I see that the USDA/ERS report on "snack taxes" was picked up by a tax law weblog this Fall. Kudos to authors Fred Kuchler, Abebayehu Tegene, and Michael Harris at USDA's Economic Research Service for shedding some light on this issue. Various forms of the snack tax have been proposed (for example, Battle and Brownell in 1996). The cleverest (like Nestle's book Food Politics in 2002) suggest spending a snack tax on nutrition education or physical activity programs. This feature provides a "heads we win/tails we still win" quality to the proposal. If the tax causes people to stop eating junk food, then it provides a health benefit to the public. If instead the tax has little impact on consumption -- which is what the ERS economists predict -- then it is all the better for generating revenue that... you guessed it... provides a health benefit to the public.