State-Dependent Pricing and Its Economic Implications

In a packet-switched integrated-services network, the service provider can charge users a state-dependent price, which depends on the extent to which the network is congested. Alternatively, one can also charge users a long-term average price, which is set based on expected demand and capacity availability derived from prior experience at that time-of-day, but independent of instantaneous network conditions. In this paper, we study the economic implications of both statedependent and long-term average pricing. Using dynamic programming and computer simulations, we compare benefits that the service provider, society, and consumers can derive under different pricing schemes. Our results suggest that adopting state-dependent pricing improves both profit to the service provider and total benefits to the society. Those improvements are achieved for two reasons. First, state-dependent pricing functions as a traffic management mechanism that leads to a better packet throughput. Second, because under state-dependent pricing, a higher price is charged during the congested periods, limited capacity can be used to send more valuable packets at those times. Meanwhile, the service provider can extract more wealth from each packet sent. While the first effect benefits the service provider, society, and consumers, the second effect can reduce consumer benefits. Therefore, consumers may or may not benefit, depending on which of these two effects is more significant. We demonstrate that with smaller buffer size or inelastic demand, the second effect is more likely to dominate the first, which will result in a lower consumer benefit. 1 The authors gratefully acknowledge Brendan Dunham for his hard work in coding simulation programs, and Professor Marvin Sirbu for his helpful comments. Responsibility for errors are solely retained by the authors. 2 Member of Technical Staff in Bell Laboratories and doctoral candidate in the Department of Engineering and Public Policy, Carnegie Mellon University. Address: 600 Mountain Avenue, Bell Labs, Murray Hill, NJ 07974. Tel: (908)-582-3798. Fax: (908)-582-3161. E-mail:[email protected] 3 Associate Professor at Carnegie Mellon University in the Department of Electrical and Computer Engineering, and the Department of Engineering and Public Policy. Address: Carnegie Mellon University, Dept. of ECE, Pittsburgh, PA, 15213-3890. Tel: (412)-268-7126. E-mail: [email protected]. http://www.ece.cmu.edu/~peha 2 Section