In existing papers on dynamic incentive contracts, the dynamic structure of the principal-agent relationship arises exclusively from the ability of the principal to learn about the hidden information over time. In this paper we deal with a different source of dynamics, which is considered standard in all areas of economics other than the information literature: we study situations where current opportunities depend on past and current actions, notwithstanding any information conveyed by the actions. Standard examples include investment, Learning by doing, and R&D. In order to focus on this neglected source of dynamics, we restrict our attention to situations involving asymmetric information in each period, but without any intertemporal informational correlation, so that no dynamic effect arises from informational asymmetries directly. This makes comparisons with static results both easier and more interesting.