Extending Regression Discontinuity Models Beyond the Jump Point

This paper proposes a new estimation method for regression discontinuity models, allowing for estimation of a treatment effect beyond the jump point (with additional assumptions). The proposed procedure consistently estimates the treatment effect function, as well as the average outcome in the absence of treatment. The treatment effect estimator is root-N consistent. We apply the method to an important question in health economics–what is the effect of having Medicare insurance on admissions and health care costs after age 65? Our preferred models shows an increase in both admissions and costs after age 65 due to Medicare.