National Roaming Pricing in Mobile Networks
Published on: Mar 3, 2016
Transcripts - National Roaming Pricing in Mobile Networks
National Roaming Pricing in Mobile Networks
Jonathan Sandbach 1,2
Head of Regulatory Economics
Tel: + 44 7795 300653
National roaming enables a new mobile entrant to compete against existing incumbent mobile
networks in the absence of full national coverage by its own infrastructure. A typical situation is
where the new entrant provides its own network infrastructure only in major urban areas, and relies
on roaming onto a competitor’s network in the remainder of the country. New 3G network
operators invariably require national roaming to provide full national coverage, because of the high
cost of building a WCDMA network at 2.1 GHz in rural areas.
What wholesale price should the new entrant be expected to pay in any commercially negotiated
deal, or a rate set under regulatory oversight?3 From an economic efficiency perspective there are
two criteria. Firstly, an efficient economic price needs to cover the marginal costs (including the full
economic costs of additional capacity) imposed on the host network. A price below this could
result in the new entrant providing services at a price to the end user below the true economic cost
of the resources it uses. However, in rural areas, where the network has been installed for
coverage rather than capacity, this is likely to be low. Secondly, and more significantly, where the
new entrant focuses its own network build on urban areas, the national roaming price needs to
contribute to the host network’s fixed coverage costs in rural areas. Otherwise the host network
will be victim to “cream skimming” by the new entrant, with the result that competition will be
distorted and ultimately dynamic efficiency of the competitive market will be damaged.
This second criteria is likely to be the dominant factor in determining national roaming prices. The
issue will be accentuated if the new entrant operates a WCDMA 2.1 GHz network. Other things
being equal, spectral efficiencies should provide 3G operators with greater traffic capacity, and so
in areas where the network is dimensioned for capacity (rather than coverage), as will often be the
case in urban areas, unit costs can be expected to be lower than for a GSM 900/1800 network
carrying the same amount of traffic and with the same absolute amount of allocated spectrum.
However, the situation reverses in rural areas, where the network needs to be dimensioned for
The views expressed in this paper are those of the author, and should not necessarily be attributed
BA (1st Class) Mathematics/Economics University of York
MSc (Distinction) Statistics London School of Economics & Political Science
PhD Economics City University, London
1978-1990: British Telecom, Economist & Statistician
1990-1998: National Economic Research Associate (NERA), Senior Consultant
1998-2004: Cable & Wireless Director of Economic & Regulatory Analysis
2004-2007: Vodafone Group Head of Regulatory Economics
National roaming prices are normally commercially negotiated, since the new entrant will have a
competitive choice of host networks.
coverage rather than capacity. Here the cost advantage will lie with GSM 900/1800, rather than
WCDMA at 2.1 GHz.4
Therefore, national roaming rates that seek to preserve a competitive neutral market (to maximise
the dynamic efficiency between competitors) will need to allow for:
1. The cost differential that the incumbent faces in rural areas, where the demand for national
roaming will be greatest;
2. The cost disadvantage it may face in urban areas, if it is restricted to GSM 900/1800
The need for (1) should be obvious. The need (2) is less immediately clear but equally important.
To see this imagine a WCDMA 2.1 GHz new entrant that was granted national roaming in rural
areas at a price that reflected only the GSM 900/1800 incumbent’s cost in those rural areas. Since
the new entrant may have a cost advantage in urban areas, the overall cost of an efficient new
entrant, would be below the overall cost of the incumbent (because of the cost differences in
urban areas). This simply reflects the efficiency of the new entrants 3G technology in urban areas,
and could be replicated by an efficient incumbent if they were permitted to install the same 3G
technology. If, however, they are not able to deploy equivalent technology, they will be at an
unavoidable cost disadvantage compared to the new entrant, that could ultimately effect the
viability of the incumbent network, on which the new entrant depends.
It follows, therefore, that in determining national roaming charges (by either commercial negotiation
or regulatory over-sight) the geographical cost structure of both the incumbent and the new entrant
network will be relevant (the latter only in the case where the incumbent doesn’t have access to the
same technology as the new entrant). The time profile of national roaming prices is also relevant.
As the new entrant network expands its network coverage, a greater proportion of the roaming
traffic is in more remote areas, with progressively higher unit cost. Setting a time profile for
roaming according to an ex-ante anticipated network build will provide the correct incentives for the
new entrant to complete its geographical network build.
Bottom-up Long Run Incremental Cost (BU-LRIC) models developed by regulators allow us to
explore a representation of the cost structure of a host mobile operator, and also its client network.
By comparing the costs of networks with different coverages it is possible to estimate the
incremental cost to the host network of the particular coverage over which national roaming is
provided, and also the cost advantage of the new entrant in urban areas. This it is possible to
derive an efficient price, consistent with criteria (2). This paper will provide results based on the
2G BU-LRIC model of OPTA (Netherlands), and the 2/3G model of Ofcom (UK).
Although the situation would be different if spectrum re-farming were permitted.