A0212
Title: Harnessing genetic variants for local average treatment effect estimation
Authors: Michela Bia - Luxembourg Institute of Socio-Economic Research (Luxembourg) [presenting]
Abstract: The contribution to the literature on genetic epidemiology and health economics is by estimating local average treatment effects (LATE) using genetic data. Using the Understanding Society dataset, a flexible instrumental variables (IV) framework is applied that leverages genetic variants to estimate heterogeneous effects of arthritis on equivalized income. Results show significant negative effects within some genetic subgroups and positive effects in others, possibly due to reliance on welfare programs among low-income individuals. This heterogeneity highlights the importance of subgroup analysis, which standard 2SLS may overlook. To improve the credibility and precision of the estimates, the testing approach was adopted by a recent study, which identifies valid instruments under the assumption that a relative majority is valid, conditional on compliance. This method is implemented using genetic data. Additionally, the findings are benchmarked with Mendelian randomization (MR), an alternative IV approach that supports the robustness of the results. New insights are offered into the economic effects of chronic conditions like arthritis, with relevant implications for public health and economic policy.
