Improving calibration of state and national SPM rates

[DTrim99]

Our state and nationwide SPM poverty rates in policyengine-us-data continue to diverge meaningfully from the Census-reported figures. The current discrepancies are documented here. As we expand into higher-visibility deliverables like the Child Poverty Impact Dashboard, these gaps increasingly risk undermining the accuracy and credibility of the analyses we publish.

Why this matters now

External-facing dashboards: The Child Poverty Impact Dashboard and state-level tools (RI CTC calculator, Hawaii, Montana, etc.) put baseline poverty rates directly in front of policymakers, journalists, and advocates. If our baselines don't match Census, every reform impact we model inherits that discrepancy.
Legislative outreach: When we share state-specific analyses with congressional offices and state legislators, the first sanity check stakeholders run is comparing our baseline against published SPM figures. Mismatches force us to caveat and explain rather than focus on the policy question.
Cross-state comparability: Inaccurate state rates affect how reforms rank across states — a core use case for legislators considering peer-state comparisons.
Reform-level estimates: Baseline drift compounds when we model reforms. Even if relative effects are directionally right, levels matter for the headline numbers we publish (e.g., "X children lifted out of poverty").

Adding SPM rates as a calibration target, auditing SPM unit construction, and reviewing state-level benefit imputation accuracy could all help ensure rates are more accurate. Happy to review options and assist in any way I can.