Revealed preference and collective action

There's been an interesting exchange between Henry Farrell and Megan McArdle on revealed preference arguments. McArdle says that if people wanted higher taxes, higher levels of foreign aid and better public services, they would donate money freely. Henry replied that this kind of revealed preference argument assumes that choices are independent, which they are not. People might want to give more money to the government only if other people give more money too. If that's true, then there might be the kind of collective action problem that revealed preference arguments ignore.
(Two further installments of the debate: McArdle here, Farrell here)

I think Henry is clearly right here, but also that he doesn't pick the strongest argument against revealed preference arguments like this.

Preferences can be faulty. A desire to commit rape isn't a desire we should aim to fulfil at all. Of course, it's controversial as to which preferences are irrational (so for my two cents, marketing is the art of making preferences faulty, but I suspect people like McArdle would deny this). But one way in which it's uncontroversial that they can be faulty is when they're based on false information. If I desire to buy that car because it's reliable, when in fact it will break down next week, then my desire is faulty. Revealed preference arguments, on the other hand, imply that many people want to buy second hand cars of bad quality. "Why else would they buy them?"

This is obviously relevant here. If I desire to keep hold of my money because charities would only waste it, and it's false that charities waste money, then my preference is faulty and potentially ignorable. That means that we first need to establish how effective governments and charities are before we can decide whether or not revealed preference arguments work. That, of course, would make the revealed preference arguments redundant anyway.

On the same topic, Will Wilkinson suggests that there's an inconsistency in those who agree with Henry here, but reject the idea that there are collective action problems for things like carbon emissions and meat consumption. In the latter case many people assert that you ought to do what you can even if others don't.

This is partly why I think Henry picks the wrong counter-argument, since it leaves him open to this problem. I think we should accept that we have obligations to freely send money where it's needed - perhaps to charity - if doing so is effective, even if others don't do so as well.

But I also think that Wilkinson here assumes that both goods are equally "lumpy" (i.e. linear in benefit to an equal degree). But this might be false. Whilst buying one less chicken this week has a reasonable chance of actually lowering the number of chickens killed next week/month/year, it's not obviously true that sending £10 to a charity has a reasonable chance of increasing the benefit they do by £10 worth. Perhaps charities need larger sums to begin and continue projects.

You might think that this is false (I suspect it is), and that £10 at Oxfam will indeed buy one more packet of AIDS-retrovirals. But if people only incorrectly imagine that charity expenditure is lumpy - and I think it is plausible that many people do believe this - then Wilkinson is locating the problem in the wrong place. Such people are not inconsistent, but just misinformed on the lumpiness of charity expenditure.