Elsevier

Information Economics and Policy

Volume 45, December 2018, Pages 47-51
Information Economics and Policy

Quantifying and interpreting network effects: ‘Own-network’ is not the same as ‘firm-level’

https://doi.org/10.1016/j.infoecopol.2018.08.002Get rights and content

Highlights

  • Own-network and firm-level network externalities are two different concepts.

  • The ratio of cross- to own-network effects is not a ‘clean’ measure of compatibility.

  • As an alternative, I propose the ratio of industry- to firm-level network effects.

Abstract

This note revisits existing research on network effects in the mobile telephony industry. It highlights that – defined correctly – own-network and firm-level network externalities are two different concepts, and that mixing them up leads to inaccurate interpretations. In particular, I demonstrate that the ratio of cross- to own-network effects is an imprecise measure of compatibility because it does not accurately isolate network effects that convey market power from effects that do not. As an alternative, I propose the ratio of industry- to firm-level network effects.

Introduction

An increasing number of papers on an expanding range of industries try to identify network effects and/or try to quantify the value of ‘platforms’ that benefit from such effects (Van Hove, 2016a, Van Hove, 2016b, Alabi, 2017). This is not surprising. In a world of networks, a proper understanding of the nature and strength of network externalities – whereby the utility of consuming a good or service increases with the number of users – is of crucial importance for both company strategy and government regulation. A particularly critical issue is whether the network effects are predominantly situated on the firm or the industry level. As Bhargava (2014, p. 201) explains, if the relevant network “comprises all users of all firms, […] each firm's own network size ceases to be a differentiator”, and thus loses its strategic relevance. Regulators, for their part, can take comfort in the fact that network size is, ceteris paribus, less likely to command market power.

This note points out that the terminology that the literature uses when examining the relative importance of firm- and industry-level network effects can be misleading. For one, there is no shortage of terms. Besides firm- vs. industry-level, there is own- vs. cross-network, intra- vs. inter-firm, operator- vs. industry-specific, individual vs. global, and even other variants.1 Sometimes authors use several of these dichotomies in one and the same article but purely as synonyms and with a consistent definition – in what amounts to a relatively innocuous display of ‘elegant variation’; see Behringer (2014) for an example. That one author's intra-firm effects are another author's individual effects is also only mildly annoying at most.

There are, however, authors who overlook that the concepts behind the terms can be completely different. And if one then starts to mingle terms, interpretations can go awry. In what follows, I illustrate this by revisiting Grajek's (2010) analysis of the Polish mobile telephony industry and comparing it to other research, also on the telecom industry. In particular, I first show that own-network effects – the extent to which growth of a network is driven by an autonomous increase in its own installed base – comprise not only firm-level or operator-specific effects (which are exclusive to the network) but also industry-level or cross-network effects (which, as the names signal, are common to all networks). Obviously, cross-network effects are not a source of market power. As a result, Grajek's ratio of cross- to own-network effects is not a ‘clean’ measure of compatibility, because it compares effects that do not command market power with effects that in part do and in part do not. I argue that the ratio of industry- to firm-level effects is more intuitive to interpret (even though, unlike Grajek's measure, it does not lie neatly between 0 and 1).

In what follows, Section 2 first sets the scene by summarising Grajek's approach, and Sections 3 and 4 point out the interpretation problems. Section 5 concludes.

Section snippets

Grajek's approach

Grajek's (2010) goal is to gauge inter-network compatibility in the Polish (2G) mobile telephone industry. His point of departure is that, even if all mobile phone networks in a country are technically fully compatible, on-net call discounts (and possibly other mechanisms) can lead subscribers of a given network to value users of other networks less than users of their own network. In order to examine this, Grajek develops a structural model of consumer demand for mobile phone services, which

‘Own-network’ is not the same as ‘firm-level’

Let me first, in order not to detract from the point that I will make, dramatically simplify Grajek's diffusion equation toyi,t=ωyi,t1+χyi,t1.That is, I ignore entry by new operators, I set the standalone value of mobile phone services at zero, and I ignore pricing. In other words, I only want to examine how growth of a network is driven by its own installed base and by that of competing networks. In so doing, I assume away any non-linearities. Compared to Eq. (3), I have also changed the

4. What is the problem then?

The upshot of the above analysis is that one should not equate own-network with firm-level network effects. Unfortunately, this is what Grajek does. He initially reasons in terms of industry and firm-level effects, along the road – on p. 132 – starts using ‘own-network’ as a synonym for ‘firm-level’ (and ‘cross-network’ as a synonym for ‘industry-level’), and, as Eq. (4) shows, ends up estimating own- (and cross-)network effects. Using ‘cross-network’ and ‘industry-level’ as synonyms is not a

Discussion and concluding remarks

The main point of this note is that one should not equate own-network with operator-specific (or firm-level) network effects, and that doing so can – depending on the goal of the analysis – lead to erroneous interpretations. The paper by Grajek (2010) is a case in point. Grajek uses the ratio of cross- to own-network effects (1) to determine to what extent Polish mobile phone subscribers value their own network more than competing networks (which is not problematic), and (2) to derive insights

Acknowledgements

I would like to thank my colleague Marc Jegers and two anonymous referees for their comments. Special thanks go to Michal Grajek for the interesting discussion, which helped crystalise my thoughts.

References (11)

There are more references available in the full text version of this article.

Cited by (2)

  • Enhancing customers' prevention efforts: An incentive feedback mechanism design

    2021, Journal of Southeast University (English Edition)
View full text