Introduction
This essay has developed a model which compares the effectiveness of capacity expansion and congestion tolls in dealing with internet congestion. It has also suggested that in the presence of positive usage externalities, the optimal congestion toll may need to be reduced or a subsidy introduced. This section of the essay will discuss a number of areas for possible research which would allow the model developed in this essay to be tested and improved.
Areas for Future Research
"Build it... and they will come" - such are the words that drive Kevin Costner to build a baseball diamond in the movie 'Field of Dreams'. But it may be argued to also apply to the internet if the model developed in this essay holds true. Recall from the discussion on capacity expansion that the reason it does not alleviate congestion is because users trade-off the costs and benefits until the original level of congestion is restored. Demand and usage is increasing in response to the expansion in capacity and associated lowering of congestion. Therefore, if an IAP wishes to increase demand for their service and increase usage, expanding capacity may promote an increase in usage. In evaluating capacity expansion however it was assumed that increasing capacity produced constant increases in marginal benefit, at any network size. This was necessary to ensure that the expansion in capacity would be matched by an increase in demand which restored the same level of congestion. An avenue for further research thus lies in determining over what range might this assumption be true and what level of capacity expansion will produce diminishing marginal benefit and increases in demand which do not restore the initial congestion equilibrium.
The Nature of Internet Demand and the Network Effect
The nature of demand for internet usage is an area which still exhibits extensive possibilities for further research. There does not yet appear to be a wide body of theoretical or empirical academic literature available on the nature of demand for internet usage.1 Elasticities of demand for usage are a key area which deserve further research. This is important when considering the actual implementation of a congestion toll. IAP's may only be prepared to introduce congestion tolls if they believe that demand is relatively inelastic, as elastic demand would mean a fall in revenue for the IAP. Furthermore, if the congestion toll is levied on the IAP by an external agency, the elasticity of demand will determine the shares of the toll borne by consumers and producers.
An additional area of research would be empirically determining the strength of the network effect as it relates to different measures of network size, and different network sizes. In this essay it was argued that increasing network size would yield initial positive and increasing marginal benefit, followed by decreasing positive marginal benefit and eventually a negative marginal benefit. Further research into the network effect may allow this hypothesis to be tested and for some estimation to be made of where the turning points and intercepts lie for the marginal benefit function. Ideally this research should endeavor to produce information on the social benefit as well as the private benefit of increased size.
Costs of Expansion vs Costs of Implementing a Congestion Toll
Before implementing a congestion toll it may be beneficial to conduct a comparison of the costs of expansion as compared to the costs of implementing the congestion toll. As was argued earlier in this essay, a major drawback to congestion pricing, especially dynamic pricing, is the implementation cost and the associated development of new systems and technologies to implement the such a scheme. Given that the costs of capacity expansion are argued to be falling (Hazlett (1994:2)), it may be appropriate to consider the cost effectiveness of capacity expansion. The argument in favour of capacity expansion may also be advanced if it is found that there is significant time lags in the adjustment and restoration of the congestion equilibrium.
Positive Externalities to Usage
Earlier in this essay it was assumed that usage may exhibit some positive externality. This was done so as to consider the possible impact on the optimal congestion toll. It was shown that in the presence of positive usage externalities, the optimal toll would be reduced and in extreme cases a subsidy might be required. The opportunity therefore exists to explore whether or not internet usage does display a positive externality, and if so, to attempt to gauge the size of it. Of most importance is perhaps not the absolute size of the externality, but its size relative to the negative congestion externality as this will determine whether a toll or subsidy should apply to usage.
Technical Trends - New Applications and Changing ADU's
Changes in ADU sizes and the application mix have been argued to alter the size of the congestion externality. It may therefore be important to consider the impact of changes in the behaviour of internet applications on congestion. For example, if there is an increase in push content, how will this impact on the dynamics of congestion? If web pages become more graphic intensive, will this have a significant effect on congestion? Could ADU sizes be reduced by greater use of compression? These are all issues which have the potential to affect the size of the congestion externality and thus the size of congestion tolls on the internet. Continuing research into trends in the way the internet is being used is therefore also important.
Conclusion
As this section has highlighted, there remains several areas for research in relation to internet usage and congestion tolls. The relationship between capacity expansion and increases in demand should be regarded as a significant area as it will influence the effectiveness of capacity expansion in dealing with congestion. It will also influence the extent to which IAP's can promote increases in demand for usage by expanding the infrastructure. At the same time however, the other issues raised also deserve attention. The nature of internet demand and the network effect are areas in which it was necessary in this essay to make a number of assumptions about. The validity of the model developed could therefore be more vigorously evaluated and the model itself improved by more research in this area. The presence of positive usage externalities, costs of capacity expansion and future trends in internet applications are also areas which deserve more attention.
Endnotes
1. The author is willing to speculate however that because of the high commercial value of such information that it may already be known and organisations may simply be protecting their commercial interests.
Last update 12-Jun-98. To visit my personal web site click here.