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/fondsgebundene_rentenversicherung/, 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 FRV_DE_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_file	mort_table()	mort_table.csv
lapse_file	lapse_table()	lapse_table.csv
charge_file	charge_table()	charge_table.csv
fund_scenario_file	fund_scenario_table()	fund_scenario_table.csv
rentenfaktor_file	rentenfaktor_table()	rentenfaktor_table.csv

Every file but ``model_point_table.csv`` carries a ``provenance`` column, one tag per row, and the library’s conventions suite asserts it. That is this library’s second ruling: the citation discipline reaches the data files rather than stopping at the prose. The model point table is the single exemption, because a model point is a configuration — one policy’s own terms — and not an assumption.

The two shipped decrement proxies, and what a replacement must preserve

Nothing in this product’s source corpus was retrieved, and no charge level, lapse rate or *Rentenfaktor* was established at any carrier. Three of the five assumption files are therefore standardizations end to end, and two of them stand in for tables this library may not ship.

mort_table.csv is a [std] Gompertz-form proxy of the first-order DAV 2008 T death table:

qx_tariff(x) = 0.00080 x 1.10^(x - 37),        ages 18 to 100

anchored so that ``qx_tariff(37) = 0.00080`` exactly. That single value is what the notes’ worked example rests on — it produces mort_rate_tariff_mth(0) = 0.00080/12 and, through mort_be_factor = 0.75, mort_rate(0) = 0.00060 — so a substitute table must reproduce it at the anchor cell’s entry age if the example is to close. What else a replacement must preserve is the direction of the two bases: DAV 2008 T is a death table with a first-order margin above best estimate, which is the opposite direction from an annuity table, and the 10 % per year of age is an insured-lives gradient — a proxy built on population mortality overstates claims at the working ages this product lives at. DAV 2008 T is the property of the Deutsche Aktuarvereinigung, is not public and is cited by name here rather than redistributed.

rentenfaktor_table.csv is [std] and derived, not observed. At a Rechnungszins of 0 % — the only conversion-basis statement with any corroboration anywhere in the delib corpus, and that at one remove and for a classic tariff — a monthly annuity of R per 10 000 € for an expected T years has present value 12 T R, so:

T_eff(x) = 100/3 - 0.75 (x - 67)            ages 60 to 75
rentenfaktor_guar(x) = 10000 / (12 T_eff(x)) = 10000 / (400 - 9 (x - 67))

which is exactly 25.00 at age 67, 22.47 at 62 and 26.81 at 70. Read the other way, 25.00 prices the guarantee as though the insurer holds the capital for 33⅓ years and earns nothing on it — the Sicherheitsabschlag made concrete. The std_2026 row sets the current factor equal to the guaranteed one, so the max(guaranteed, current) rule is exercised without injecting an unsourced uplift; rich_current sets it 12 % higher so that the max() visibly bites on one model point. The underlying table, DAV 2004 R, is generational and is likewise cited and not shipped: a replacement must preserve a generational annuitant basis with a first-order margin, which is a margin in the opposite direction from the death table above.

charge_table.csv, lapse_table.csv and fund_scenario_table.csv are standardizations of the same kind, each row carrying its own rationale in the provenance column. The one anchor in the charge stack is the Höchstzillmersatz of 25 ‰ of the Beitragssumme, and std_gross takes the cap rather than a guessed interior point.

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.

Thirteen policies indexed by point_id; model point 1 is the worked example’s anchor cell. This is the library’s one provenance-exempt input, because a model point is a configuration rather than an assumption.

mort_table()

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

A [std] proxy of DAV 2008 T, not the table itself; see the Space docstring for what it is, what it is anchored on and what a replacement must preserve. It is the basis of the Risikobeitrag the tariff charges — not of the projection’s own decrement, which is this table scaled by mort_be_factor, and not of the Rentenfaktor, which rests on an annuity table.

lapse_table()

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

The policy_year key is the contractual 1-based label, not the model’s t: Projection.lapse_rate_base(t) maps through policy_year(t) = t // 12 + 1, so the file’s values did not move when the frame became 0-based.

[std] throughout: no German unit-linked Stornoquote was established anywhere. The front-loading in years 1 to 5 is a structural inference from the exit terms — the acquisition charge is being taken and the value is furthest below the premiums paid — and the dip in years 11 and 12 anticipates the twelve-year tax threshold, whose step up is applied in Projection rather than stored here, because it depends on the attained age as well as the duration.

charge_table()

The charge scales by charge_id, read from charge_table.csv.

One row per tariff: the acquisition rate on the Beitragssumme and the months it is spread over, the premium-based and fund-based administration rates, the monthly Stückkosten, the Zuzahlungskosten rate and the Stornoabzug rate. Four rows ship — std_gross, std_netto, std_high, std_low — and the difference between the first two is the acquisition load, the parameter this library most needs and cannot source.

fund_scenario_table()

The gross fund return and TER by scenario and policy year, from fund_scenario_table.csv.

Four [std] deterministic paths — base, etf, zero, stress. The policy_year half of the key is the contractual 1-based label, reached through policy_year(t) = t // 12 + 1, so it is unaffected by the 0-based frame. The TER is a return item, never a policy charge: it is borne inside the Anteilspreis and never appears in the ledger, so the projection nets it off the gross return. Charging it explicitly double-counts; ignoring it overstates the policyholder’s return. Nothing here is a PRIIPs performance scenario and nothing here may be compared with one — those are derived from an underlying’s own return history, not chosen.

rentenfaktor_table()

The guaranteed and current Rentenfaktoren by factor id and age at Rentenbeginn.

Read from rentenfaktor_table.csv: euro of monthly annuity per 10 000 € of Fondsguthaben. [std] and derived rather than observed — see the Space docstring for the derivation and for what the underlying DAV 2004 R basis requires of a replacement. The table is indexed by the age at *Rentenbeginn*, not by the attained age in the last projected month, which is one lower.