This paper studies dynamic principal-agent models when the length of a contracting period shrinks to zero. The information structure of all the models converges to the same limit, in which the output, observable by the principal, is a diffussion proces with drift determined by the agent’s unobservable effort (see Sannikov 2008). The novel optimal contracts in the time limit are described by ordinary differential equations. The contracts depend on the distribution of the noise in the approximating sequence, as well
as whether the timing within each period reflects Hidden Action or Hidden Information environment. The paper further explores how those details of the environment affect the optimal contracts.