You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Imputed savings put CPS SSI reporters over the resource limit, so reforms 'add' people who already report SSI #1184
On the certified default (populace-us-2024-spm-20260915, policyengine-us 2.2.1), some people who reported SSI in the CPS have imputed savings above SSI's resource limit. The take-up stage keeps their claiming flag on, but the resource test keeps them off SSI under current law. Any reform that raises the limit then "adds" them, although they report already receiving SSI.
Here a CPS SSI reporter is a source person with SSI_VAL > 0 on any row, the take-up stage's anchor. Every such row carries takes_up_ssi_if_eligible = True.
New SSI recipients
All
Who report SSI in the CPS
Share
$10k/$20k limits, aged 18–64
153,211
23,346
15.2%
No limit, aged 18–64
253,026
38,918
15.4%
No limit, aged 65+
693,845
42,726
6.2%
Source: results/admin-checks/new_recipients_by_reporting.csv in PolicyEngine/ssi-asset-limits (merged in #1).
For the $10k/$20k limits, this is a sixth of the adult row that otherwise matches CBPP's 154,000.
Why
scf_wealth draws each household's liquid assets (50/50 SIPP 2023 / SCF 2022) without conditioning on SSI receipt. In the data, current SSI recipients hold almost no countable resources, because they cannot hold more than the limit.
Suggested fix (attributes only)
Use reported SSI receipt (CPS SSI_VAL > 0) as a predictor in the asset imputation. SIPP observes both receipt and assets, so reporters would draw from recipients' observed asset distribution. This conditions on an observed attribute, not on the resource limit, so it stays policy-invariant.
Acceptance
Among CPS SSI reporters with the take-up flag, the weighted share failing current-law meets_ssi_resource_test falls to roughly SIPP's share of recipients above the limit.
SSA age-band reconciliation does not regress.
Related: #584 (asset gradient in claiming), #1178 (Social Security erased by the PUF channel), #880 (married couples overcounted).
Updated numbers on the same certified default (6496cc43…), with PolicyEngine-US fab18b7d6b, January 2026. These count own-reason SSI reporters, excluding the on-behalf-of-child payees in #1313.
Resource-test failures: 454,000 weighted reporters are aged, blind or disabled but fail the resource test (median countable resources $9,087).
Upper bound on fixing assets: zeroing the liquid assets of everyone in a reporter's household, an input change with no recalibration, brings only 157,000 more reporters into payment (2.79M → 2.95M). Most of the others then fail the income test.
Effect of that bound: federal SSI rises $1.93B, which includes non-reporters in those households.
Activity
MaxGhenis commentedon Oct 18, 2026
Updated numbers on the same certified default (
6496cc43…), with PolicyEngine-USfab18b7d6b, January 2026. These count own-reason SSI reporters, excluding the on-behalf-of-child payees in #1313.Evidence: #1312 (
docs/evidence/us-reported-ssi-receipt/, variantreceipt_assets0).