Andrew Samwick doesn't favor agricultural subsidies, really, but he says something odd in this post. In response to a statistic noting that "nearly 70 percent of the subsidies go to the top 10 percent of agricultural producers," he asks how much these 10 percent of farms actually produce:

If it is about 70 percent, then we would figure that there is probably nothing perverse about the way the subsidies are being doled out.

But that doesn't seem like a good way to evaluate a distribution at all!

Let's imagine I have 10 farmers in my imaginary country, none of them receiving subsidies. One farmer, Farmer Joe, produces 70 percent of the country's food. But for various reasons, he also turns massive profits while doing so. Now the other nine farmers produce the other 30 percent of the country's food, but they are all in danger of going bankrupt, or are exceedingly poor. Clearly, giving 70 percent of all agricultural subsidies to Farmer Joe is somewhat frivolous here. Not only that, but it ends up being harmful—you get overproduction of food, spur environmental waste, promote poor farm management, and indirectly hurt Third World farmers.

Now my imaginary world sounds very similar to the real world, where farms are expected to pull in a record net cash income of $73.7 billion in 2004—most of that going to the top 10 percent. And yet agricultural subsidies continue to rise. So while I'm not sure exactly how to measure the efficiency of subsidies, doing it like Professor Samwick did above seems to miss something. The rest of his post is quite interesting, though.