"Control fraud" theory explains why the most damaging forms of fraud are situations in which those that control the company or the nation use it as a fraud vehicle. The CEO, or the head of state, poses the greatest fraud risk. A single large control fraud can cause greater financial losses than all other forms of property crime combined - they are the "super-predators" of the financial world. Control frauds can also occur in waves that can cause systemic economic injury and discredit other institutions essential to good government and society. Control frauds are commonly able to defeat for several years market mechanisms that neo-classical economists predict will prevent such frauds.
Black is arguing that fraud is a problem for economists, because we assume that it can easily be prevented through audits and such.
I think it is a problem for libertarians in general. Suppose that we do away with lots of government laws and government imprisonment. Then someone who has a Ponzi scheme or somebody who wants to pass off possibly tainted meat as pure is in a position of saying "If you catch me, I lose a little bit. If you don't, I win a lot." The fraud perpetrator has much to gain from outwitting the auditing agency and little to lose by failing to do so.
The government is, like the private sector, probably unable to eliminate fraud by auditing. However, it is in a position to send perpetrators to prison, which has some deterrent value.
Let's assume that the cost of preventing fraud solely through auditing is ridiculously high, perhaps even infinite. However, suppose that the threat of imprisonment, when accompanied by an auditing regime, prevents most fraud at relatively low cost. In a deregulated, libertarian-ish society, how would you create the deterrence value that the threat of imprisonment provides?