407RESEARCH · MIGRATION LAB
Migration is a referendum on progress.
Explore who moves where across 41 OECD countries since 1995 — then paste a policy, an op-ed or a bill and see how it could change those flows, with every assumption on the table.
WHAT WOULD THIS POLICY DO?
Describe a policy in your own words.
Paste an article, a manifesto or a single sentence. The lab finds the levers in it, then projects them against the OECD record.
Projected annual inflows to the selected destinations
Levers found in the text
Untick a lever or change its strength — the projection updates.
| Use | Lever | Direction | Strength | Years | Effect at full strength |
|---|
By destination, year 10
| Destination | Recent inflow | With policy | Change |
|---|
Not captured by the model
Caveats
THE RECORD
Three decades of movement.
OECD International Migration Database, 1995–2023. Only flows above 1,000 people a year are included.
HOW IT WORKS
Two steps, kept apart.
1. Reading the policy
Claude reads the text and lists the levers in it — sanitation, healthcare, corruption, admission rules and so on — with a direction, a strength from 1 to 5, the years to full effect, and the passage it relied on. Goals without an instrument are marked aspirational and count half. Claude produces no numbers about migration. If it is unavailable, a keyword reader does the same job more crudely, and says so.
2. Projecting the effect
The baseline is the latest three-year average of OECD inflows from the origin to each destination, held flat. Each lever then scales that baseline, phased in linearly over its years to full effect.
Admission rules at a destination are estimated from the data. Across destination countries and country-years, a 0.01 rise in the Immigration Policy Score goes with inflows , after removing each country’s own level and each year’s shocks (fixed effects, standard errors clustered by country: β = , s.e. ). A strength-1 lever moves the score by 0.01. This is an association over time within countries, not proof of cause.
Conditions in an origin country cannot be estimated from this dataset, which has no data on sanitation, health or governance at home. These effects use the coefficients below: the percentage change in emigration per point of strength, at full effect. They are illustrative assumptions chosen to be modest, not estimates, and the low/high setting halves or adds half to all of them.
| Lever | % emigration change per point | Why |
|---|
What this is not
A forecast. Real migration also answers to wages abroad, visas, wars, exchange rates and family networks. Use the lab to compare proposals and to see which levers a text actually leans on — not to predict a number.