[Food] subsidies enjoy significant political and popular support because it is believed they at least protect the nutrition of the poor. This is especially the case when comparing subsidies to other forms of welfare, like cash payments, since people worry that the poor may spend the cash on things other than food, or at least may not use the money in a way that improves their nutrition.
But when we subsidized the price of rice and wheat, people consumed less of them, not more.
And in a follow-up paper, we show that when you take together all the consumption substitutions people make, our large subsidies did not improve nutrition at all.
In fact, in Hunan nutrition actually declined in response to the subsidy. In Gansu, the effect on nutrition was essentially zero. And our sample included only the very poorest households, malnourished by international standards and earning much less than a dollar per person per day — i.e. the exact group whose nutrition subsidies are intended to improve.
Now, we’re absolutely, definitely not saying that we should therefore do nothing to help the poor. ... But while there may be some reasons to prefer subsidies as a form of welfare, it’s time to abandon the long-standing presumption that they are the best policy because they improve nutrition.
--Robert Jensen, Freakonomics blog, on the implications of Giffen goods for anti-poverty policy