With the growth of fibre networks in Europe, the proportion of overbuild (two or more network infrastructures in a single area) is steadily increasing in some countries. In the UK a recent post by Robert Kenny (Kenny) calculated that around a third of homes passed are served by two networks, around 40% of new built fibre networks are overbuilds of an existing network and that these proportions are steadily rising. In the UK it is easy to imagine areas in which three network infrastructures are built: Openreach (the incumbent telco), Virgin Media (the incumbent cableco) and a new entrant (the largest of which is CityFibre).
Given the significant fixed costs of rolling out a fibre network, overbuild results in higher costs per connection as the addressable market is shared between the competing networks. So it is perhaps surprising that networks are willing to overbuild. Much will depend on the take-up rate (connections per home passed) that the second network can expect.
Three questions on take-up
There are three important questions on take-up that are relevant to the commercial viability of overbuild:
(i) What is the overall rate of take-up that can be expected from homes passed by a fibre network?
(ii) What proportion of take-up can be “captured” by the first builder?
(iii) Is total take-up enlarged by the presence of two or more competing infrastructure networks in the same area?
Of these, (i) can be estimated from the dynamic relationship between homes passed and take-up. At any one point in time the new take-up of a network will come from a blend of areas that have had fibre built at different times, from those where the network was built five or six years ago, to those only newly built, and all points in between. By econometrically modelling take-up in any one year against the historical build profile over time we can deduce an estimate of the ultimate take-up when the market is fully mature in the current build cohort of actual and potential customers. My own modelling, based on a panel dataset across European fibre networks, suggests we can expect fibre take-up to reach at least 90% in the current cohort of homes passed by a fibre network. This, of course, is a lower bound for a forecast, since future cohorts of actual and potential customers may well have a higher willingness to take-up fibre gigabit service, and so we can reasonably expect future take-up to comfortably reach somewhere within the 90-100% range.
On question (ii), the same modelling of European take-up helps us take a view on the proportion of take-up “captured” by the first builder. My econometrics points to an initial slow start for take-up in the first year followed by a fairly steady increase in take-up over the subsequent six years following the completion of network build in any individual area, to eventually reach around 90%.[1] Therefore, a second fibre infrastructure, that completes its build a year after the first may find about 10% of the potential market already taken – especially if consumers are reluctant to have a second fibre entry point into their home. The second network, therefore, can expect an eventual take-up of no more than 40%.[2]
The final question, (iii), is difficult to answer without time series data on overbuild which is less readily available. Overbuild, with two or more competing network infrastructures, brings price and service quality competition at the wholesale network level, and more choice to consumers in respect of network quality[3] and, potentially, service capability.[4] McKinsey considers that a second fibre network will grow overall take-up from a little over 50% to 75%.[5] My own economic modelling suggests that competition from a second infrastructure helps accelerate eventual take-up from 70% to 90%.
To conclude
An overbuild network, built up to a year after the completion of the first network, can still reach an ultimate take-up of around 40% of homes passed and provide a competitive spur to ensure the market reaches its potential. The questions remains about whether this provides sufficient customers to cover the costs of network build, although this may well be the case in urban and suburban environments where the fixed costs per home passed are lower than rural builds where customers are more sparsely distributed.[6]
Commercially, returns will always by lower under overbuild situations and, with lower take-up and fixed costs, profit volatility will be higher with respect to any over or under performance on take-up, so investors will require a higher return (cost of capital). Nevertheless, it may still be profitable provided the second network does not delay network build beyond around a year, thus avoiding too much of the market being already taken.
Additionally, of course, any second network disadvantage may be offset by product differentiation (e.g. a 10 Gbps XPS-PON network) or a wholesale only business model that emphasises the role of retail ISPs in developing the market.
[1] In round numbers, 10% in the first year, followed by 16% in each of the subsequent 5 years.
[2] i.e. (90%-10%)/2. Although this would be lower if there were a second overbuild network.
[3] For example, network availability and reliability.
[4] For example, CityFibre is rolling out XGS-PON capable of 10Gbps connections (Progress Update on CityFibre's UK Rollout of 10Gbps XGS-PON Broadband - ISPreview UK).
[5] The keys to deploying fiber networks faster and cheaper | McKinsey, Exhibit 1.
[6] For example, see SPC Network, "Economically Sustainable Ultrafast Broadband Markets: A two sided market analysis", Economic_Market_Sustainability_Report_Final_V4.pdf, in particular Figure 5.
