The Trade-Off between Flexibility and Robustness in Instrumental Variables Analysis

In additive instrumental variables models, the robustness to some failure of instrumental validity or additive separability depends on the strength of a priori restrictions on the structural relationship between outcomes and treatments. I provide theoretical analysis of the problem, discuss the implications for empirical practice, and demonstrate with a numerical study calibrated on real-world data. (JEL C26, G51, H24, J13)