No Grundsicherung im Alter if a child's annual Gesamteinkommen is 100,000 € or more (§ 43 Abs. 5 SGB XII, 2005–2019)

[MImmesberger]

Summary

• Implement the 100k children exclusion per § 43 Abs. 5 SGB XII (originally Abs. 2, BGBl. I 2003 S. 3022) for policy dates 2005–2019: applicants are excluded from Grundsicherung im Alter if any first-degree descendant has an annual Gesamteinkommen (§ 16 SGB IV) of 100,000 EUR or more. The threshold applies to each child individually.
• The Angehörigen-Entlastungsgesetz (BGBl. I 2019 S. 2135) repealed the exclusion as of 2020-01-01. Since then the 100,000 EUR limit only restricts the Unterhaltsrückgriff of the Sozialhilfeträger against the children (§ 94 Abs. 1a SGB XII), which does not affect the benefit — betrag_m therefore no longer depends on children's income from 2020 on.
• Add einkommensgrenze_kinder YAML parameter (100,000 EUR/year, 2005–2019, expired with a note on the repeal).
• New policy functions: hat_gesamteinkommen_über_kindeseinkommensgrenze, hat_kind_mit_einkommen_über_einkommensgrenze_als_elternteil_1/_2 (AggType.ANY aggregations over both p_id_elternteil_* slots), and hat_kind_mit_einkommen_über_einkommensgrenze.
• Gate betrag_m on the new check until 2019; separate betrag_m variant from 2020 on.
• Add a documentation page explaining both regimes (docs/tt_explanations/childrens_income_grundsicherung_im_alter.md).
• Policy cases for the exclusion live at 2018-07-01 (pre-repeal).
• Bump ttsim-backend to current main for the any_by_p_id aggregation.

Closes #1148

🤖 Generated with Claude Code

[hmgaudecker]

Just wondering whether the rule is really that simple cutoff? No exceptions, hardship, ... ?

[MImmesberger]

Just wondering whether the rule is really that simple cutoff? No exceptions, hardship, ... ?

Good that you push back. Apparently since 2020 parents can actually receive Grundsicherung im Alter if their children earn more than 100k annually (before 2020 it was indeed a restriction that excluded the parents from receiving the transfer; no hardship rules, at least in the SGB). Today, parents receive Grundsicherung im Alter, but the Sozialamt can reach out to the children to get that money back afterwards if they are more than 100k.

AFAICT the Sozialämter usually do this but since it is not an automatically applied rule, I'd say we don't model if past 2019.

I added an explanation to GETTSIM's documentation and the corresponding functions.

[hmgaudecker]

Could you please fix the test-jax errors before I look into this? Thanks!

[hmgaudecker]

but the Sozialamt can reach out to the children to get that money back afterwards if they are more than 100k. AFAICT the Sozialämter usually do this

In this case, I think we should add an input variable sozialamt_rückgriff_auf_kindereinkommen: bool (or so). We don't want to have a hard policy break when in practice, there often is none.

[hmgaudecker]

Actually, the more important reason: We want to be able to represent the clawback to children's income. The situation is not quite symmetric to take-up because we can always leave out a target and it won't be computed. If we left out the possibility here, we would not be able to represent it.

[MImmesberger]

I'd prefer to postpone a proper treatment. It is complex because

• children are obliged to pay Unterhalt according to the Düsseldorfer Tabelle; in that sense the problem is similar to determining parental Unterhalt for children
• the Sozialhilfeträger's claim can never exceed the minimum of {individuelle Unterhaltsverpflichtung, geleistete Sozialhilfe, individuelle Leistungsfähigkeit}
• Especially in the presence of multiple children its not trivial how the Amt distributes their claim across children. To me it seems like the children below earnings of 100k are relevant too (their liability share is not lost but the Amt takes care of it).

We could ofc provide a target maximaler_unterhaltsrückgriff_des_sozialamts which equals the parental Grundsicherung im Alter. But I'd be afraid of giving users a false sense of security regarding the implementation.

If you agree, I'll post the corresponding issue.

[hmgaudecker]

Agree, but I think I prefer keeping the pre-2020 rule as the default then despite this going against our philosophy. Again, easier to turn that off than to turn Rückgriff on if we don't model it.

[hmgaudecker]

Thanks!