A regressive tax on junk food might actually stop it from being consumed. This goes to the question of sin taxes that I believe we had here earlier, although I might have engaged in it somewhere else. Ideally, the sin tax would provide remediation from any externalities coming from the activity being taxed. For example, if public funds are necessary to treat type II diabetes in children or adults funded by Medicaid, then the "soda tax" should tap that amount. If there is zero consumption, then there will be zero cost due to that consumption. If taxes are too low, there will be too many external costs and not enough revenue to cover them. The question is, when does the tax on a dollar's worth of soda correct the damage done by that amount of product? Can it be estimated? If so, there is your tax amount.