alone, and the survey’s diagnostic instrument is unverified.

  • KOSIS redirected to an SSO endpoint REG-R39 and the 노인장기요양보험법 index page returns only chrome, so the statute was read article by article at the 2018 consolidation R1 with the current text taken from REG-R54 and REG-R55; the 유효기간 comes from a vendor restatement R13 and is used only inside a sensitivity.

  • The IFRS17 계리가정 가이드라인 attachment was never converted from HWP R14, so the lapse model’s functional form is unverified at instrument level while its two endpoint values are verified from the 보도자료 REG-R27; and 보험업감독규정 별표 22 was not retrieved REG-R26, so the K-ICS 대량해지 shock is second-hand.

  • Two carrier documents were lost: a fifth carrier’s 간편심사 간병보험 약관 (HTTP 503) and a major non-life carrier’s current disclosure, which is a scanned image PDF with no text layer. No 삼성생명 product parameter appears anywhere in this product, the largest life carrier being represented only by its non-life affiliate [S3] and by secondary mentions.

What no source in this file provides, and no amount of further searching would. There is no published Korean long-term-care incidence table, no post-onset mortality table by grade, no recovery or grade-transition matrix, no durational persistency series for a 보장성 contract, no expense or commission scale, and no 질병 / 상해 split of certifications. Every one of those is a std parameter in model.md, each with its rationale and whatever observed range the retrieved documents bound — and for several of them, the honest answer recorded there is that the documents bound nothing at all.

On this page
Show Source