For Gary Becker the point of criminal penalties was not to drive crime rates towards zero. The point was to make sure that the criminals internalized the cost of committing crime. You may speed to get a relative to hospital on time. You might steal food when you are starving. According to Becker, the penalties shouldn't be so high that people won't do those crimes. The point is similar for firms, even the anti-trust crimes that the Economist discusses. After all, antitrust violations can be efficiency creating even if the government doesn't recognize that possibility as often as it should. Becker was arguing that there was likely a non-zero optimal amount of crime and it was not for the reasons that the Economist magazine says regarding false conviction, though that might also play a role.
Indeed, Mr Becker’s crime calculus might lead to the conclusion that fines should be as draconian as possible—seizing all a wrongdoer’s assets, for example. Anything lower reduces the expected cost of criminality, without doing anything to improve the probability of detection. (Treating whistleblowers leniently is consistent with this logic: letting them off punishment raises the odds of truth-telling, and therefore of detection.) There are plenty of arguments against ultra-high fines, however. One is that false convictions carry too high a cost. Another is that fines of this sort could cripple firms, reducing competition. . . .