Understand. Simulate. Improve. The Economy Lab turns economic claims into experiments. We begin with a practical problem, state what would count as improvement and compare more than one way of organising the system. AI can help generate scenarios, inspect patterns and explore large parameter spaces. It cannot supply missing evidence by assertion. We publish assumptions, data sources, model versions, uncertainties and conditions under which a result changes.
Our working sequence
- Define the problem in terms people experience: cost, time, quality, access and control.
- Describe the existing mechanism and at least one competing explanation.
- Specify the proposed alternative, including who decides and who pays.
- Run ordinary conditions, difficult conditions and attempts to exploit the model.
- Compare the distribution of gains and losses, not only averages.
- Publish the model, result, limitations and the next test.
Experiment status
Proposed: the question and mechanism are being defined.
In development: the model is being built and checked. Open for review: others can inspect and challenge the approach. Results available: findings are published with their conditions and limitations. Revised: a new version changes assumptions, code or interpretation. A result should never display “proved” simply because one simulation returned a favourable number.
Simulation access
Read research and download public material without an account. Try available basic simulations without logging in. Log in to save, compare and share your runs. Advanced AI features may use credits; show the charge before a run starts.