Service Bundling and Churn of Fiber Users: Evidence from the Netherlands

Abstract

In Europe, switching of broadband users has been a key problem for internet service providers (ISPs) for some time now (CEC, 2009). In the Netherlands, ISPs have introduced a form of self-regulation in the industry in order to influence the switching behavior of broadband users. Churn of broadband users has been at rates of around 3.5 percent over the past five years (OPTA, 2010). To deal with churn of broadband users, ISPs have offered a large variety of service bundles aimed at segmenting the residential broadband market based on speed and bundling differences. There has been a “sharp uptake” of triple-play and double-play bundles in the Netherlands since 2006 (CEC, 2009). As a result, the Netherlands has not only achieved one of the highest levels of broadband penetration in the world (OECD, 2012), but is also leading with respect to service bundles offered in Europe (CEC, 2009). Interestingly, residential broadband users often choose a service bundle not only based on a (lower) price tag, but are willing to pay (premium) prices for higher internet speed even if the service bundle remains (practically) unchanged. In particular, the (collective) switching pattern of residential users preferring fiber access over traditional (xDSL and Cable modem) broadband access technologies has been an example of this trend (Sadowski, Nucciarelli, & de Rooij, 2009). Due to competitive pressure and technological challenges posed by traditional broadband providers, ISPs offering fiber based access have become affected by churn of residential users. In the literature, competitive bundling has been considered as a strategic option to reduce churn by users (Belleflamme & Peitz, 2010). In contrast to conventional product bundling, service bundling (i.e. the promotion of two or more separate services e.g. TV, Internet or Telephony in a package at a single price) by ISPs has been aimed at either attracting new users to or preventing existing users from leaving the network (Prince & Greenstein, 2011). In this context, the paper examines the extent to which bundles provided by an ISP have an effect on residential users leaving the fiber network between 2009 and 2011 at a time when triple play service became pervasive in the Netherlands. Due to data limitations, research on churn of residential users on fiber networks has been rather scarce as most large scale surveys do not sufficiently distinguish between different categories of broadband. As a result, a number of propositions have been put forward in the literature ranging from churn rates which are “no different” (compared to traditional broadband networks) to churn rates which are “close to zero” . In the paper, churn, i.e. the abandonment of an ISP by an existing user (household), is linked to extent to which residential users adopt different service bundles by the ISP. To examine churn, we used data on the socio-economic characteristics of residential users from an on-line survey carried out in July 2009 (about 2,700 respondents) in the Eindhoven area in the Netherlands and linked it to survey data in 2011. The paper employs a dynamic probability panel model to explain household choices. The paper shows that service bundling has been an important factor in retaining residential users on the fiber network and to reduce churn. In particular, different service bundles involving TV services and a (high) preference of residential users for (higher) internet speed have been crucial for residential users to keep their existing subscription.