A0613
Title: Testing in the functional linear model with Anderson-Rubin-type robustness and without dimension reduction
Authors: Won-Ki Seo - University of Sydney (Australia) [presenting]
Abstract: Statistical tests are developed for the slope parameter in a functional regression model with potential endogeneity, using a functional instrumental variable (IV) without relying on dimension reduction of the functional variables. The proposed tests are not only easy to implement in practice but also remain consistent even under minimal IV identification strength. In fact, they exhibit Anderson-Rubin-type robustness within the functional regression framework. These tests are particularly appealing in practical situations where verifying all the technical conditions required by existing methods is not feasible. Monte Carlo simulations demonstrate that the proposed tests perform well. The methodology is applied to a functional regression model involving residential electricity consumption and temperature distribution in the Republic of Korea.
