A quick behavioral economics quibble to the USDA paper I summarized yesterday showing that subsidizing fruits and vegetables would do little to increase their purchase: A smart friend pointed out that there's a real question of study design. The economic model used by the researchers is based off consumer responses to natural changes in the price of produce. They took Nielsen Homescan Data -- which tracks self-reported food-purchase information from a random panel of households, much like the normal Nielsen's tracks self-reported viewing choices -- and tracked that against produce prices. As I understand the methodology, they used the actual change in buying patterns that resulted from actual changes in prices to estimate the likely impact of the hypothetical subsidy program. To take that out of study speak, they took what happened when a dollar's worth of apples naturally fell by 10 cents in order to estimate what would happen if a government program made a dollar's worth of apples 10 cents cheaper.But that's not how the subsidy program would work. Presume a big subsidy -- 30 percent, say -- that is explained directly to the targeted population. Rather than produce prices quietly changing without warning -- which is the natural data the study used for its estimations -- the targeted population is given a card -- maybe it looks like this one -- that they present to the supermarket cashier, at which point 30 percent is knocked off the cost of the produce in their cart. Under this regime of stated and predictable savings, the relative cheapness of produce would be more apparent to the targeted population, and so they might be more likely to try and increase their purchases of produce relative to other foods. This, incidentally, seems like the sort of question that some foundation -- or even the USDA -- could easily and fairly cheaply test in a pilot program and, for all I know, already has. COMPARING APPLES TO APPLES.
A quick behavioral economics quibble to the USDA paper I summarized yesterday showing that subsidizing fruits and vegetables would do little to increase their purchase: A smart friend pointed out that there's a real question of study design. The economic model used by the researchers is based off consumer responses to natural changes in the price of produce. They took Nielsen Homescan Data -- which tracks self-reported food-purchase information from a random panel of households, much like the normal Nielsen's tracks self-reported viewing choices -- and tracked that against produce prices. As I understand the methodology, they used the actual change in buying patterns that resulted from actual changes in prices to estimate the likely impact of the hypothetical subsidy program. To take that out of study speak, they took what happened when a dollar's worth of apples naturally fell by 10 cents in order to estimate what would happen if a government program made a dollar's worth of apples 10 cents cheaper.But that's not how the subsidy program would work. Presume a big subsidy -- 30 percent, say -- that is explained directly to the targeted population. Rather than produce prices quietly changing without warning -- which is the natural data the study used for its estimations -- the targeted population is given a card -- maybe it looks like this one -- that they present to the supermarket cashier, at which point 30 percent is knocked off the cost of the produce in their cart. Under this regime of stated and predictable savings, the relative cheapness of produce would be more apparent to the targeted population, and so they might be more likely to try and increase their purchases of produce relative to other foods. This, incidentally, seems like the sort of question that some foundation -- or even the USDA -- could easily and fairly cheaply test in a pilot program and, for all I know, already has. 
