On the falsification of instrumental variable models for heterogeneous treatment effects

Ricardo E. Miranda

In this paper I derive a set of testable implications for econometric models defined by three assumptions: (i) the existence of strictly exogenous discrete instruments, (ii) restri…