The Data Space

Input data shared by every by-policy projection.

The six 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/temporaire_deces/, 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 TD_FR_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
premium_rate_file	premium_rate_table()	premium_rate_table.csv
mort_table_file	mort_table()	mort_table.csv
lapse_file	lapse_table()	lapse_table.csv
freq_loading_file	freq_loading_table()	freq_loading_table.csv
benefit_schedule_file	benefit_schedule()	benefit_schedule.csv

Two of the six deserve a word about what they are and are not.

premium_rate_table.csv is the one genuinely French artefact in the set: a published tarif de base annuel by attained age, in per cent of the guaranteed capital, shipped here as the decimal fraction the cotisation rule multiplies by. It is a real rate card of a 2019–2021 vintage, and its shape — including the +38 % step from age 59 to 60 against a trend of about +8 % — is what a fitted curve would smooth away. Use it for shape, not for level.

mort_table.csv is the opposite: a [std] Gompertz-form proxy, 0.00400 x 1.09^(age - 58) over ages 18–74, not a fitted or homologated table. The regulatory non-annuity tables TH 00-02 and TF 00-02 are annexed to the arrêté du 20 décembre 2005 and are cited by name in this library rather than redistributed, and no French insurer publishes a basis of its own. The 9 % per year of age slope is calibrated on the published tariff grid — the one observable French artefact — and sits at the top of it rather than inside a tight band: the grid rises at roughly 7–9 % per year of age from age 35, compounding at 7,7 % over ages 42–58 and 8,98 % over the whole rated span 35 to 74. A tariff gradient is not a mortality gradient, so both the level and the slope are placeholders. The anchor a substitute table must preserve is the rate at age 58 = 0.00400, so the notes’ worked example still closes. INSEE’s national series is the intended base for a replacement; it is population, not insured, mortality, and the INSEE page states no licence or reuse conditions — standard open-data terms are assumed there and that assumption is [unverified], so confirm before redistributing derived CSVs. Dropping a licensed or company table in place of this one changes the basis with no formula change.

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.

premium_rate_table()

The tariff rates by rate id and attained age, from premium_rate_table.csv.

An annual tarif de base by attained age, held as the decimal fraction of the guaranteed capital, so the published 1,05 % at age 58 is stored as 0.0105 and sum_assured x prem_rate is the annual cotisation before the surprime and the fractionation loading.

mort_table()

The annual death rates by attained age, from mort_table.csv.

A [std] proxy, not a homologated table; see the Space docstring for what it is and what a replacement must preserve.

lapse_table()

The lapse rates by policy year, read from lapse_table.csv.

The rates are annual rates by policy year; Projection.lapse_rate_mth converts them to the month. policy_year is the contractual 1-based label, 1 to 4, not the model’s 0-based month index: Projection.lapse_rate_base reads it through Projection.policy_year(t) = t // 12 + 1.

freq_loading_table()

The fractionation loadings and fees by payment frequency.

Read from freq_loading_table.csv: the number of instalments per policy year, the multiplier prem_freq_load embedded in the cotisation TTC and the fixed annual frais d’échéance prem_freq_fee, which is a euro amount and not a second percentage. The last two are the only disclosed charge figures in the whole source corpus, and the first is what Projection.prem_cycle and Projection.prem_inst_pp collect them on.

benefit_schedule()

The benefit factors by schedule id and policy year, from benefit_schedule.csv.

One schedule ships, constant, whose factor is 1.0 in every year: the capital of a French standalone temporaire décès does not amortize. The table exists so that a decreasing shape can be dropped in without a formula change; no source in the corpus gives one, so none is shipped.

policy_year is the contractual 1-based label, 1 to 57, not the model’s 0-based month index: Projection.benefit_factor reads it through Projection.policy_year(t) = t // 12 + 1.