This paper proposes a new strategy to identify causal effects. Instead of finding a conventional instrumental variable correlated with the treatment but not with the confounding effects, we propose an approach which employs an instrument correlated with the confounders, but which itself is not...