Abstract of the Dissertation

Some Aspects of Internet Portal Market Competition

by

Volha Chuvakin

Doctor of Philosophy

in

Economics

Stony Brook University

2007

This dissertation investigates consumer behavior using the web portal industry as an empirical setting. Specifically, it explores the connection between features proposed by the portal and its success on the market and models the relationship between consumer characteristics and his online choices.

Two important questions concerning the online market competition are explored. First, the behavior of users on the Internet market is studied by examining the market shares of Internet portals, and establishing the connections between different portal characteristics and their attractiveness for people. Later, users’ switching decisions are analyzed as a function of their own demographic characteristics and portal attributes using the survival analysis methods.

It was demonstrated that individual portal features such as Portal age, Mail and Search quality, are very important in explaining the overall market share, but less powerful in explaining the market shares of separate services. Although Mail and Search can be treated as major determinants of market shares: increase in Mail and Search quality can lead to an increase of market share for 5.6% and 4.4% respectively; the existence of Greetings, News service, Messenger and Weather service plays positive role in forming the consumer preferences towards the portal, adding up to 5% to the number of existing customers. In addition, separate market shares for the most popular portal services are estimated and interconnections between them are analyzed. The results of the estimations point that market shares of search and services associated with virtual communities are determined not only by the overall quality of portal attributes, but also by the demographic characteristics of users, namely, Age and Education. It was also discovered that it is not the number of services, but the presence of high quality services improves the portal market share.

Survival analysis for the portal switching was introduced to further investigate the patterns of consumers’ behavior on the online portal market. Logit probability estimation together with duration models in three different specifications is utilized to understand what factors lead to potential users’ drop off.

Again, it was confirmed that the main factors contributing to the survival probability are existence of high quality portal services. Portal can raise the probability of survival by a factor of 1.22 by offering the high quality Mail service; existence of such portal features as Shopping, Finance, News, which increase the probability of survival by factors of 1.93, 1.64 and 1.21 respectively. Among the demographic characteristics, User age and Household size increase the probability of survival by 0.7% and 0.4%, and higher levels of user education reduces the probability of survival by 3.6%.

This dissertation is among the first to explore online consumer heterogeneity. The portal users’ behavior is explored by dividing them into two groups based on the level of activity online. It is assumed that more active users, who surf the Internet intensively while switching constantly from site to site, will demonstrate different rates of portal drop-offs than regular users. This result is used to support the assumption that Internet users’ population is heterogeneous and the behavior of different groups of users should be modeled separately. Kaplan-Meir model suggests that active users are attracted by the higher quality of services and have 25-35% higher proportion of switching. A question whether users from multiple member households produce higher switching rates is also introduced. Due to the nature of information spillover, it is hypothesized that increased number of switches will result from household information sharing. The work suggests the existence of such extra switching phenomenon.

The approach developed and applied in this dissertation can be effectively used for new, more detailed click stream data from today's portals. More importantly, the results can be used for the modern market of mobile portals, which is currently undergoing the stage equivalent to one explored in this research.