The Data Space

Input data shared by every by-policy projection.

The five input CSVs are read here, once per model, and referenced from Projection as data. Projection is parameterized by point_id, so each Projection[N] is a separate ItemSpace with its own cells cache; if the readers lived there, every model point would re-read every file. Holding them in an unparameterized Space reads each file once no matter how many policies are projected.

Inputs are external files: plain CSVs in the model folder’s parent directory, products/medical/, rather than data stored inside the model. The model folder therefore holds nothing but formulas — no _data/, no IOSpec, no embedded values — so a diff of the model shows logic changes only. This follows annuallife.TradLife_A; contrast basiclife.BasicTerm_S, which keeps its inputs inside the model through modelx’s IOSpec machinery.

The consequence worth knowing: the model is not portable on its own. Copying the Medical_JP_S folder without its parent’s CSVs produces a model that reads and then fails on first evaluation.

input_dir() resolves the directory from _model.path.parent at run time, so the model works wherever the repository is checked out. Each table has a filename Reference and a reader Cells:

Reference	Cells	File
model_point_file	model_point_table()	model_point_table.csv
mort_table_file	mort_table()	mort_table.csv
lapse_table_file	lapse_table()	lapse_table.csv
incidence_table_file	incidence_table()	incidence_table.csv
los_table_file	los_table()	los_table.csv

The mortality table is a proxy, and deliberately so

第三分野標準生命表2018 is public, free and machine-readable at a stable 日本アクチュアリー会 URL — the sharp contrast with uklib, which had to proxy subscriber-only CMI tables. Anyone can retrieve it and check a rate. But the publisher’s site terms prohibit reproduction, alteration and transmission to third parties without written consent, so this library does not ship a copy of it.

mort_table.csv is therefore a [std] construction, and it is the library-wide canonical one: the same file, value for value and provenance string for provenance string, ships in every jplib product that reads a third-sector rate, so a cell cannot disagree with itself across products. It is built on the union of the rates the research pass actually read out of 第三分野標準生命表2018 — 22 sourced anchors, male and female, age 0 to the terminal age — with piecewise log-linear (geometric) graduation between adjacent anchors: for a <= x <= b, q(x) = q(a) (q(b)/q(a))^((x-a)/(b-a)). That reproduces every anchor exactly by construction, so nothing sourced is disturbed, and it is locally the Gompertz family the publisher’s own table follows. The terminal age anchor — 116 male, 118 female, q = 1 — is what closes the projection horizon. Every row says so in its provenance column, which points at the IAJ entries rather than reproducing them: an anchor row records the quoted rate, an interpolated row records the two anchors it sits between. No conclusion about Japanese third-sector mortality should be drawn from it.

Two further things the provenance columns record, because they change what the numbers mean. The table is a valuation table whose margin runs the wrong way for a best estimate on a morbidity product — death releases the liability, so the table is set deliberately below national mortality — which is why Projection.mort_be_factor scales it up. And 第三分野標準生命表2018 excludes 高度障害, so a severe-disability state cannot be read out of it and the premium waiver module carries its own incidence.

The morbidity tables are constructions on public statistics

There is no published morbidity table in Japan: 日本アクチュアリー会 publishes the mortality basis only, and every insurer’s 危険発生率 sits in its unpublished 算出方法書. incidence_table.csv and los_table.csv are built from 患者調査, a 基幹統計 of 厚生労働省:

incidence_table.csv

入院受療率 per 100,000 by five-year age band and 退院患者平均在院日数 by the four broad bands each statistic is published at. 受療率 is a point-in-time prevalence, not an incidence, and the conversion inc = (juryoritsu / 100,000) x 365 / alos is an explicit [std] step — treating the published figure as a claim frequency is the commonest error in a Japanese medical model. The 概況 prints 入院受療率 for every five-year band, so all fifteen bands the model needs carry its citation and none is interpolated. The sex factors are [std]: the age x sex cross-tabulation lives in an e-Stat table that was not downloaded.

los_table.csv

A five-band discrete length-of-stay distribution per broad age band, with the probabilities in each row solved so that the row mean equals the sourced 平均在院日数 for that band exactly. The shape is [std] — the 32-band e-Stat grids were not downloaded — and it matters more than the mean it reproduces, because the 60-day limit bites on the tail and not on the mean.

To swap in a company basis, replace the CSVs with same-schema files, or point the *_file References at different names, and clear the cache. No formula changes.

Cells Descriptions

input_dir()

The directory holding the input CSVs: the model folder’s parent.

Inputs are external files, not data stored inside the model, so the model folder is pure formulas. The path is resolved at run time from where the model was read, following annuallife.TradLife_A.

model_point_table()

The model point table, read from model_point_table.csv.

Indexed by point_id. point_id = 1 is the technical notes’ worked-example anchor cell; the others exercise the product’s 型 elections, its riders, its 特則 and its edge ages.

mort_table()

The [std] mortality construction by sex and age, from mort_table.csv.

Not a copy of 第三分野標準生命表2018, whose publisher’s terms forbid redistribution: the library-wide canonical [std] construction, log-linear between the sourced anchors and exact at every one of them, including the male q40 the technical notes quote and the real table’s terminal ages. See the Space docstring. Read as the valuation rate; Projection.mort_be_factor turns it into a best estimate.

lapse_table()

The [std] annual lapse rates by policy year, from lapse_table.csv.

Anchored by construction to the only published industry-wide Japanese persistency figure, 解約・失効率 5.6% p.a. on 個人保険 — which is measured on opening in-force sum assured, a basis a 医療保険 with no sum assured cannot even enter. Policy years beyond the last row take that row.

incidence_table()

入院受療率 and 平均在院日数 by age band, from incidence_table.csv.

Indexed by age_start, the lower edge of the five-year band. alos_days repeats the broad-band 平均在院日数 that the five-year band falls in, because that is the granularity each statistic is published at [std]. inc_factor_m and inc_factor_f are the [std] sex factors on incidence.

los_table()

The [std] length-of-stay distribution, from los_table.csv.

Five stay_days bands per broad age band, keyed by band_start, with prob solved so the row mean equals the sourced 平均在院日数 for that band. A single mean is not usable on this product: the sourced per-cause means run from 2.4 days to 290.4, and a 60-day cap bites at one end and never at the other.