Conventional Cost-Benefit Analysis with Distorting Taxes and the Revised Samuelson Condition

When projects are evaluated using a conventional Harberger (1971) cost-benefit analysis the welfare effects are separated with lump-sum transfers. But this does not appear possible when governments raise revenue with distorting taxes. Evidence to support this view can be found in Mayshar (1990) and Wildasin (1984) who derive a marginal social cost of public funds (MCF) that depends on how the government spends the extra revenue raised. Ballard and Fullerton (1992) use this MCF in place of the conventional Harberger (1964) measure to amend the revised Samuelson condition obtained by Pigou (1947). We show that a conventional cost-benefit analysis is possible in this setting by decomposing their revised condition into conventional Harberger terms. The welfare effects of marginally increasing the public good are isolated by hypothetical lump-sum transfers that are offset separately with a distorting tax. We also demonstrate that when the marginal costs and benefits of providing the public good are measured by changes in utility (denominated in units of a chosen numeraire), the income effects are irrelevant because they impact equally on each dollar of cost and benefit. Consequently, projects can be evaluated correctly using uncompensated welfare changes