Open-source tools that came out of our applied economics research. Each one is documented for reuse, ranging from provenance audits of AI-assisted analyses to BigQuery demand-segmentation patterns we keep reaching for.
A five-step sequence for putting an AI coding agent into a real applied economics study, from the first literature scan to the replication package, grounded in published analyses with open code.
The work below tends to start the same way: a research thread runs into the same plumbing problem twice, and we end up extracting the plumbing into something we can reuse. Some of these tools support the analyses we publish here; others scaffold the public-facing infrastructure we share with collaborators.
Everything is on GitHub. Where a tool is small enough to drop into a one-line install, we note the command; for larger projects, the repo README has the setup walk-through.
Looking for the smaller artifacts? Prompt and checklist templates extracted from our methodology articles live at → Browse templates.
Local, deterministic Python framework for qualitative synthesis and econometric parameterization. Features an interactive 1-click Codebook Wizard that converts emergent topic medoids into Boolean YAML rules on your laptop—with zero recurring seat fees, zero cloud API leakage, and cryptographic SHA-256 quote receipts.
Provenance audit for AI-assisted empirical research. Reads Claude Code session logs and surfaces the decision graph, so we can self-audit our own analysis path and give referees a transparent record of what was tried and what was kept.
Code on GitHub →Replication materials for research published on Too Early To Say. Where an article shows a chart or a result, the underlying script, data pull, and notebook live here, organized by topic.
Code on GitHub →Full-stack platform for the California Association of Public Health Economists. 37 interactive methods labs, a Medi-Cal Access Explorer, and a Public Health ROI Calculator. Node and Express on Heroku, Supabase with row-level security on the back end.
Code on GitHub →Geographic demand segmentation for subscription pricing using Census ACS via BigQuery. A walk-through of CTEs, window functions, QUALIFY, and ROLLUP, written to run inside the BigQuery free tier so anyone can fork and follow along.
Code on GitHub →Centralized visual regression testing across the websites we maintain, built on Playwright. Captures baseline screenshots, flags layout drift on each deploy, and lets us share one testing pipeline across several properties.
Code on GitHub →Analysis code linking grocery access and mobility data for the food security research thread. Includes the data-prep scripts behind several of our food-access pieces and the spatial joins we use to characterize mobility deserts.
Code on GitHub →Most of these started as one-off scripts inside a project before becoming reusable enough to publish. If something is helpful, please fork it, file an issue, or open a PR. We treat the documentation gaps in our own tools the same way we treat them in published research: as places to come back to once we understand the use case better.
If we have built something that would help our work and is not yet here, let us know at [email protected] and we will look at extracting it.
New analysis delivered to your inbox. Policy-relevant findings with methodology. No spam, unsubscribe anytime.
Subscribe for Free