The article explores how insurance companies deal with moral risk by using a principal-agent model. It shows that optimal insurance contracts can't always perfectly share risks and may need partial coverage to motivate good behavior. As the size of a loss increases, the insurer's actual losses can also go up.