Content uploaded by Maurizio Naldi
Author content
All content in this area was uploaded by Maurizio Naldi on May 21, 2016
Content may be subject to copyright.
Concentration in the mobile operating systems
market
Maurizio Naldi
Universit`y of Rome Tor Vergata
Department of Computer Science and Civil Engineering
Via del Politecnico 1, 00133 Roma, Italy
naldi@disp.uniroma2.it
Abstract. Concentration phenomena concern the ICT market. Though
the regulatory action has been active mainly in the telecom network
operators industry, even more significant worldwide concentration phe-
nomena affect other industries. The market of mobile operating systems
is analysed through two concentration indices to get a quantitative pic-
ture of the current situation and its evolution over time: the Hirschman
Herfindahl Index (HHI) and the Four-Firm Concentration Ratio (CR4).
A strongly imbalanced oligopoly is shown to exist, where the four major
operating systems take over 99% of the market, but the dominant oper-
ating system Android alone is installed on over 80% of the new devices.
Keywords: Operating Systems; Concentration; Competition; HHI
1 Introduction
Market structure and the presence of dominant operators (manufacturers and/or
service providers) has been a significant field of activity in industrial policy since
long [18]. An operator holding a very large share of the market, or even acting
as a monopolist, may take advantage of its position and enforce unfair poli-
cies towards its customers, which in turn have little or no room to oppose. The
attention for the appearance of dominant positions is at the root of the birth
of a number of national anti-trust agencies, both at the national and superna-
tional level [6], which enforce rules against anticompetitive agreements, abuses
of dominant position as well as concentrations (e.g., mergers and acquisitions,
joint ventures) which may create or strengthen dominant positions detrimental
to competition.
The issue is particularly delicate in ICT industries, where operators may
often benefit of economies of scale, which would lead to a natural monopolistic
structure as the most efficient one [15]. Noam has carried out a broad analysis
of concentration phenomena in several ICT and ICT-related industries [13] [14]:
–Books
–Film
–ISP
arXiv:1605.04761v1 [cs.CY] 16 May 2016
2 M. Naldi
–Magazines
–Multi- channel
–Newspapers
–Online News
–Radio
–Search Engines
–TV
–Wireless
–Wireline
In that survey, the highest HHI value is observed for search engines and is
roughly 0.75, quite above the second highest value, which is 0.55 and pertains
to the wireline telco market.
However, the survey of [13] leaves out a market that has often been at the
center of anti-trust disputes in recent years, which is the operating systems one.
The most notable ones have been the U.S.A. vs Microsoft case for the Windows
desktop operating system [4], and the very recent Statement of Objections raised
by the EU vs Google for the mobile operating system Android [1].
In that Statement of Objections, the European Commission alleges that
Google has breached EU antitrust rules by:
–requiring manufacturers to pre-install Google Search and Google’s Chrome
browser and requiring them to set Google Search as default search service
on their devices, as a condition to license certain Google proprietary apps;
–preventing manufacturers from selling smart mobile devices running on com-
peting operating systems based on the Android open source code;
–giving financial incentives to manufacturers and mobile network operators
on condition that they exclusively pre-install Google Search on their devices.
However, in both cases just brief quantitative data are given about the market
structure.
Very few scientific papers have been devoted instead to competition in the
operating systems market. The effects of Microsoft Windows dominance on pos-
sible competitors and the run to standardization has been analysed in [7]. In
[3] the case of a dominant player (Windows) competing with a zero-price player
(Linux) has been considered. An analysis of the desktop/laptop market for oper-
ating systems using an organizational ecology approach has instead been carried
out in [8], where a Lotka-Volterra model has been employed to investigate the
evolution of competition and forecast equilibrium states, in the light of the pres-
ence of open source software.
The relevant literature exhibits a gap of quantitative studies concerning mo-
bile operating systems.
This paper provides a few data to analyse the mobile operating systems mar-
ket and its evolution through the 2007-2015 period. Two concentration indicators
(the Herfindahl-Hirschman Index - HHI - and the Four-firm Concentration Ratio
- CR4) are employed to provide a quantitative indication of market concentra-
tion. It is shown that the dominance by Android has replaced that by Symbian
Mobile OS concentration 3
(with the cross occurring in 2011) and that the current market structure is closer
to a monopoly than to a tight oligopoly (with the dominant OS holding more
than 80% of the market and the four largest OSs’ share in excess of 99%).
The paper is organized as follows. The rough data concerning the OSs’ shares
are reported in Section 2, while the concentration indices are computed in Section
3.
2 Mobile operating systems market 2007-2015
In this section we provide information on the data employed to carry out the
concentration analysis.
We use the raw data as provided by the quarterly market analyses released
by Gartner (Gartner smartphone market share). In particular, we consider the
worldwide smartphone sales, arranged by operating system, in the 2007-2015
period. The numbers refer therefore to new devices and not to the overall number
of actually working phones.
The operating systems expressly considered are reported hereafter; those not
appearing are cumulated as others:
–Windows Mobile;
–RIM;
–Symbian;
–iOS;
–Android,
–Bada;
–Windows Phone.
The respective market shares (as computed over the total number of smart-
phone sold in the same period) are shown in Fig. 1
As can be seen, the evolution over the 9 year-interval can be roughly divided
into two periods: the first one, where the dominance of Symbian has been steadily
declining (till the negligible market share held today), and the second one where
the dominance of Android has steadily risen, till a plateau reached in 2014, which
is about the 80% share of today. The crossing point, where Android has taken
the lead, can be roughly identified in 2011. It is to be noted that the share held
by Android today is much larger than that held by Symbian in the days of its
dominance.
3 Market concentration index
In order to assess the market structure, in this section we compute two concen-
tration indices for the raw data mentioned in Section 2. In addition, we analyze
the relevance of the operating system holding the maximum share.
As concentration indices, we consider the two most relevant indices employed
in industry studies: the Herfindahl-Hirschman Index (HHI) and the Four-firm
Concentration Ratio (CR4).
4 M. Naldi
2008 2010 2012 2014 2016
Year
0
20
40
60
80
Market share
Windows Mobile
RIM
Symbian
iOS
Android
Bada
Windows Phone
Fig. 1. Market share of operating systems 2007-2015
The HHI of a market where ncompanies operate, and the market share of
the i-th largest company is si, is
HHI =
n
X
i=1
s2
i.(1)
Its value ranges in the [1
n,1] interval, with the minimum value representing the
perfect competition case (all companies having the same share of the market)
and the maximum value (1) representing the absolute monopoly case: larger
values represent increasing level of market concentration. Its relationship with
the Zipf law, a well-known rank-size model, has been described in [9].
In our case, we consider operating systems instead of companies.
Unfortunately, we do not have a complete picture of the market composition,
since we know market shares just for the more widespread OSs. In such a case,
we can however compute bounds for the HHI. Here we employ the formulas
proposed in [12] and reported in Table 1, where Mis the smallest OS whose
market share we know, R= 1 −PM
i=1 si, and Q=bR/sMc.
When we apply the formulas of Table 1 to the data of Section 2, we get the
curves shown in Fig. 2. We note that the bounds are very tight, since the two
curves are practically indistinguishable: the difference between upper and lower
bound is at most 3.5% and below 1% in 89% of the quarters examined.
Mobile OS concentration 5
Type Bound
Lower PM
i=1 s2
i
Upper (R≤sM)PM
i=1 s2
i+1−PM
i=1 si2
Upper (R > sM)PM
i=1 s2
i+s2
MQ+1−PM
i=1 si−sMQ2
Table 1. Bounds on HHI
2008 2010 2012 2014 2016
Yea r
0.3
0.4
0.5
0.6
0.7
HHI
Lower bound
Upper bound
Fig. 2. HHI in the mobile operating systems market 2007-2015
6 M. Naldi
The HHI curve shows anyway that after a slight dip in 2011 (corresponding
to the crossing point, where Symbian and Android held a roughly equal share
of the market), the concentration indicator has steadily risen, reaching peaks in
excess of 0.7. Is this a high value? Though a precise correspondence cannot be
drawn between the numerical value of the HHI and the qualitative indication of
a level of concentration (or, equivalently, of competition), we can resort to the
guidelines provided by the U.S. Department of Justice for horizontal mergers
(first in 1985 and later revised several times). In Section 5.3 of the latest version
of 2010, a classification of markets into three types is proposed, as reported in
Table 2 [16]. According to this classification, the HHI levels observed at any
time during the 9-year long interval considered, and in particular in the latest
years, are typical of a highly concentrated market. Actually the current value,
exceeding 0.7, is by far above the threshold (0.25) that separates moderately
concentrated markets from highly concentrated ones.
HHI Competition level
<0.15 Unconcentrated Markets
0.15–0.25 Moderately Concentrated Markets
>0.25 Highly Concentrated Markets
Table 2. Levels of competition and the HHI
In addition to the HHI, we consider another well known concentration index:
the Four-Firm Concentration Ratio (CR4). The CR4index has been the most
relevant index to measure concentration before the advent of the HHI [17]. Its
relationship to the HHI has been investigated in [10,11]. It is given by the sum
of the market shares of the four largest firms in the market [2]
CR4=
4
X
i=1
si.(2)
While it is clear that a low value of the index represents a larger competition
level, and a high value (close to 100) represents an oligopoly situation, there is
not a general consensus on the correspondence between the value of the index
and intermediate concentrations. Typically, if CR4<0.4, the industry is consid-
ered as very competitive. A complete classification table is proposed in [5] and
reported here in Table 3.
We show in Fig. 3 the CR4 for the same data for which we computed the HHI.
We observe that even in 2007 the CR4 was representative of a tight oligopoly
situation, but the following years have seen a steady rise in concentration (with
the slight dip again around 2011), and since 2013 the value is extremely close
to 1, which means that the 4 dominant firms take all the market. Even in the
Mobile OS concentration 7
CR4Competition level
0 Perfect Competition
0–0.4 Effective Competition or Monopolistic Competition
0.4–0.6 Loose Oligopoly or Monopolistic Competition
>0.6 Tight Oligopoly or Dominant Firm with a Competitive Fringe
Table 3. Levels of competition and the CR4
years of HHI depression, the CR4 was well above the 0.6 barrier, with most of
the market just redistributing among the dominant players.
2008 2010 2012 2014 2016
Yea r
0.84
0.86
0.88
0.9
0.92
0.94
0.96
0.98
1
CR4
Fig. 3. CR4 index for the mobile operating systems market 2007-2015
Finally, in order to assess whether the market structure is closer to an
oligopoly or to a monopoly, we plot in Fig. 4 the market share owned by the
top operating system. We observe a behaviour quite similar to the HHI, with
a dip around 2011, where the dominant operating system (then Symbian) held
roughly 1/3 of the overall market. But both before and after that dip, the market
saw instead an operating system getting an absolute majority. Since 2013, the
8 M. Naldi
dominant operating system is quite steady around 80% of the overall market, a
situation never attained before, which we can classify as an imbalanced oligopoly
(where the imbalance is within the oligopolist circle itself).
2008 2010 2012 2014 2016
Yea r
0.3
0.4
0.5
0.6
0.7
0.8
Top market share
Fig. 4. Market share of the dominant operating system 2007-2015
4 Conclusions
The structure of the mobile operating systems market has been investigated
through two concentration indices: the Hirschman-Herfindahl Index and the
Four-Firm Concentration Ratio. Their evolution in the 2007-2015 period shows
that a tight oligopoly has been a characteristic of this market ever since. A dip in
HHI around 2011 marks the passage from the dominance by Symbian to the cur-
rent dominance by Android. Currently, the four dominant operating systems are
installed on over 99% of new devices. The present oligopoly is however strongly
imbalanced, since Android alone holds a share larger than 80% of the market of
new devices.
Mobile OS concentration 9
References
1. Antitrust: Commission sends statement of objections to google on android operat-
ing system and applications factsheet. http://europa.eu/rapid/press-release_
MEMO-16-1484_en.htm, accessed: 2016-05-10
2. Arnold, R.: Economics. Thomson South-Western (2008)
3. Casadesus-Masanell, R., Ghemawat, P.: Dynamic mixed duopoly: A model moti-
vated by linux vs. windows. Management Science 52(7), 1072–1084 (2006)
4. for the District of Columbia, U.S.D.C.:
5. Gwin, C.: A guide for industry study and the analysis of firms and competitive
strategy. Available at http://faculty.babson.edu/gwin/indstudy/
6. Jacquemin, A.: Towards an internationalisation of competition policy. World Econ-
omy 18(6), 781–789 (1995)
7. Kretschmer, T.: Upgrading and niche usage of pc operating systems. International
Journal of Industrial Organization 22(8), 1155–1182 (2004)
8. Lakka, S., Michalakelis, C., Varoutas, D., Martakos, D.: Competitive dynamics
in the operating systems market: Modeling and policy implications. Technological
Forecasting and Social Change 80(1), 88–105 (2013)
9. Naldi, M.: Concentration indices and Zipf’s law. Economics Letters 78(3), 329 –
334 (2003)
10. Naldi, M., Flamini, M.: Correlation and concordance between the cr 4 index and
the herfindahl-hirschman index. Available at SSRN 2502764 (2014)
11. Naldi, M., Flamini, M.: The cr4 index and the interval estimation of the herfindahl-
hirschman index: an empirical comparison. Available at SSRN 2448656 (2014)
12. Naldi, M., Flamini, M.: Interval Estimation of the Herfindahl-Hirschman In-
dex Under Incomplete Market Information. In: 16th International Conference on
Computer Modelling and Simulation, 2014 UKSim, Cambridge, United Kingdom,
March 26-28, 2014. pp. 317–322 (2014)
13. Noam, E.M.: Who owns the worlds media? In: TPRC 41: The 41st Research Con-
ference on Communication, Information and Internet Policy (2013)
14. Noam, E.M., International Media Concentration Collaboration: Who Owns the
World’s Media?: Media Concentration and Ownership Around the World. Oxford
University Press (2015)
15. Nooteboom, B.: Information technology, transaction costs and the decision to’make
or buy’. Technology Analysis & Strategic Management 4(4), 339–350 (1992)
16. The U.S. Department of Justice and the Federal Trade Commission: Horizontal
Merger Guidelines (19 August 2010)
17. Weinstock, D.S.: Using the Herfindahl Index to measure concentration. The An-
titrust Bulletin pp. 285–301 (1982)
18. Zlinkoff, S.S.: Monopoly versus competition: Significant trends in patent, anti-trust,
trademark, and unfair competition suits. The Yale Law Journal 53(3), 514–552
(1944)
