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/per_assurance/, 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 PER_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
allocation_grid_file	allocation_grid()	allocation_grid.csv
mort_table_file	mort_table()	mort_table.csv
exit_table_file	exit_table()	exit_table.csv
annuity_factor_file	annuity_factor_table()	annuity_factor.csv

Four of the five carry a compound key, because every one of them is a ladder rather than a single number. allocation_grid is indexed by (allocation_profile, years_to_horizon) so the four qualified profiles of the de-risking arrêté sit in one file and an insurer ladder finer than the four regulatory bands substitutes without touching a formula; mort_table by (sex, age); exit_table by (compartment, duration), because compartment 3 is closed to the main-residence early-release case and therefore does not carry the same decrement as compartments 1 and 2; and annuity_factor_table by (sex, age).

The glide path being a file is the point of the file list. It is the product’s dominant financial lever — moving the équilibré grid to the prudent one replaces most of a 5.00% UC return with a 3.38% euro return over the anchor cell’s twelve years — and the regulatory grid is a minimum that insurers may sit above and restate unilaterally. A published grid is a snapshot, so it belongs in a table a reader can edit and diff, not in a chain of if statements.

What is not in a file, and why. The charge levels — the entry loading, the two management charges, the arbitrage rate, the frais d’arrérages — are Projection References rather than table columns. The encadré requires maxima to be disclosed and caps nothing, so a charge level is never a contractual constant, and the sampled range is wide: entry loadings from 0% to 4.80%, euro management charges from 0.50% to 2.30%. Every adopted level is a standardization, and putting them where a reader trips over them is better than filing them in a table that looks like data.

To swap in a licensed basis — TH 00-02 / TF 00-02 for the death benefit during accumulation, or the generational TGH05 / TGF05 for the annuity conversion, neither of which is redistributed here — replace mort_table.csv or annuity_factor.csv with a same-schema file, or point mort_table_file / annuity_factor_file at a different name, then 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.

allocation_grid()

The gestion pilotée par horizon glide path, from allocation_grid.csv.

Indexed by (allocation_profile, years_to_horizon), with an euro_share and a uc_share column. The four qualified profiles of the de-risking arrêté are shipped at their regulatory minimum — the équilibré ladder is 0 / 20 / 50 / 70% in the low-risk bucket as the horizon closes — because in this market the regulatory grid is not a floor insurers beat; it is the product. dynamique and offensif are shipped identical, which is what the arrêté says rather than an oversight.

The band edges are a [std] convention: the tighter minimum applies at a boundary, so k = 10 reads 20% and not 0%. Sorted on read, because Projection.alloc_euro indexes into it.

mort_table()

The base annual mortality rates by sex and age, from mort_table.csv.

A [std] proxy shaped on French population mortality, which through INSEE is the only freely redistributable French mortality data. The homologated tables that actually govern the death benefit during accumulation — TH 00-02 and TF 00-02, and the décalage d’âge age shifts annexed with them — are cited and not shipped.

The level is anchored, not measured: the whole table is scaled by one constant so that Projection.mort_be_factor times q(M, 52) reproduces the technical notes’ flat 0.00500 placeholder exactly, which is what makes the worked example close on either mortality basis. That places the table above French population mortality, because the placeholder is above it. Sorted on read, because Projection.mort_rate indexes into it.

exit_table()

The two accumulation-phase exit decrements, from exit_table.csv.

Indexed by (compartment, duration), with an early_release_rate and a transfer_out_rate column. The duration key is the plan’s own ancienneté year and is 1-based: its first row is the plan’s first year, so Projection.early_release_rate keys it with plan_year(t) — the 1-based label — and not with the 0-based duration(t).

Neither is a lapse rate and the file deliberately does not use the word: the plan carries no surrender right, and these are a déblocage anticipé on one of the seven statutory cases and a transfer of acquired rights to another PER.

Both are [std]. The one citable anchor is an aggregate — early releases and transfers together were €1 651 m against €63.0 bn of accumulation-phase provisions in 2024, i.e. 2.62% — which the shipped split of 1.60% / 1.00% reproduces to 2.60%. Two caveats travel with it: it is an amount ratio adopted as a policy decrement rate, and it is contaminated by a book growing 18.7% a year, so it is not a steady-state rate.

The compartment key earns its place. Compartment 3 is excluded from the main-residence limb of L. 224-4, which is the only discretionary case of the seven, so it carries a lower early-release rate; the transfer rate does not vary by compartment, because a transfer moves the rights without changing them. Sorted on read, because Projection.early_release_rate indexes into it.

annuity_factor_table()

The annuity conversion factors by sex and age, from annuity_factor.csv.

A [std] placeholder ladder anchored at 22.0000 for a male aged 64 — annual in arrears, no reversion — which is the figure the technical notes’ worked example uses. No sampled insurer publishes an annuity rate card: the contracts say only “la table de mortalité en vigueur” and “le taux d’intérêt technique en vigueur”, and the homologated generational TGH05 / TGF05 are cited and not shipped.

At the 0% technical rate a PER tariff is capped at, the factor is an undiscounted expected-instalment count rather than a discounted annuity factor — so this file holds a number of payments, and applying any discount to it would be a second, unauthorised technical rate. Sorted on read, because Projection.annuity_factor indexes into it.