Extrapolating Away from the Cutoff in Regression Discontinuity Designs

Yiwei Sun

Canonical RD designs yield credible local estimates of the treatment effect at the cutoff under mild continuity assumptions, but they fail to identify treatment effects away from t…