ecopol-identification

• SKILL.mdcore · 6,550 bytes

# Identification Strategy (ecopol-identification)

## When to trigger

- A policy effect rests on OLS + controls, or TWFE on staggered policy rollout
- An IV's exclusion restriction is institutional hand-waving, not an argument
- A structural/quantitative policy model is estimated but it is unclear *what in the data* pins each parameter
- The counterfactual policy scenario relies on parameters whose policy-invariance is undefended
- You must convince **two discussants at once** — one who will probe the econometrics, one who needs the design to be legible enough to trust the policy number

## The EP identification bar

EP papers are debated by two named discussants and read by policymakers, so identification must be **both rigorous and legible**. The academic discussant will apply the modern frontier; the policy discussant must be able to follow *why* the estimate is causal without reading the appendix. The discipline: make the **mapping from data/variation to the policy claim** explicit in the main text in plain language, and carry the formal defense in a technical appendix (EP's house split — accessible main text, rigorous appendix). Because results feed a policy recommendation, the magnitude — not just the sign or the stars — must be defended; report standard errors and confidence intervals, never lean on significance asterisks for the headline claim.