This paper presents an original econometric model for estimating a dose-response function through a regression approach when: (i) treatment is continuous, (ii) individuals may react heterogeneously to observable confounders, and (iii) selection-into-treatment may be potentially endogenous. After...