Demand estimation techniques and investment incentives for the digital economy infrastructure: An econometric and simulation-based investigation

Large data communication networks such as The Internet have become the vital part of the global business infrastructure. However, many issues remain to be resolved before the present network infrastructure can facilitate tremendous anticipated growth in electronic commerce. This dissertation blends economic theory with management science techniques to provide important insight into anticipated problems with resource management and to provide viable solutions, especially as the level of use of public data networks keeps increasing. The first theme of this dissertation is the estimation of data network users' demand characteristics. Internet technology has drastically changed the availability, depth and frequency of update of relevant customer information, which has become a virtual data flow updated minute by minute. Effective harnessing of this raw mass of real-time information into accurate estimates of user's demand characteristics can be immensely useful due to the ease of rapid customization of digital products. However, as this dissertation will show, the dynamic nature of this information renders classical estimation techniques unsuitable. Hence a new non-parametric method was devised, which can fully exploit the advantages of the rich stream of publicly available information about users' choices to produce real-time estimates of demand characteristics. This method was tested in a simulation environment and its superiority is clearly demonstrated. The existence of such a method will also prove significant in dealing with the second topic of this dissertation: The investigation of impact of two different network resource pricing schemes on the level of future investment in network's infrastructure. This dissertation will describe the effects of implementation of flat rate pricing and congestion based pricing, two commonly proposed alternatives, on long-term network-wide investment. It will be shown that the ability of each of the pricing schemes to provide incentive for vigorous investment in future network expansion will greatly depend on the relative per-unit cost of capacity of network resources as well as on the presence of competition. Potential of these findings to provide applicable guidelines for management of expansion of computing resources in data networks is emphasized.