Shining a light on state
waste and stupidity.
Through data and analysis. nstate audits every public pound against its stated objective, proves what fails, and publishes the evidence for everyone to see.
audit
Every scorecard starts with publicly available data. We cite the source, state the method, and publish the code. If you can't reproduce it, we haven't done the job.
benchmark
We don't just say "this is bad." We build counterfactuals — what the same money could achieve under a different policy, compared against comparable countries and historical evidence.
publish
Every finding is public, versioned, and correctable. Got better data? Open a PR. Disagree with the analysis? Submit a counter-scorecard. The methodology is the contract.
automate
Phase 3: open-source agents that shadow administrative processes — not to replace human judgement, but to show which tasks deterministic software could do cheaper, faster and without political capture.
first wave in progress
We are building the data pipelines before publishing any numbers. Priority areas based on fiscal scale, data availability, and benchmarkability.
Energy levies on bills
Policy costs embedded in household bills vs. general taxation: fiscal impact, distributional incidence, international comparison.
VAT on independent schools
Projected revenue vs. fiscal displacement through state sector admissions, cross-border migration, and teaching staff costs.
Civil service headcount growth
Headcount and real-terms pay vs. service delivery metrics, 2000–present. Productivity per civil servant vs. OECD peers.
No numbers are published until the data pipeline is live and peer-reviewed. That is the point.
nstate shows you exactly how governments waste your money — and builds the open-source tools to fix it. The end goal: a minimum viable state that costs less, does more, and leaves you alone.
start your chapter
The methodology is country-agnostic. Fork the template, adapt your data sources, open a PR. Every country deserves an audit.
join the audit
Data analysts, economists, developers, journalists — if you can write a Python pipeline or check a source, you can contribute.