We study full surplus extraction and implementation in dynamic environments.
We exploit intertemporal correlations of agents' types to construct
within-period ex post incentive compatible mechanisms. First, we formulate
one-shot environments, in which a single agent has a hidden type and
the planner observes a public signal about the agent's type after
a type-contingent allocation is chosen. We propose necessary and sufficient
conditions for full surplus extraction (strong detectability)
and for implementability of the targeted allocation rule (weak
detectability) in this one-shot problem. We decompose the general
dynamic problem into one-shot problems, and obtain sufficient conditions
for surplus extraction and implementation.