Instrumental variable only works if the only connection between the instrumental variable and the outcomes is through the treatment. For an instrumental variable to be effective it must be 1. Relevant (i.e., it must be directly related to treatment) and 2. Exogenous (i.e., it must be related to the outcome only through the treatment). Failure of either condition is a problem.Exogeneity is difficult to prove and requires theoretical justification. Can you think of an example where an instrumental variable might be appropriate for estimating causal effect?