This paper considers the problem of the identification of local average treatment effects in a framework where dynamic sequences of programs are attended. Different sets of assumptions are settled depending on the dynamic form of the exclusion restriction. We clarify the concept of compliers in this multi-period framework. The introduction of a second period implies a loss of identification for some effects even if we do not have any endogeneity problem. Nevertheless all the effects for the population who complies each period are identified even in presence of the endogeneity problem coming from a participation decision based on past successes.