Summary
Close the eCPS-populated layers the certified populace-us default leaves all-zero (the missing-layer side of the parity drift in PolicyEngine/microcosm-benchmarks#1). These are pure inputs eCPS provides that populace does not impute — distinct from the formula-owned masking in #24.
Pure-input gaps to impute (PUF-derived; no engine formula — must be stored)
eCPS populates these; certified populace = 0% (verified by simulation-level parity vs current eCPS):
| variable |
eCPS populates |
amt_foreign_tax_credit |
9.3% |
general_business_credit |
5.4% |
excess_withheld_payroll_tax |
4.6% |
prior_year_minimum_tax_credit |
4.0% |
early_withdrawal_penalty |
3.3% |
other_credits |
2.8% |
They feed net federal tax, so the certified default computes lower-magnitude tax than current eCPS for the affected 3–9% of records. Source them from the PUF donor/impute stage.
Formula-owned, but yields 0 because upstream is missing — fix the UPSTREAM, do not store the output
Reported / comparison aggregates — decide carry-or-document
spm_unit_net_income_reported (99.4%), spm_unit_total_income_reported (97.7%), ssi_reported (1.5%) — not formula-consumed; used in reported-vs-modeled analysis. Decide explicitly whether to carry them (they're survey-reported observations, legitimately storable) or document their absence as a named exemption.
Related: PolicyEngine/microcosm-benchmarks#1 (parity drift), #24 (formula-owned), PolicyEngine/microcosm-benchmarks#2 (blocking).
Summary
Close the eCPS-populated layers the certified
populace-usdefault leaves all-zero (the missing-layer side of the parity drift in PolicyEngine/microcosm-benchmarks#1). These are pure inputs eCPS provides that populace does not impute — distinct from the formula-owned masking in #24.Pure-input gaps to impute (PUF-derived; no engine formula — must be stored)
eCPS populates these; certified populace = 0% (verified by simulation-level parity vs current eCPS):
amt_foreign_tax_creditgeneral_business_creditexcess_withheld_payroll_taxprior_year_minimum_tax_creditearly_withdrawal_penaltyother_creditsThey feed net federal tax, so the certified default computes lower-magnitude tax than current eCPS for the affected 3–9% of records. Source them from the PUF donor/impute stage.
Formula-owned, but yields 0 because upstream is missing — fix the UPSTREAM, do not store the output
self_employed_health_insurance_ald(7.4% in eCPS): registered formula exists but its upstream self-employed health-premium inputs are absent in populace, so it computes 0. Populate the upstream input(s); do not persist the ALD itself (that would violate the no-formula rule, see Certified populace-us persists 18 formula-owned variables eCPS doesn't (formula_owned_export gate never wired into the build) #24 / the precedence issue).Reported / comparison aggregates — decide carry-or-document
spm_unit_net_income_reported(99.4%),spm_unit_total_income_reported(97.7%),ssi_reported(1.5%) — not formula-consumed; used in reported-vs-modeled analysis. Decide explicitly whether to carry them (they're survey-reported observations, legitimately storable) or document their absence as a named exemption.Related: PolicyEngine/microcosm-benchmarks#1 (parity drift), #24 (formula-owned), PolicyEngine/microcosm-benchmarks#2 (blocking).