One last thought about the topic of Brian Galle's colloquium paper earlier this week. The paper notes that there is a lot of overlap between "internality" and "externality" explanations for policy interventions such as inducing retirement saving through Social Security and employee pension rules, or addressing obesity through taxes on junk food or a "Big Gulp" ban.
What makes the two explanations overlap so much is the fact that, if people mess up badly, society may end up rescuing them at some fiscal cost to taxpayers. In effect, if you are drowning in deep water and the lifeguard has to go out there, the fact that you are in such trouble both creates the inference that you may have erred and leads to rescue costs being borne by others.
The big exception that occurs to me is smoking. It's arguably a bad choice most of the time, if we view smokers' testimony that they want to quit as trumping the Becker-Murphy rational addiction theory. However, while sick smokers may get treatment at others' expense, one of the less widely known facts about smoking is that it can be fiscally advantageous to state and federal budgets. The states collect cigarette taxes, and the feds don't have to provide years of Social Security and Medicare benefits to smokers who die relatively fast and young.
Obviously, we don't want these people to die, and encouraging smoking due to its fiscal benefits would be a sick and inhuman policy. What's more, passive smoking by nonsmokers is a clear externality that needs to be part of the calculus, although I've heard differing estimates of how medically significant it is, outside a given household that combines smokers and non-smokers. But the fiscal point does weaken the argument for basing anti-smoking initiatives on externalities, rather than on beneficence.
Many of the students in my colloquium class expressed distaste for regulation addressing internalities, and for the associated dirty word "paternalism." I agree that paternalism can be odious in practice, and that our usual starting point should be to expect that a given individual, whether perfectly rational or not (and, of course, for all of us the answer is not), combines access to unique psychic information about his or her own preferences and welfare, with having the right incentive to promote said welfare. But I nonetheless am perfectly fine with paternalist arguments for, say, Social Security forced saving, and don't require converting them first into fiscal externality arguments (although those matter as well).
What makes paternalism emotionally odious, I think, is the view that some smug individual is doing it to you, based on an unfounded superiority complex masking lofty ignorance (or even active dislike for people who are different). Thought of that way, it's a status insult, and possibly even an aggressive rather than benevolent act. But I think of paternalism, in appropriate cases, as something that I want to do to myself because I recognize the internal conflict between different drives and the fact that these may not always end up being resolved in the way that I reflectively prefer.