We consider an agency model in which the agent acquires private pre-decision information after contracting. The agent is allowed to breach the contract by paying the principal pre-determined damages. We demonstrate the relationship of this model to the standard no-breach agency model and argue that simplifying the analysis by restricting attention to no-breach models may yield incorrect conclusions. We then demonstrate that the damage-payment component of an optimal breach contract cannot take the form of a severence payment. Next, we consider a “substantitutive” institution to breach of contract whereby the agent may purchase access to private information prior to contracting. In this case all the advantages of the breach institution are maintained, while possible exogenous (legal) restrictions on damage payments are avoided. We conclude by suggesting that implications our study may have for legal research on contracts and judicial systems.