Book excerpt: Three models of a privately informed contract designer (a principal) are examined. In the first, I study how much private information the principal wants to acquire before offering a contract to an agent. Despite allowing her to acquire all information for free, I prove in a general environment that there is a nontrivial set of parameters for which it is strictly suboptimal for the principal to be completely informed, regardless of the continuation equilibrium following any information acquisition choice. This result holds even when the principal is able to employ the most general mechanisms available and, in particular, when she can choose her most favourable full-information continuation equilibria. Further, in a specialized environment I characterize the principal's optimal information choice. The second is a two-state principal-agent model with moral hazard in which the principal knows the state but the agent does not. This model is relevant to situations where an employer has private information about the productivity of a worker in a particular task while the worker has private information about the effort she exerts on the job. Much of the literature on this subject restricts the employer to offer contracts that leave her no discretion once a contract is accepted, while more general contracts may allow the employer to exercise discretion after acceptance; such contracts are called menu-contracts. I show when the employer can obtain strictly higher expected payoffs by offering menu-contracts than by offering the restricted contracts used in the literature. The final model studies the ability of a bidder in an auction to organize collusion among her rival bidders and the resulting impact of this collusion on the seller. Bidders valuations are private information. I show that in a two bidder, discrete, independent private-value auction, the seller earns less when a bidder can offer her rival a collusion proposal than in the absence of collusion. This contrasts with a cele- brated result by Che and Kim ("Robustly collusion-proof implementation". Econometrica, 74(4):1063-1107, July 2006) stating that for such auctions there is a mechanism that eliminates all the effects of collusion. Che and Kim and much of the literature assume an uninformed third-party organizes collusion.