سرویس رومینگ و سرمایه گذاری در بازار اینترنت تلفن همراه

This model discusses mobile network operators' (MNOs) incentives to invest in their network facilities such as new 4G networks under various regimes of data roaming charge regulation. Given an induced externality of investments (spillovers) due to the roaming agreements it will be shown that MNOs, competing on investments, widely set higher investments for below cost regulation of roaming charges. Otherwise, if MNOs are free to collaborate on investments, they set higher investment levels for above cost roaming charges. Both below and above cost charges may be preferred from a welfare perspective. Furthermore, the paper discusses effects of the roaming charge regulation on roaming quality and MNOs' coverage.

The telecommunications industry faces the transition from third generation (3G) to 4G networks such as long-term-evolution (LTE) networks. ‘The boom in data traffic has caused mobile operators a lot of problems, because they need to invest in their existing 3G infrastructure and soon in new technologies like LTE and Wimax while maintaining parallel technologies like GSM (and in some cases CDMA),’ says Uwe Steffen, the head of Nokia Siemens network's radio access solutions. The demand for mobile data traffic is persistently increasing throughout recent years. Global mobile data traffic grew 2.3-fold in 2011, mobile data traffic was more than three times greater than total Internet traffic in 2000.1 Multimedia content through 3G services requires significantly larger capacities than voice or text messages through 2G, though. An email is normally between 1 and 50 kB, a page of an online newspaper can be 100 kB or more. The download of a song requires 2–5 MBs of data. Hence, the transition from 2G to 3G and the rollout of LTE and Wimax networks, also referred as 4G, causes mobile network operators (MNOs) to consider new infrastructure models. Network operators begin to collaborate on infrastructure building or rely on infrastructure or network sharing. Collaborations on infrastructure tend to play a key role in the rollout of 4G. In Spain and Sweden networks already collaborate in building capacities for the new standards. In Germany, the incumbent operator Deutsche Telekom recently stated it would be open for collaborations in building up the fast LTE-Networks.2 Under certain restrictions also Bundesnetzagentur (2010) (the national regulatory authority) would embrace such collaborations among competitors. Network sharing as an infrastructure model can be implemented at different levels of a mobile network. It may take a form of passive sharing of masts and antennas or sharing active elements of radio active access networks (RNAs) or roaming in the core of the network.3 The present paper focuses on cross sharing of infrastructure in terms of national data roaming. National roaming is considered as a form of network sharing where MNOs in the same country code, which are usually direct competitors, use each others' networks. The agreement permits subscribers to roam onto a host network if the home network is not available in a particular location or in a time of congestion. This kind of agreement has especially been employed at the early stage of the 3G rollout and in peripheral areas. In the early 2000s operators in Europe (T-Mobile/O2 in the UK and in Germany or Tele2/Telia in Sweden) or Telstra/3 Australia in Australia entered into national roaming agreements to rollout 3G networks quickly to provide services in rural areas with low subscriber density and to provide additional capacity in congested urban areas or in times of congestion. Likewise the 3G rollout national roaming agreements will likely become important in the 4G rollout. Any agreement on collaboration and sharing of infrastructure is naturally of regulatory and competition authorities' interest. Collaborations are typically subject to Article 101 of the EU Treaty which defines criteria under which such agreements could be considered as anticompetitive, however, also allows for potential efficiency gains which are weighted against competitive harm. Referring to national roaming, two competition cases underpin the European Commission's current view on the potential impact of roaming on competition. In 2006 the European Commission found that the agreement between T-Mobile and O2 in Germany would restrict competition at the wholesale level with potential harmful effects in the downstream markets. According to the European Commission roaming would undermine infrastructure-based competition, since it would significantly limit competition on quality and transmission speed. Moreover, it would reduce scope for price competition at the retail level. The European Court of First Instance finally annulled the decision holding that the European Commission had not presented sufficient evidence of harmful effects on competition. However, it generally agreed that national roaming agreement may limit competition between operators, in particular when roaming occurs in urban areas or markets which can take more than one or two operators. The link between roaming agreements and their induced effects on competition on infrastructure building and competition on retail prices serves as the starting point of the present paper. It takes a roaming agreement between two MNOs, competing on the retail level and possibly competing or collaborating on investments, as given and analyzes its impact on providers' incentives to invest in infrastructure. Hitherto regulators tend to rely on operators to engage in negotiations to set a wholesale roaming price on each others' networks. Due to cross sharing of infrastructure the roaming charge for data services shows some similarities to widely analyzed two-way externalities in voice telecommunications. The effect of wholesale prices on competition and its regulation is extensively debated in the literature of voice telephony but is rarely addressed with data services, yet. The recent academic literature on voice telecommunications discusses the regulatory concerns under two-way network competition, where networks may use the termination charge as an instrument of tacit collusion because of a raise-each-other's-cost effect (see Armstrong, 1998 and Laffont et al., 1998). Fabrizi and Wertlen (2008) stress, though, that interconnection agreements within the mobile Internet services do not have entirely the same nature as interconnection agreements between voice communications operators. With voice telephony interconnection refers to enabling end-to-end users telecommunications traffic, which thus involves the origination of a given traffic within a network, its transportation, and its termination either in the same or the rival network. Data roaming instead refers to the access of the unilateral service by the rival network, origination and termination. The present model shows that although operators negotiate roaming charges above costs this does not necessarily harm retail competition, since the effects of roaming charges on investments in network quality have additionally to be considered. Investment incentives may be encouraged both by high and low roaming charges. Moreover, both roaming charges above and below costs might lead to under- and overinvestments from a welfare perspective. Recent research on roaming in the mobile Internet market is conducted by Fabrizi and Wertlen (2008). Their focus is on optimal market coverage given roaming agreements among networks. In their model MNOs are free to enter sharing agreements. They show that MNOs avoid network duplication in order to maximize rents from roaming. Valletti (2003) considers national roaming for mobile telephony and shows that only colluding operators have an incentive to engage in roaming agreements. The present model is in line with Valletti and Cambini (2005), who analyze voice communications providers' incentives to invest given different regulation regimes. In their two-way access model, networks tend to underinvest in quality, which is exacerbated if they can negotiate reciprocal termination charges above costs. The present model is on one-way access and builds up on a model of Foros, Hansen, and Sand (2002), who analyze demand-spillovers due to voice roaming and joint investments in the mobile voice communications market. They abstract from any wholesale pricing and regulation of roaming charges and show that under collusion on investments, firms' and a welfare maximizing regulator's interest coincide, whereas with non-cooperative investments, firms even overinvest. The present model extends the model of Foros et al. (2002) by analyzing data traffic and wholesale regulation of roaming charges. The paper is organized as follows: Section 2 describes the basic model. Section 3 solves the equilibrium in the retail market, whereas Section 4 analyzes incentives to invest given different roaming charge regulation regimes. Section 5 provides a social welfare analysis. Section 6 provides two extensions of the basic model, where MNOs choose the roaming quality and decide on their geographical coverage. Section 7 concludes.

Due to the widespread use of smartphones in recent years, MNOs have to discover new infrastructure models to meet the demand for third generation (3G) and to rollout 4G networks. Likewise the transition from 2G to 3G also in an early stage of the rollout of 4G providers tend to engage in network sharing and collaborations on infrastructure investments. From a competition policy perspective it is a relevant policy question whether MNOs should be allowed to collaborate in the investment stage and if and how to regulate network sharing. The results of the paper imply that MNOs' incentives to invest depend on: (i) the regime of roaming charge regulation, (ii) on the choice whether to allow MNOs to collaborate on investment levels, and (iii) on the extent of the investment spillover. Providers prefer an above cost roaming charge, whereas a social planner prefers a below cost charge, which however, might both lead to over- and underinvestments from a social point of view. The paper provides two extensions to the base model where providers are free to determine the roaming quality and their coverage. MNOs will choose maximal roaming quality whenever they are free to collaborate on investments and quality. However, if MNOs compete on investments and the roaming quality, the social planner might face a dilemma. For below cost charges MNOs increase investment levels but decrease the roaming quality, whereas for above cost charges they decrease investment levels but increase the roaming quality. The effect of roaming charges on coverage turns out to be ambiguous and depends on the relative magnitude in the retail and the roaming market. The model is very stylized and is one attempt to capture the impact of roaming charges on strategic interactions among competing providers. It is able to show that the level of roaming charges might quite diversely affect MNOs' choices on investments, quality, and coverage. Although, there is clearly room for further developing and putting more structure to the model. One extension is to model non-linear prices such as two-part tariffs. Subscribers in different countries considerably vary in adopting pre- and post-paid tariffs. In 2010, 87% of mobile contracts for over 25 year old customers in China were estimated to be pre-paid, in Italy two third, and in the UK roughly one half. Contrary, in Spain and the US roughly 80% were post-paid. With two-part tariffs providers have two instruments to capture consumer surplus, an increase in per-minute prices will be accordingly adjusted by a reduction in the fixed fee since for higher per-minute prices firms are more willing to compete for subscribers. Hence, the standard voice communications literature shows that access charges have a neutral effect on profits which reduces regulatory concerns of high access charges. Access charges, however, have a non-neutral effect on investment levels. The paper has shown that both high and low charges may foster investments and hence consumers surplus. Thus, likewise the model of Valletti and Cambini (2005) access charges will have a non-neutral effect on profits. In their model above-cost charge have an investment-reduction effect. This may not necessarily hold true in the present setup and will depend on the extent of the investment spillover and the choice of whether to compete or to collaborate on investments.