Repository navigation
additional_residential_property_purchased treats a household's whole other-property stock as an additional purchase #2238
Description
What happens
additional_residential_property_purchased returns other_residential_property_value × property_purchased (variables/household/consumption/additional_residential_property_purchased.py). property_purchased flags any residential purchase in the year, including a main home, so a household that bought any home has its entire stock of other residential property charged as an additional-dwelling purchase. The variable's own documentation says "Only include the value of a single purchase."
Reproducer
Run at 2.89.2. The formula is unchanged on main as of 9 October 2026.
One adult, £80,000 earnings, London, 2025, main_residence_value £600,000, other_residential_property_value £2,000,000:
| property_purchased | additional_residential_property_purchased | stamp_duty_land_tax |
|---|---|---|
| False | £0 | £0 |
| True | £2,000,000 | £233,750 |
Impact
Found by the fiscal-event replay in PolicyEngine/policyengine-scorecard#157. Autumn Budget 2024's 2pp rise in the higher rates for additional dwellings costs £2.0bn to £2.3bn a year in PolicyEngine, against the OBR's £0.2bn to £0.5bn. The flagged base is about £94.7bn, and 2pp of that is about £1.9bn.
Fix
Separate main-home purchases from additional-dwelling purchases, and impute one additional purchase value (or a purchase probability on the other-property stock) rather than the whole stock. Calibrate the result to HMRC's SDLT statistics on higher-rate transactions.
Activity
MaxGhenis commentedon Oct 10, 2026
Fix in #2241: a main-residence purchase no longer charges the household's other residential or non-residential stock. The additional-dwelling and non-residential purchase prices are inputs, nil unless set directly or imputed by a dataset. Household higher-rates SDLT is therefore nil on data without imputed purchases (an inert leg, not an inferred base) until policyengine-uk-data imputes them; the PR has the HMRC comparison.