Economics
Saved & recent

Natural experiments

A natural experiment uses an external event or rule that creates plausibly exogenous exposure. Its causal value depends on the assignment argument, not simply on a striking before-after change.

Read first
Causal identification
Related
Difference-in-differences, Regression discontinuity
Further reading
World Bank: Research design for impact evaluation
birth-date cutoff Dec 31 ineligible no training offer Jan 1 eligible training offer compare nearby births
Hypothetical policy-generated variation, not researcher assignment. The eligibility event alone is not a causal argument: sorting, coincident rules and other pathways must be investigated. Nearby comparison supports a local claim rather than a population-wide one.

View in the full Economics collection