We offer a new theory as to why ethical people engage in corporate misconduct such as accounting fraud, rogue trading, and suppression of drug side-effects. Modelling ethics as other-regarding preferences we show all agents, however ethical, have an in-too-deep threshold. If private knowledge that an activity they have begun is harmful arrives after a threshold time then the agent will choose to hide this fact and continue the activity, knowingly harming third parties -- until publicly caught. We show that unethical agents can sell cash-flows linked to their activities at higher prices. That unethical agents inhibit the rate at which the market learns the value of their activities. That ethical agents are more likely to lose innovation races. And yet ethical agents are welfare enhancing. We show the in-too-deep threshold is always robust to patents and that only punitive punishments after public revelation can mitigate the problem.