Duomo di MilanoAbstract
EEA-ESEM 2022 proved to be the perfect setting to present the Instrumented Common Confounding paper to an audience, which truly got the intimate link between proximal learning and instrumental variables. The Instrumented Common Confounding approach not only emphasises the methodological similarities between proximal learning and instrumental variables, but combines their identifying assumptions into a new identification method. I want to personally thank both Frank Windmeijer and Whitney Newey for their useful feedback and guidance. Their ideas helped me improve the basics of my approach, including formulation of the identifying assumptions in potential outcomes, as well as the improved graphical depiction of how exogenous instrument variation is recovered by blocking a backdoor path between the instruments and their unobserved confounders using proxies.
