Clarify simultaneous dollar-and-count calibration vs TRIM caseload-only#33
Open
MaxGhenis wants to merge 2 commits into
Open
Clarify simultaneous dollar-and-count calibration vs TRIM caseload-only#33MaxGhenis wants to merge 2 commits into
MaxGhenis wants to merge 2 commits into
Conversation
PE-US's L0 reweighting and TRIM3's caseload-driven calibration sound
similar at the surface ("aligned to administrative totals") but differ
on what's actually solved for. The previous YAML wording underplayed
the difference. Tightened three imputation rows:
PE-US SNAP: description now states that L0 simultaneously matches
state and national SNAP outlay dollars AND state-level SNAP
recipient-household counts in a single constrained optimization
pass — total dollars and caseloads are both pinned to admin
targets, not just one or the other. Calibration-targets list adds
the previously-implicit recipient-household-count target (it's in
policyengine-us-data/calibration/target_config.yaml but wasn't
surfaced here).
TRIM3 SNAP: clarifies that total SNAP benefit outlay dollars are an
emergent property of (participation × rule-computed benefit), not
an explicit calibration target. The benefit-band composition
target approximates the dollar total via the participant
distribution, but doesn't solve for dollars directly. Cites
Wheaton & Tran (Urban) on SNAP anti-poverty effects.
TRIM3 TANF: same structural note — TANF outlay dollars not an
explicit constraint; PE-US's L0 has them as a state and national
target alongside recipient-unit counts.
Other rows (TRIM3 SSI explicitly calibrates both caseload AND
benefits per their own docs, which our existing row already
captures; PE-US TANF, SSI, Medicaid rows already list both dollar
and count targets) are unchanged.
61/61 tests pass; lint clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
…ormula Per-program calibration framing needs one more distinction. PE-US's SNAP / TANF / SSI rows have rule-computed per-unit benefits (benefit = f(income, household size, deductions, etc.)), so calibrating BOTH caseload counts AND dollar outlays is genuinely informative — the two targets constrain different dimensions of the imputation. Medicaid is different. PE-US has no per-individual benefit formula in the model; per-capita spending is assigned as a CMS-derived constant. So "Medicaid spending" effectively falls out as (enrolled people × per-capita spend), and adding a separate L0 dollar target provides little information beyond the enrollment-count target. Updated the PE-US Medicaid imputation row to call this out explicitly, and re-labelled the calibration targets to mark enrollment counts as primary and national spending as derivative. 61/61 tests, lint clean. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Tightens three imputation-row descriptions to make the methodological contrast on calibration targets explicit, in response to a question about how TRIM aligns benefit outlays to admin:
PE-US SNAP
calibrationTargetslist also gains the recipient-household-count entry that was already inpolicyengine-us-data/calibration/target_config.yamlbut wasn't surfaced here.Test plan
policyengine-us-data/calibration/target_config.yaml🤖 Generated with Claude Code