Summary
The dataset imputes total auto-loan interest but not the qualifying subset the OBBBA deduction keys on, so the auto-loan-interest provision is structurally zero in reform validation — unvalidatable, and any analysis touching the provision silently shows no effect.
Evidence
On the released populace_us_2024.h5 (…f0af251…20260620), household table:
| variable |
nonzero households |
weighted total |
auto_loan_balance |
18,508 / 75,112 |
$1,302.7B |
auto_loan_interest |
18,507 / 75,112 |
$68.7B |
But PE-US's auto_loan_interest_deduction is computed from qualified_passenger_vehicle_loan_interest — interest on loans meeting the OBBBA vehicle/loan requirements (new vehicles, US final assembly, 2025–2028; IRS Pub 6126). That variable is a pure input with no formula, and populace does not impute it, so it defaults to $0 for every household → the deduction is $0 by construction.
Reform-validation result: populace $0.0 vs JCT −$5.4B (FY2026) / −$8.1B (FY2027) (JCX-35-25 Ch.2 line 3, effective interest incurred after 12/31/24).
Suggested fix
Impute qualified_passenger_vehicle_loan_interest as a share of the existing auto_loan_interest — e.g. apportion by the share of loans that are for qualifying new, US-assembled vehicles (new-vehicle financing shares are published; US-final-assembly share can be approximated from sales data), or calibrate the qualifying share so the simulated deduction is benchmarkable.
Found via the reform-validation page in PolicyEngine/calibration-diagnostics; sibling of #225 (SSN/citizenship imputation gap).
Summary
The dataset imputes total auto-loan interest but not the qualifying subset the OBBBA deduction keys on, so the auto-loan-interest provision is structurally zero in reform validation — unvalidatable, and any analysis touching the provision silently shows no effect.
Evidence
On the released
populace_us_2024.h5(…f0af251…20260620), household table:auto_loan_balanceauto_loan_interestBut PE-US's
auto_loan_interest_deductionis computed fromqualified_passenger_vehicle_loan_interest— interest on loans meeting the OBBBA vehicle/loan requirements (new vehicles, US final assembly, 2025–2028; IRS Pub 6126). That variable is a pure input with no formula, and populace does not impute it, so it defaults to $0 for every household → the deduction is $0 by construction.Reform-validation result: populace $0.0 vs JCT −$5.4B (FY2026) / −$8.1B (FY2027) (JCX-35-25 Ch.2 line 3, effective interest incurred after 12/31/24).
Suggested fix
Impute
qualified_passenger_vehicle_loan_interestas a share of the existingauto_loan_interest— e.g. apportion by the share of loans that are for qualifying new, US-assembled vehicles (new-vehicle financing shares are published; US-final-assembly share can be approximated from sales data), or calibrate the qualifying share so the simulated deduction is benchmarkable.Found via the reform-validation page in PolicyEngine/calibration-diagnostics; sibling of #225 (SSN/citizenship imputation gap).