There has long been a controversy in economics about whether or not a firm with monopoly power should be allowed to exploit its monopoly power by engaging in third-degree price discrimination.' The effect of third-degree price discrimination on welfare is two-fold: (1) output under price discrimination is misallocated because different groups are charged different prices but (2) if total output is different under price discrimination than under uniform pricing, this will also affect welfare.2 Schmalensee [8], Varian [11; 12] and Schwartz [9] have shown that price discrimination must decrease welfare if output under price discrimination is the same or less than output under uniform pricing. This fundamental result is quite robust and holds regardless of the number of sub-markets or the behavior of marginal cost (rising, constant or falling); it also holds even if the sub-market demands are interdependent. A necessary, but not sufficient, condition for price discrimination to be beneficial is that it increase total output. Perhaps the most common situation in which price discrimination will be beneficial is the case where only one sub-market is served under uniform pricing but two or more sub-markets are served under price discrimination. In this case Hausman and Mackie-Mason [3] have pointed out that price discrimination will result in a Pareto welfare improvement if marginal cost is constant or falling. Even when price discrimination does not open up new markets. Hausman and Mackie-Mason [3] and Nahata, Ostaszewski, and Sahoo [5] have shown that price discrimination can result in a Pareto improvement by lowering prices in all sub-markets. Hausman and Mackie-Mason [3, 257] argue that price discrimination is most likely to be beneficial when economies of scale are present:3