- Three households, five outputs (four dollar amounts plus one
+ Three households, four outputs (three dollar amounts plus one
eligibility flag), two models. Every step on this page uses
the same numbers.
@@ -656,9 +656,9 @@
- Average the five row scores into a household score, then
+ Average the four row scores into a household score, then
average the household scores into a model score. Each output
- gets a weight of 1/5 in every household.
+ gets a weight of 1/4 in every household.
@@ -695,7 +695,7 @@
4Per-household impact weights
Now an output gets weight in proportion to its dollar size
for that household. A $5,000 income tax line carries more
- weight than a $500 ACA premium tax credit.
+ weight than a $500 benefit line.
@@ -877,9 +877,8 @@
Worked examples on realistic households
The "share of net income that flowed through this program"
reading is intuitive. Medicaid eligibility carries a
meaningful weight because it shows up with sizeable paired
- values in two of the four households; ACA PTC carries
- smaller weight because only one household uses it. None
- of these numbers required a tunable parameter.
+ values in two of the four households. None of these numbers
+ required a tunable parameter.
@@ -1309,7 +1308,6 @@
12Side by side
{ id: "income_tax", label: "Income tax", kind: "amount" },
{ id: "payroll_tax", label: "Payroll tax", kind: "amount" },
{ id: "snap", label: "SNAP", kind: "amount" },
- { id: "aca_ptc", label: "ACA PTC", kind: "amount" },
{
id: "medicaid_eligible",
label: "Medicaid eligibility",
@@ -1329,7 +1327,6 @@
12Side by side
income_tax: -4000,
payroll_tax: -5000,
snap: 0,
- aca_ptc: 0,
medicaid_eligible: false,
medicaid_value: 0,
},
@@ -1342,7 +1339,6 @@
12Side by side
income_tax: -1000,
payroll_tax: -1500,
snap: 6000,
- aca_ptc: 8000,
medicaid_eligible: true,
medicaid_value: 6000,
},
@@ -1355,7 +1351,6 @@
12Side by side
income_tax: -2000,
payroll_tax: 0,
snap: 2400,
- aca_ptc: 0,
medicaid_eligible: true,
medicaid_value: 9000,
},
@@ -1374,7 +1369,6 @@
12Side by side
income_tax: 3000,
payroll_tax: -1530,
snap: 5000,
- aca_ptc: 3000,
medicaid_eligible: true,
medicaid_value: 8000,
},
@@ -1387,7 +1381,6 @@
12Side by side
income_tax: -4500,
payroll_tax: -4590,
snap: 0,
- aca_ptc: 0,
medicaid_eligible: false,
medicaid_value: 0,
},
@@ -1400,7 +1393,6 @@
12Side by side
income_tax: 0,
payroll_tax: 0,
snap: 2400,
- aca_ptc: 0,
medicaid_eligible: true,
medicaid_value: 9000,
},
@@ -1413,7 +1405,6 @@
12Side by side
income_tax: -40000,
payroll_tax: -9900,
snap: 0,
- aca_ptc: 0,
medicaid_eligible: false,
medicaid_value: 0,
},
@@ -1425,18 +1416,18 @@
12Side by side
id: "tax",
label: "Tax-focused",
predictions: {
- H1: { income_tax: -4000, payroll_tax: -5000, snap: 0, aca_ptc: 0, medicaid_eligible: false },
- H2: { income_tax: -1000, payroll_tax: -1500, snap: 0, aca_ptc: 0, medicaid_eligible: false },
- H3: { income_tax: -2000, payroll_tax: 0, snap: 0, aca_ptc: 0, medicaid_eligible: false },
+ H1: { income_tax: -4000, payroll_tax: -5000, snap: 0, medicaid_eligible: false },
+ H2: { income_tax: -1000, payroll_tax: -1500, snap: 0, medicaid_eligible: false },
+ H3: { income_tax: -2000, payroll_tax: 0, snap: 0, medicaid_eligible: false },
},
},
{
id: "benefit",
label: "Benefit-focused",
predictions: {
- H1: { income_tax: -2000, payroll_tax: -4000, snap: 0, aca_ptc: 0, medicaid_eligible: false },
- H2: { income_tax: 0, payroll_tax: -1000, snap: 6000, aca_ptc: 8000, medicaid_eligible: true },
- H3: { income_tax: -1000, payroll_tax: 0, snap: 2400, aca_ptc: 0, medicaid_eligible: true },
+ H1: { income_tax: -2000, payroll_tax: -4000, snap: 0, medicaid_eligible: false },
+ H2: { income_tax: 0, payroll_tax: -1000, snap: 6000, medicaid_eligible: true },
+ H3: { income_tax: -1000, payroll_tax: 0, snap: 2400, medicaid_eligible: true },
},
},
];
diff --git a/app/public/paper/web/index.html b/app/public/paper/web/index.html
index f9d3528..a3f0adb 100644
--- a/app/public/paper/web/index.html
+++ b/app/public/paper/web/index.html
@@ -362,7 +362,7 @@
Frozen snapshot an
| 30 |
Output groups |
-19 US and 7 UK |
+18 US and 7 UK |
| 31 |
@@ -539,7 +539,7 @@ Frozen snapshot an
Data and scenario construction
United States
The US benchmark is built from Enhanced Current Population Survey (CPS)-derived households using PolicyEngine US. The sampled households are filtered to keep a single-tax-unit, single-family, single-Supplemental Poverty Measure (SPM)-unit structure with at least one adult and a supported filing status. The 2024 Enhanced CPS source contains 41,314 households; 30,173 (73.0%) pass the filter and form the eligible draw. The 27.0% excluded by the filter include multi-tax-unit households (e.g., adult roommates), multi-family households, multi-SPM-unit households, and households whose head reports a filing status outside the supported set. These excluded compositions are exactly the kind of cases where federal/state credit allocations and benefit-unit rules become hardest, so the eligible draw is a tractable subset rather than the full distribution of US households. Prompts include nonzero promptable raw inputs across relevant entities rather than a hand-curated summary, so the models see many of the same facts the simulator receives. Filing status is not stated in the prompt; the reference computation infers it from tax-unit role flags. Models therefore see the same household facts that drive the reference filing-status assignment, but they do not receive that assignment as a label.
-The current US release evaluates 19 output groups spanning federal income tax, refundable credits, payroll and self-employment tax, state and local income tax, Supplemental Nutrition Assistance Program (SNAP), Supplemental Security Income (SSI), Temporary Assistance for Needy Families (TANF), Affordable Care Act (ACA) premium tax credits, school-meal eligibility, and person-level coverage eligibility for the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), Medicaid, the Children’s Health Insurance Program (CHIP), Medicare, Head Start, and Early Head Start.
+The current US release evaluates 18 output groups spanning federal income tax, refundable credits, payroll and self-employment tax, state and local income tax, Supplemental Nutrition Assistance Program (SNAP), Supplemental Security Income (SSI), Temporary Assistance for Needy Families (TANF), school-meal eligibility, and person-level coverage eligibility for the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), Medicaid, the Children’s Health Insurance Program (CHIP), Medicare, Head Start, and Early Head Start.
The output scope is intentionally narrower than the full PolicyEngine model. Table 3 summarizes the inclusion rule. The benchmark asks for WIC eligibility rather than a WIC dollar amount; WIC dollar values are used only as impact-weight proxies for coverage flags, not as requested model outputs.
@@ -583,18 +583,13 @@
United States
Excluded |
Intermediate tax bases, payroll subcomponents, and outputs that mainly require unavailable history, restricted local market data, restricted program-administration data, or take-up assignment rather than rule calculation. |
-
-| 2 |
-ACA Premium Tax Credit |
-Retained as a deliberate health-support output; when local benchmark premiums are not listed, the model must estimate them from the household facts. |
-
-| 3 |
+2 |
Binary coverage outputs |
Requested as 0/1 eligibility flags and scored as classification tasks; their dollar values are used only as impact-weight proxies, not as requested model outputs. |
-| 4 |
+3 |
WIC |
The benchmark asks for person-level WIC eligibility. It does not ask models to estimate a WIC dollar amount. |
@@ -1999,7 +1994,7 @@ Appendix A: Structu
| 4 |
Final parse coverage |
-The repaired manuscript snapshot has zero missing parsed numeric values and zero missing explanations across all 34,656 model-output rows. |
+The repaired manuscript snapshot has zero missing parsed numeric values and zero missing explanations across all 33,456 model-output rows. |
| 5 |
@@ -2690,4 +2685,4 @@ Competing interests
-