Social Welfare Maximization and Network Pricing

The growing number of users of the Internet are facing the unsettling realization that the communication services they have received in the past, for the most part free of
charge, will probably no longer be free in the near future.(note 1) As the government stops playing the role of Internet sponsor and private companies become the
providers of network communication services, no longer will there be anyone willing to lose money on networks for the good of the people. The networks that people will
be using in the future, whatever form they may take, will be subject to the same economic forces guiding other markets--consumers\' demand and producers\' supply.
Whereas under government sponsorship supply was not dictated by monetary profit, private network providers will need to make at least zero profits. Charging users for
their services will be a natural means of achieving this goal.

One of the questions that should be addressed before producers begin setting prices is what price will maximize the consumers\' surplus regardless of profits. Next, we
can examine how this pricing strategy can be amended to give producers a return on their costs of production with as little loss to social welfare as possible. Analysis of
social welfare maximization in the arena of information networks, including cost recovery, is explored by MacKie-Mason and Varian (1994). Their work is extended
here to include not only the sunk costs of capacity, but also the more realistic case with fixed costs and costs of usage as well.

Two facets of information networks make the analysis of pricing and its effects on social welfare significantly more complex than the simple \'price equals marginal cost\'
solution usually applied by economists. One is the externality of delay, and the second is the near-zero marginal production cost of usage. In other words, an extra
increment of network usage by any consumer has virtually no cost of production to the supplier yet creates congestion that affects other users, which will be an impetus
for increasing the network\'s maximum capacity. Therefore, not only does \'price equals marginal cost\' pricing fail to recover the large fixed costs of the network, but it
also fails to address the effects of the externality cost. A necessary characteristic of a pricing system that will maximize social welfare, then, is attention to the marginal
cost of delay imposed by one consumer\'s extra usage upon others. In addition, although marginal costs of usage to producers are negligible, as network provision
becomes more commercialized one can predict that extra increments of usage will create a cost to producers. For example, commercialization will most probably cause
increased complexity of the network and the services it provides, and debugging to ensure smooth operation of such a complex network will present a cost of production
proportional to the amount of traffic on the network.

Adding fixed costs to the analysis is more easily envisioned. For private firms, fixed costs would have to include the costs of billing users. If billing becomes more
difficult as the amount of usage by a particular customer increases, or logging each increment of usage presents a cost to the producer, there is also a marginal cost of
usage associated with billing as well. Providing technical support to users also represents a fixed cost, as there will be a minimum number of employees and a minimum
infrastructure needed no matter how many people make use of the support services. Again, since more usage will probably lead to more need of technical advice for
users, this will be another source of marginal cost to the producer.

Therefore, the basic model, which only considers the sunk costs of capacity, will be of less practical use in the future, as producers must also be compensated for the
marginal and fixed costs discussed above. Producers will have three simple pricing strategies to consider: flat-fee access charges, usage-based pricing, or a combination
of both. These three cases are considered and the problems and characteristics of each are elaborated upon to provide a basis of understanding for the application of
economic concepts to information network pricing.

The benchmark for comparison of the various pricing schemes will be the case of maximizing social welfare regardless of producers\' profits, or the \'first best\' case.(note
2) Cases with the constraint that producers must make at least an arbitrary amount of profit, or the \'second best\' cases, are considered