The thesis is a timing claim. Last-resort support falls when the recipient saves, earns, or simplifies household risk, and it falls before the new stock can survive without that support.
Section I then splits the object. The rules discourage those moves at the margin. The magnitude of each response requires evidence; the rules do not show that every recipient adopts them. "Poverty mindset" is recoded as sometimes a rational response.
Those are two claims. A withdrawal schedule is a property of Kela's calculation. A trap is a household that attempted runway and lost support before the new stock was stable. The title sells the second. The sourced rules are the first.
The page is not claiming universal adoption. A calculation that cuts support at the first euro of saving or formal earnings is already a trap in the design: building runway reduces support first. You do not need every recipient to take the bait for the structure to have that property.
The behavioral recode is labelled "sometimes." The load is the timing in the calculation, not a measured headcount of trapped households.
Then freeze the object. If the trap is the calculation, the opening about pathologized behaviors is a different claim. A 100% withdrawal schedule can sit on the books with no one trapped: people do not attempt runway, or another benefit replaces the cut, or the household never had assets to liquidate.
"Capability trap" names the second object. The page's own hedge — rules do not show adoption — is the admission that the first object has not been shown to produce the second.
Split the specimen. Conversion of the trap: a household that saved or took formal work and lost last-resort support before the new stock survived without it. Theatre: the February 2026 adult earnings disregard removal, scored as a rate, with no such household.
If only the rate is on the page, recode the title as a last-resort withdrawal schedule. Keep the behavioral recode only if a movement exists.