VoxEU Column Competition Policy Europe's nations and regions

How does access regulation affect broadband penetration?

Most governments in Europe are interested in boosting broadband penetration. This column presents new evidence on how various forms of market structure affect penetration ratios. It turns out that inter-platform competition (e.g. DSL vs cable) promotes broadband penetration while intra-platform competition doesn’t.

The evolution of broadband penetration has shown substantial regional differences, both between and within OECD countries. Are these differences mainly due to external circumstances such as the level of development or geographic and demographic conditions? Or do regulatory policies explain the differences, i.e. the existence or absence of different modes of competition between broadband suppliers? These questions are of key importance to policy makers aiming to fulfil the productivity growth objectives in the Lisbon agenda of 2000, or to achieve other goals such as reducing inequality. The research on this to date has been incomplete.  Wallsten (2006) looked at the relationship between broadband penetration and different forms of broadband regulation without distinguishing between different platforms, and Denni and Gruber (2006) studied the effects of intra-platform competition (incumbent vs. entrants) and inter-platform competition on broadband penetration. 

In a recent study we addressed these questions from two different perspectives. First, we conducted an international study of the determinants of broadband penetration, based on its evolution in twenty OECD countries during 2003-2008. We considered the role of different modes of platform competition after taking into account other determinants of broadband penetration. Second, we conducted a case study for Belgium. From being a frontrunner in the early years, Belgium has gradually fallen behind a significant number of other European countries. As a result, Belgium now belongs to the group of OECD countries with an intermediate broadband penetration. Our international study distinguishes between three modes of competition between different broadband suppliers; see Figure 1.

  1. Inter-platform competition refers to competition between the incumbent DSL (Digital Subscriber Line) operator and operators on different platforms, most notably cable (but possibly also fibre-to-the-home and wireless).
    Intra-platform competition refers to competition between different DSL operators making use of the incumbent’s network.
  2. Under facilities-based intra-platform competition (so-called unbundled local loop, or ULL), entrants lease bare unbundled local loop elements, but have to invest in own equipment and facilities.
  3. Under service-based competition (such as bitstream access), entrants are merely reselling the incumbent’s services and therefore incur few investments themselves.

To measure the degree of these three modes of competition we constructed Herfindahl indices of market concentration. A high Herfindahl index means that market shares between or within platforms are asymmetric, thereby reflecting limited competition. Our econometric analysis then assessed how the degree of inter- and intra-platform competition may affect broadband penetration, after accounting for other factors, such as a country’s GDP per capita, its population density, etc.

Figure 1. Three forms of competition and the construction of Herfindahl indices

Our findings indicate the following regarding competition:

  • Inter-platform competition has a significant, positive effect on broadband penetration. For example, broadband penetration will tend to be 10% higher in a country where DSL and cable have equal market shares, compared to a country without a cable operator.
  • Facilities-based intra-platform competition has an insignificant effect on broadband penetration.
  • Service-based intra-platform competition has a significant, negative effect on broadband penetration. But the magnitude of this effect is less important than that of inter-platform competition.

We also found that other factors contribute to explaining differences in broadband penetration across countries.

  • A $1000 increase in GDP per capita tends to raise broadband penetration by 0.9%.
  • A 10% increase in PC-penetration is associated with a 2% increase in broadband penetration.
  • A cross-country difference in population density of 100 inhabitants/km2 tends to increase broadband penetration by 3%.

In sum, these findings show that external factors are important in explaining the differences in the evolution of broadband penetration across countries.

Regulation matters too

Differences in regulatory policies have also played a crucial role. Countries that promoted competition between different platforms (e.g. by investing in cable infrastructure) have done significantly better. In contrast, countries that mainly promoted service-based intra-platform competition on the incumbent’s network have on average done worse. This is consistent with the view that service-based competition does not provide sufficient investment incentives to new entrants and discourages investment of the incumbent operator.

Our case study of Belgium reveals interesting additional insights. The country as a whole was a frontrunner in the early years, but has recently slowed down and now belongs to the intermediate group of OECD countries with a household penetration level of 59% in 2007; see Table 1. Behind this national number there are large regional differences: in Flanders (northern part) penetration reached 63%, compared with 60% in Brussels and only 52% in Wallonia (southern part). An important part of these differences may be explained by external conditions. Table 1 shows that Flanders had a higher GDP per capita, a higher PC-penetration and a higher population density, which are all conducive to broadband penetration.

Table 1. Regional differences within BelgiumVariable Belgium Flanders Brussels Wallonia

Variable Belgium Flanders Brussels Wallonia
Dependent variable        
Broadband penetration (fraction of households) 59% 63% 60% 52%
Competition variables        
Market share cable 39% 55% 33% 28%
Market share DSL incumbent 46% 36% 49% 53%
Market share other licensed operators (OLOs) 13% 7% 9% 17%
 - of which unbundled local loop 2% 2% 6% 3%
 - of which bitstream access 11% 6% 3% 14%
Market share others 2% 1% 10% 1%
Control variables        
Net taxable income/capita (€, fiscal year 2006) 13,655 14,483 11,550 12,807
PC-penetration (2007) 67% 72% 64% 61%
Population density (pop./km2, 2007) 345 448 6,601 202

*: Belgian market shares are taken from 13th EU Implementation Report (2007); regional market shares are estimates based on 2008 market survey by Belgacom; split of OLOs between unbundled local loop and bitstream access is based on estimates by Belgacom

However, the differences in broadband performance are also in part due to differences in regional policies. Investment in cable broadband was actively promoted in Flanders from the start, which resulted in a strong degree of inter-platform competition. In Brussels and especially in Wallonia the promotion of inter-platform competition did not take place, and instead there was more emphasis on service-based intra-platform competition (bitstream access). Table 1 indeed confirms that in 2007 cable broadband had reached a market share of 55% in Flanders, compared to only 33% in Brussels and 28% in Wallonia. Conversely, bitstream operators had reached a 14% market share in Wallonia versus only 6% and 3% in Flanders and Brussels. Combining this with the results of our international study we can conclude that Flanders has correctly promoted inter-platform competition, whereas Wallonia focused on service-based intra-platform competition which entails adverse investment incentives.


Based on our international econometric study and our case study for Belgium, we make the following policy recommendations to improve broadband penetration:

  1. Policy-makers should develop a long-term vision and promote inter-platform competition: The “ladder of investment” theory argues that it is good to promote intra-platform competition as a stepping stone for new entrants to induce them to invest. Our study shows there is no support for this theory, and that to the contrary intra-platform competition may even give adverse investment incentives. To improve broadband penetration, the promotion of inter-platform competition should become a priority.
  2. Policy-makers should gradually phase out bitstream access regulation. Our econometric results suggest that service-based intra-platform competition is worse than facilities-based intra-platform competition, which is neutral in this respect (it neither encourages nor discourages penetration). We therefore recommend as a first priority to phase out bitstream access regulation, for example by gradually increasing regulated bitstream access prices relative to ULL prices, or by not imposing mandatory bitstream access on new optical fibre networks, where substantial investments still need to be made.
  3. Policy-makers should pay attention to regional differences that affect broadband penetration: Our case study for Belgium shows that external conditions may explain large regional differences, but policy is also partly responsible. In regions that are lagging behind, policy makers should assess whether the right regulatory and investment policies are followed and focus sufficiently on inter-platform competition to preserve the right investment incentives.

The authors acknowledge Jonas Rosenstok for his excellent research assistance and Belgacom for financial support. The opinions expressed are exclusively those of the authors.


Jan Bouckaert, Theon van Dijk and Frank Verboven (2008). Regulation and broadband penetration – What is required to regain speed in Belgium?

Denni, M. and H. Gruber, 2005, The diffusion of broadband telecommunications: the role of competition, working paper 

Wallsten, S., 2006, Broadband and unbundling regulations in OECD countries, AEIBrookings Joint Center for Regulatory Studies, Working Paper 06-16.




st1\:*{behavior:url(#ieooui) } */ /*-->*/ /*-->*/

420 Reads