Three essays on applied econometric methods

Abstract

This dissertation endeavors to scrutinize biases arising from infringements of fundamental assumptions in econometrics and establish sturdy frameworks to tackle these issues. To this end, the first essay is concerned with deriving nonparametric bounds on the average treatment effect using an imperfect instrument. The framework thus derived is immune to the violation of the perfect exogeneity assumption of the instrumental variable. The second essay strives to develop a generalized difference-in-differences framework by deploying a notion of selection bias that can withstand violations of the parallel trends assumption. In the third essay, we examine the performance of three Tobit-based approaches to estimate demand systems with zero consumption. Monte-Carlo simulations corroborate the misspecification bias and the over-rejection problem of the Tobit-based approaches. The findings of this dissertation offer guidance on the applications and interpretation of results and illuminate future research avenues in econometrics literature. The frameworks developed in this dissertation, being robust to the key assumptions that are likely to be violated, can substantially augment the field of applied economics.