Strategic action in network industries: an empirical analysis of the European mobile, страница 12

^*  Concentration: Herfindahl index of market shares per country computed monthly:

X

Concentration ¼ S_i²  10000;

where S_i is the market share of each phone operator in its own country.

^*  Interaction PenetrationConcentration: The interaction between the two variables above was also computed to account for any possible joint effect.

^*  Nature of action: Each action was categorized as competitive or cooperative depending on whether they involved strategic collaboration with other network firms in any area of the world, e.g., joint-ventures, strategic alliances and acquisitions.

^*  Type of action: Each action was categorized as being related to one of the three primary activities of networks: infrastructure operations, network promotion and services layering.^[3]

6.  Results

Exhibit 1 provides descriptive statistics for the data collected, while Exhibits 2 and 3 show the results of the analysis of cooperative/competitive moves and the nature of the moves, respectively. The 199 actions in the dataset reported in any of the markets during the studied period were categorized as cooperative actions (73 moves in cooperation with at least one other national or foreign phone operator), competitive actions (103 moves made by individual operators), or other actions (23) moves, such as divesting or closing down facilities, that could not be categorized. In view of the categorical nature of the dependent variable (cooperative/competitive action), the likelihood of an action being cooperative as opposed to competitive was analyzed using a logit model. The log of the probability of an action being cooperative divided by the log of the probability of the action being competitive (log odds) was regressed over Market Penetration, Concentration, PenetrationConcentration Interaction, and Period (time dummy variable), plus a set of country dummy variables representing the 12 countries.

Hypotheses 1, 2, and 3 received clear support at the 0.05 confidence level. The likelihood of a strategic action being cooperative rather than competitive (i.e., taken by one firm alone) is greater at an earlier stage in time and in markets with a higher degree of penetration and concentration. As expected, cooperation is essential among operators in the early stages of the PLC as well as in markets with a high degree of penetration and concentration. As market penetration increases, firms engage in a greater number of alliances. Agreements to improve interconnection and roaming serve to increase connectivity across the established networks, hence also increasing traffic. New services, or service layers, are also aimed at increasing traffic. Cooperation allows for the sharing of development costs, while benefits from network externalities in new service layers are made possible by inter-operable technology and service agreements. Cooperation is beneficial in that it increases traffic within established networks. However, perfect or non-discriminatory interconnects also limit the operators’ ability to differentiate their services. Companies may favor cooperative over purely competitive moves when subscriber bases become more stable. Cooperative moves are also more likely to occur when concentration is high, since firms cannot disregard the terms of access to the

Exhibit 2

Likelihood of Cooperation versus Competition

Logit Regression Analysis