From 76f20e10785a645ab7587f4435d6ff28e6e44220 Mon Sep 17 00:00:00 2001 From: PavelMakarchuk Date: Tue, 17 Feb 2026 15:20:54 -0500 Subject: [PATCH 1/4] Add blog post: TANF modeling for all 50 states and DC Announces nationwide TANF coverage, AI-assisted workflow, and availability through web app, Python package, and API. Co-Authored-By: Claude Opus 4.6 --- .../assets/posts/tanf-50-state-coverage.svg | 121 ++++++++++++++++++ .../posts/articles/tanf-50-state-coverage.md | 36 ++++++ app/src/data/posts/posts.json | 9 ++ 3 files changed, 166 insertions(+) create mode 100644 app/public/assets/posts/tanf-50-state-coverage.svg create mode 100644 app/src/data/posts/articles/tanf-50-state-coverage.md diff --git a/app/public/assets/posts/tanf-50-state-coverage.svg b/app/public/assets/posts/tanf-50-state-coverage.svg new file mode 100644 index 000000000..d3f1c5e3b --- /dev/null +++ b/app/public/assets/posts/tanf-50-state-coverage.svg @@ -0,0 +1,121 @@ + + + + + + AK + + ME + + + + VT + + NH + + + + WA + + ID + + MT + + ND + + MN + + WI + + MI + + NY + + MA + + CT + + RI + + + + OR + + NV + + WY + + SD + + IA + + IL + + IN + + OH + + PA + + NJ + + + + CA + + UT + + CO + + NE + + KS + + MO + + KY + + WV + + VA + + MD + + DE + + + + AZ + + NM + + OK + + AR + + TN + + NC + + SC + + DC + + + + TX + + LA + + MS + + AL + + GA + + + + HI + + FL + diff --git a/app/src/data/posts/articles/tanf-50-state-coverage.md b/app/src/data/posts/articles/tanf-50-state-coverage.md new file mode 100644 index 000000000..71d31e5e3 --- /dev/null +++ b/app/src/data/posts/articles/tanf-50-state-coverage.md @@ -0,0 +1,36 @@ +PolicyEngine now models Temporary Assistance for Needy Families (TANF) eligibility and benefit calculations for all 50 states and Washington, D.C. This completes our nationwide coverage of the federal block grant program that provides cash assistance to low-income families with children, with each state's specific rules encoded in our open-source rules engine. + +## How we built it + +Each state administers TANF under its own rules, setting benefit levels, income disregards, asset limits, and eligibility criteria independently. Encoding 51 distinct policy configurations required a systematic approach. + +We used the [multi-agent AI workflow](multi-agent-workflows-policy-research) we described in an earlier post. Claude Code agents handled the repetitive components of the implementation: reading state policy manuals, structuring parameters, drafting variable logic, and generating test cases. Human reviewers validated each state's rules against official documentation and resolved cases where policy language was ambiguous or where program interactions required judgment. + +This workflow allowed us to maintain consistent code quality and documentation standards across all 51 implementations while reducing the time required for each additional state. + +## What TANF covers + +TANF is a federal block grant that gives states broad flexibility in designing cash assistance programs for families. Key policy dimensions that vary by state include: + +- **Maximum benefit amounts** by household size +- **Earned income disregards** that determine how much employment income is excluded from benefit calculations +- **Asset limits** for eligibility +- **Income thresholds** for initial eligibility and ongoing receipt +- **Child age requirements** and school enrollment provisions + +PolicyEngine's model captures these variations, enabling direct comparisons of how TANF operates across states and how proposed reforms would affect families differently depending on where they live. + +## How to use it + +The TANF models are available through multiple channels, each supporting both household-level calculations and population-wide microsimulation: + +- **[Web app](https://policyengine.org).** Enter household details to calculate TANF eligibility and benefit amounts in any state, or use the population impact view to estimate the fiscal cost, poverty impact, and distributional effects of TANF reforms. +- **[Python package](https://github.com/PolicyEngine/policyengine-us).** Run household or microsimulation analyses programmatically, with full access to TANF variables and parameters for custom research. +- **[API](https://policyengine.org/api).** Integrate TANF calculations into external applications for benefits screening, policy design, or other tools. + + +## Looking ahead + +With TANF coverage complete, PolicyEngine's US model now includes state-level rules for income taxes, SNAP, Medicaid, CHIP, ACA subsidies, and TANF across all 50 states and D.C. A full list of covered programs is available on our [model documentation page](https://policyengine.org/us/model). We are applying the same approach to encode the Child Care and Development Fund (CCDF), SSI State Supplements, and the Low Income Home Energy Assistance Program (LIHEAP) across all states. We also continue to validate our calculations against other microsimulation models through our partnerships with [NBER](policyengine-nber-mou-taxsim) and the [Federal Reserve Bank of Atlanta](policyengine-atlanta-fed-mou-prd). + +For questions about using the TANF models or integrating them into your work, reach out to us directly or visit our [GitHub repository](https://github.com/PolicyEngine/policyengine-us). diff --git a/app/src/data/posts/posts.json b/app/src/data/posts/posts.json index c04c8414a..f0e040e85 100644 --- a/app/src/data/posts/posts.json +++ b/app/src/data/posts/posts.json @@ -1,4 +1,13 @@ [ + { + "title": "PolicyEngine completes TANF modeling for all 50 states and DC", + "description": "AI-assisted workflows enabled nationwide coverage of Temporary Assistance for Needy Families, now available through the household API and microsimulation.", + "date": "2026-02-17", + "tags": ["us", "technical", "ai", "featured"], + "authors": ["pavel-makarchuk"], + "filename": "tanf-50-state-coverage.md", + "image": "tanf-50-state-coverage.svg" + }, { "title": "Analysis of South Carolina H.3492: Making the state EITC partially refundable", "description": "PolicyEngine projects the bill would lower state revenues by $402 million while reducing child poverty by 5.4% in 2026.", From a640a0cea94f2d0c463fd7f208ecdb507b2d80a3 Mon Sep 17 00:00:00 2001 From: PavelMakarchuk Date: Tue, 17 Feb 2026 15:26:21 -0500 Subject: [PATCH 2/4] Update blog post description to list three access channels Co-Authored-By: Claude Opus 4.6 --- app/src/data/posts/posts.json | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/app/src/data/posts/posts.json b/app/src/data/posts/posts.json index f0e040e85..86d47a2c0 100644 --- a/app/src/data/posts/posts.json +++ b/app/src/data/posts/posts.json @@ -1,7 +1,7 @@ [ { "title": "PolicyEngine completes TANF modeling for all 50 states and DC", - "description": "AI-assisted workflows enabled nationwide coverage of Temporary Assistance for Needy Families, now available through the household API and microsimulation.", + "description": "AI-assisted workflows enabled nationwide coverage of Temporary Assistance for Needy Families, now available through the web app, Python package, and API.", "date": "2026-02-17", "tags": ["us", "technical", "ai", "featured"], "authors": ["pavel-makarchuk"], From 6aae2f3aa7994b08f5596c4059ddc28200b1c550 Mon Sep 17 00:00:00 2001 From: PavelMakarchuk Date: Tue, 17 Feb 2026 17:15:42 -0500 Subject: [PATCH 3/4] Address review comments: link text and add significantly Co-Authored-By: Claude Opus 4.6 --- app/src/data/posts/articles/tanf-50-state-coverage.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/app/src/data/posts/articles/tanf-50-state-coverage.md b/app/src/data/posts/articles/tanf-50-state-coverage.md index 71d31e5e3..0cd9c417f 100644 --- a/app/src/data/posts/articles/tanf-50-state-coverage.md +++ b/app/src/data/posts/articles/tanf-50-state-coverage.md @@ -4,9 +4,9 @@ PolicyEngine now models Temporary Assistance for Needy Families (TANF) eligibili Each state administers TANF under its own rules, setting benefit levels, income disregards, asset limits, and eligibility criteria independently. Encoding 51 distinct policy configurations required a systematic approach. -We used the [multi-agent AI workflow](multi-agent-workflows-policy-research) we described in an earlier post. Claude Code agents handled the repetitive components of the implementation: reading state policy manuals, structuring parameters, drafting variable logic, and generating test cases. Human reviewers validated each state's rules against official documentation and resolved cases where policy language was ambiguous or where program interactions required judgment. +We used the multi-agent AI workflow we described in [this article](multi-agent-workflows-policy-research). Claude Code agents handled the repetitive components of the implementation: reading state policy manuals, structuring parameters, drafting variable logic, and generating test cases. Human reviewers validated each state's rules against official documentation and resolved cases where policy language was ambiguous or where program interactions required judgment. -This workflow allowed us to maintain consistent code quality and documentation standards across all 51 implementations while reducing the time required for each additional state. +This workflow allowed us to maintain consistent code quality and documentation standards across all 51 implementations while significantly reducing the time required for each additional state. ## What TANF covers From dedd7d08c8db26bde7d79ed56573b7bddfa4ddfd Mon Sep 17 00:00:00 2001 From: PavelMakarchuk Date: Tue, 17 Feb 2026 17:45:27 -0500 Subject: [PATCH 4/4] Add ziming-hua as co-author on TANF blog post Co-Authored-By: Claude Opus 4.6 --- app/src/data/posts/posts.json | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/app/src/data/posts/posts.json b/app/src/data/posts/posts.json index 86d47a2c0..179744a3d 100644 --- a/app/src/data/posts/posts.json +++ b/app/src/data/posts/posts.json @@ -4,7 +4,7 @@ "description": "AI-assisted workflows enabled nationwide coverage of Temporary Assistance for Needy Families, now available through the web app, Python package, and API.", "date": "2026-02-17", "tags": ["us", "technical", "ai", "featured"], - "authors": ["pavel-makarchuk"], + "authors": ["pavel-makarchuk", "ziming-hua"], "filename": "tanf-50-state-coverage.md", "image": "tanf-50-state-coverage.svg" },