Consumer Preferences for Mobile Telecommunication Services Study

This report presents a study on consumer preferences for improvements in mobile telecommunication services, focusing on 4G upgrades and roaming. The research employs a choice experiment to estimate consumers' willingness to pay (WTP) for various attributes, including increased mobile internet speed, unlimited mobile internet use, improved quality, and unrestrained roaming in neighboring countries. The findings reveal that consumers value unrestrained roaming the most, followed by increased speed and unlimited use, while the impact of improved quality is statistically insignificant. The study concludes that bilateral roaming regulation is more valuable than 4G investments. The research also reviews prior studies on consumer preferences for fixed and mobile broadband services, providing context for the current analysis. The study was conducted in North Cyprus, a developing economy, and the results are relevant for other developing countries, particularly Turkey. The paper is structured with an introduction, background review, methodology explanation, model definition, results presentation, and a discussion with conclusions.

1
Consumer Preferences for Improvements in Mobile
Telecommunication Services
Orhan Dagli
Eastern Mediterranean University, North Cyprus
E-mail: [e-mail]
Tel.: + 90 533 865 5005
Fax: + 90 392 365 5901
Glenn P. Jenkins
Queen’s University, Canada,
and Eastern Mediterranean University, North Cyprus
Corresponding author
E-mail: [e-mail]
Tel: [phone]
Development Discussion Paper: 2015-05
ABSTRACT
We employ a choice experiment in order to estimate consumers’ willingness to pay for
improvements in mobile services, focusing on 4G upgrades and roaming services. The
attributes of an improved mobile service that we investigate in our experiment are: increased
mobile internet speed (possible with 4G), unlimited mobile internet use, improved quality
(possible with 4G) and unrestrained use in two neighbouring countries (unrestrained
roaming). The results indicate that people value unrestrained roaming services the most.
Increased speed and unlimited use attributes are next, and are similarly significant at the 1%
level. The impact of improved quality is statistically insignificant at the 5% level, suggesting
that consumers are content with the current level of quality they receive with 3G. We
conclude that bilateral roaming regulation between governments is more valuable than 4G
investments.
Revised paper published as: Dagli, O., & Jenkins, G. P. (2016). Consumer preferences for improvements in
mobile telecommunication services. Telematics and Informatics, 33(1), 205-216.
Keywords: Mobile telecommunication services; choice experiment; willingness to pay;
consumer preferences; 4G; roaming
JEL classification: C5, D12, L96

2
Research highlights:
 We model the utility consumers derive from improvements in mobile services.
 We estimate the marginal willingness-to-pay for each improvement.
 Consumers will pay 2% of their average monthly income for unrestrained roamin
 Consumers will pay for improved data rates, but not for improved quality.
 Consumers are indifferent between a 30 Mbps 4G service and a 300 Mbps 4G se
1. Introduction
Advances in telecommunications have turned the world into a more conne
‗globalized‘ place in the 20th century, and have been a major contributor to increased
efficiencyand productivityin every possible sector. Technologicalprogressin
telecommunications continues to change the way we live our lives in the 21st century.
Mobile communications (MC) has been the star of telecommunications in the pas
decades. Initially MC was a means for speaking and texting over mobile phones using
later GSM (2G) technologies. The introduction of 3G enabled mobile users to connect
internet and to send and receive various multimedia messages. Then 4G arrived
possible to access mobile internet with speeds that even some fixed broadband
cannot achieve. The International Mobile Telecommunications Advanced specification
peak speed requirements for 4G service at 100 Mbps for high mobility commun
1 Gbps for low mobility communication (ITU, 2008). Commercial 4G networks h
achieved the peak speeds of the specification, although they have spread rapidly aroun
since the early 2010s. As of November 2014, there are 331 4G LTE networks offering
data connection speeds, deployed in 112 countries. The top speeds available are offere
commercial 4G LTE-A CA networks launched in 14 countries, subscribers of wh
downlink data speeds ranging from 225 Mbps to 300 Mbps (Ericsson, 2014).

3
Numerous prior studies have focused on the MC sector. However, rapidly
technologies continue to open up new territories for academic and empirical research.
literature has touched on MC licensing and auctions (Klemperer, 2002; Fuentelsaz et a
mobile tariff discrimination (Haucap and Heimeshoff, 2011), mobile roaming (Fa
Wertlen, 2008; Stühmeier, 2012), MC adoption (Rice and Katz, 2003; Pagani, 2004; Bo
al., 2007), and consumer preferences for MC services (Kim, 2005; Shin et al., 2011; Kw
Yoo, 2012; Klein and Jakopin, 2014). This paper presents a brand-new study on the las
subject areas in this list.
We employ a choice experiment (CE) in order to estimate consumer prefer
selection of ‗current and crucial‘ improvements in MC services. The attributes we eva
increasedmobile internetspeed,unlimitedmobile internetuse, improvedqualityof
communications service, and unrestrained use abroad. These service upgrades a
most mobile markets around the world, and each one is of interest for a reason.
Although 4G is deployed in many countries, there are still many regions t
covered, and many more that are covered but lagging behind in terms of 4G t
Consumers of mobile services in these regions have yet to fully benefit from the featur
namely increased mobile internet speed and improved quality. Therefore, understandi
of introducing these features continues to be of interest. Unlimited mobile inter
interesting because most mobile broadband services on offer have data caps, w
broadbandservicesgenerallyprovideunlimiteduse. Mobilebroadbandcould becomea
competitor of fixed broadband if offered with unlimited use, so we aim to quantify the
consumers associate with this attribute. Finally, unrestrained use abroad is of i
people are travelling more than ever, and operators are charging excessively for roam
services. The EU has taken steps to regulate its roaming market (Salsas and K
Shortall, 2010; Infante and Vallejo, 2012), and recently independent countries h

4
make bilateral agreements for coordinated action on roaming services (Singapore and
2011 (The Independent, 2011), Australia and New Zealand in 2013 (MBIE, 2013
expect to see more countries follow suit in the near future, if the value for th
depicted more clearly.
Our aim in this study is to evaluate consumers‘ willingness to pay (WTP)
abovementioned attributes, as a measure of their value. We conduct 320 face-to-face i
with people from all regions of North Cyprus, asking respondents to choose be
existing mobile service and two other hypothetical alternatives with varying attribute l
estimate consumers‘ marginal WTP (MWTP) for each attribute by analysing how they t
between price and other attributes when making their choices.
North Cyprus is a developing economy in the Eastern Mediterranean with a popu
slightly below 300,000. Mobile use is spread widely throughout the country and
available mobile technology is 3G. The results of this study are useful for the
North Cyprus in designing a possible auction or tender for 4G licensing, and for mobil
operators in analysing the costs and benefits of future 4G investment. Similarly
should be of interest for all developing countries, and especially for Turkey, the
mobile market in the world in terms of number of subscribers in 2013 (ITU, 2015). Lik
Cyprus, Turkey has not yet introduced 4G, and the same operators dominate both the
market and the market in North Cyprus (Turkcell and Vodafone).
The paper is organized as follows. Section 2 reviews the previous research in thi
section 3 explains the methodology used in the study. Section 4 defines the mo
estimated and the estimation results are presented in section 5. Finally, section 6 disc
results and concludes.

5
2. Background
Estimating consumer preferences for the attributes of telecommunications
been a topic of interest among researchers since the advent of broadband internet in t
Earlier studies focused on fixed broadband services, while the focus has shifted toward
services since the 2010s as mobile technologies have caught up and overtaken fixed te
A number of notable stated preference studies that estimate consumers‘ valuations for
services and their attributes have been completed to date.
2.1 Consumer studies for fixed broadband services
Madden and Simpson (1997) were among the first to carry out research in this a
used data obtained from a national survey of households in Australia in order to determ
willingness of households to subscribe to a broadband network. The fact that broadban
were not currently available at that time was a complication for their study. O
households surveyed, 598 provided usable data. The authors employed maximum
estimation for a logit model, and found that the effects of the installation fee and incom
probability of subscription were statistically significant, whereas the effect of m
not. Other determinants for the probability of subscription were the size of the househ
age of the household head and whether the head was employed in a blue-collar occupa
Ida and Kuroda (2006) studied the Japanese market for broadband service
ADSL, CATV (cable television internet) and FTTH (fibre to the home). They em
discrete choice analysis with a nested logit model on a data set of 1,013 obser
showed that a nested choice structure of narrowband (dial-up, ISDN) versus broadban
CATV, FTTH) is the best model fit because of the sign conditions of price and speed va
their statistical significance and degrees of fitness. They also showed that the own-pric
of ADSL is inelastic, while the figures for CATV and FTTH are elastic, conclud
ADSL market is independent of other services.

6
Rosston et al. (2010) produced the most comprehensive CE study on the
internet market in the USA, and for the first time introduced the effects of att
authors employed discrete choice analysis to estimate the marginal WTP for im
eight internet service characteristics: cost, reliability, speed, laptop mobility, movie re
telehealth and videophone. The data was from a nationwide survey conducted w
respondents in late 2009 and early 2010. The results implied that reliability an
important characteristics of internet service. Estimated MWTPs were 20 USD pe
more reliable service, 45 USD for an improvement in speed from slow to fast, and 48 U
an improvement in speed from slow to very fast. MWTPs for the other attributes were
less. Valuations for broadband internet were larger for experienced households, and th
estimated two- to three-fold increase in consumer surplus between 2003, when a simil
was conducted, and 2010.
Carare et al. (2015) focused on measuring the WTP for broadband of non-adopte
USA. They reported that 28% of American households did not have a broadband subsc
of October 2012, and set out to identify the determinants of broadband adoption. The
a survey of 15,082 households conducted in 2011. Approximately two thirds of the res
stated that they would not consider subscribing at any price, for reasons such as a lack
a lack of a computer or other device. The authors found that, conditional on th
household characteristics, including education and the presence of children, the
broadband adoption increased with higher levels of income.
2.2 Consumer studies for mobile broadband services
The term ‗mobile broadband‘ was born with the advent of 3G technology in the 2
Since then, there have been a number of empirical studies evaluating consumer prefer
mobile broadband services, both 3G and 4G, and for related attributes.

7
Kim (2005) estimated consumer preferences for IMT-2000 (3G) services, fo
service upgrades including video telephony, global roaming and multimedia mob
applications. Using a survey of 250 respondents from Seoul, South Korea, Kim
variations in consumer valuation of 3G service upgrades. The results indicated that co
place a higher value on video telephony than on multimedia mobile internet and globa
services.
Shin et al. (2011) carried out a similar conjoint analysis for mobile service consu
Uzbekistan. Their primary aim was to identify the demand for mobile number portabili
which refers to consumers‘ right to keep their mobile numbers while switching betwee
service providers. Other attributes estimated in the study were price, call and
discount calls within the same network, and the mobile network operator company. Us
responses for their survey, the authors found that price and quality were the m
attributes, while subscribers did not consider MNP to be an important service upgrade
The first study evaluating consumers‘ preferences for 4G technology was by Kwa
Yoo (2012). It involved 500 person-to-person interviews held in Seoul, South Korea, in
CE was used in order to evaluate the MWTP for the following 4G attributes: data rates
communications service, number of broadcasting channels, video-on-demand (VOD) se
supplementary services. The authors found that ―consumers were interested in 4G an
prepared to pay for 4G services‖. Estimated per-month figures for MWTP were 4.03 U
improved communication service, 0.06 USD for an additional broadcasting chann
for VOD and 1.45 USD for supplementary services.
Klein and Jakopin (2014) took a different approach in their conjoint analys
attempting to investigate bundling of mobile telecommunication services. As mob
spread and competition in the mobile sector has intensified, mobile operators have aim
competitive edge by bundling services together, including, but not limited to, minutes

8
text messaging, internet access, and even financing for a mobile device. The au
data via an online survey among German consumers, and carried out their ana
responses out of a total of 355 surveyed. The results indicated that pricing wa
important attribute in a service bundle, followed by minutes included and internet acc
messaging was calculated to be the least important attribute. To account for the accur
estimated WTP figures, both linear calculation and curve fitting were conducted
parameter, with no significant change in results.
The current study is the first to estimate the importance to consumers of being a
use their local mobile package while travelling abroad (unrestrained roaming, in
increasing number of people around the world have travelling routines, and unrestrain
can be achieved through regulation. Furthermore, this study is an update on the consu
evaluating 4G, as we include in our attribute list the top data rates currently available
most advanced 4G technologies. This will shed light on the extent of the consumer dem
ever-faster mobile data rates.
3. Methodology
3.1 Choice experiments
The origins of the CE methodology date back to Louis L. Thurstone‘s 192
paired (comparison) CEs. Many authors have contributed to the literature on choice an
the final version of the CE methodology draws upon Lancaster‘s economic theor
(Lancaster, 1966) and random utility theory (McFadden, 1973; Hanemann, 1984
commonly used in various fields of economics and marketing to make choice-based val
goods, services and their attributes.
What sets CE apart from other stated preference elicitation methods is th
researchers to study not only the value of a commodity itself, but also the valu
attributes of this commodity. These attributes are the main factors influencing people‘

9
and hence, the value that individuals associate with each attribute is important inform
order to extract this information, the CE practitioner designs choice sets contai
levels of the attributes, and asks people in a survey to make choices between these set
method, the CE practitioner is able to analyse the marginal effect of each individual at
the consumers‘ utility.
In the contextof this research,CE methodologyenablesus to breakdown the
improvement in mobile services into the individual attributes that we intend to study. W
thus estimate separate marginal WTP figures for each individual attribute.
3.2 Attributes and attribute levels
The attributes for the study are identified through an extensive review of related
and through focus groups held with representative consumers.
Previous choice studies in MC literature, as mentioned in section 2.2, have each
different list of attributes. Kim (2005) selected mobile internet, video telephony
roaming as the attributes of interest for a mobile service. Shin et al. (2011) in
preferences for company name, discounts, service quality and availability of MN
Yoo (2012) identified five attributes of 4G technology, namely data rates, qualit
broadcasting channels, VOD service and supplementary services. Klein and Jakop
investigated service bundling.
In the current research, two distinct focus groups are conducted with rep
consumers to determine what they value in their mobile service, and which improveme
would be interested in. As 4G is yet to be introduced in North Cyprus, the principal co
among the focus groups is that increased data rates and improved quality through the
4G are valuable for everyone. With 4G, mobile data transfer rates increase from aroun
to a minimum of 30 Mbps and a maximum of 300 Mbps. The higher the data rates, the
users‘ experience with data-intensive mobile applications such as videophoning a

10
computing. The quality of communication is also improved with 4G because of
transition time between transmitters, which avoids freezing and disconnection while th
speaking or surfing the net on the move. Increased data rates and improved quality ar
attributes that we take into our attributes list, and that have been previously studied (
Yoo, 2012).
We do not find strong justification to include in our study the other attributes me
in the literature. The concept of broadcasting channels was difficult for most p
grasp, and its impact for the consumer is already captured by the attributes of
quality. Video telephony and VOD are already offered by third-party applications
free of charge, so people are not willing to pay extra for these services. Mobile interne
roaming are already available in today‘s standard subscriptions, and MNP, though cur
available in North Cyprus, is believed to be a consumer right and likely to rec
‗protest zero‘ valuations. Service bundling is not the focus of this paper, since we are
improvements in mobile services, rather than the bundling of existing services.
The focus groups identify two new attributes that are not mentioned in the litera
that are mentioned very frequently by our representative consumers. The first one is u
mobile internet use. Respondents are discontent with the state-run ADSL internet serv
is the only fixed broadband service available in North Cyprus. The system lacks
infrastructure is old and troubled, and, besides, ADSL technology is limited to a maxim
of 8 Mbps. The faster cable and fibre technologies would be too costly to introduce, so
remaining option is wireless connection. If mobile operators were to offer unlimited in
instead of imposing data caps, many people would be willing to replace their fi
connection with a mobile subscription.
The second new attribute is unrestrained use of mobile services when abroad. C
small island in the Eastern Mediterranean, and citizens of North Cyprus frequently tra

11
destinations: Turkey and South Cyprus. They travel for business, for entertainment, fo
or simply to take a flight to a third destination. However much they travel, they cannot
home mobile subscription freely, so they end up paying extra roaming fees or purchasi
local mobile number. If their mobile service offered unrestrained use in Turkey
Cyprus, which is possible through bilateral roaming regulation between governm
could use their home minutes and data plans in these destinations.
The final list of attributes and attribute levels is presented in Table 1. M
speed (more comprehensible than ‗data rates‘) has three levels: 3 Mbps with th
service, 30 Mbps with basic 4G implementation, and 300 Mbps with the most a
technology. Mobile internet limit and quality each have two levels. The present level fo
refers to the quality experienced with 3G, and better quality refers to the 4G experienc
freezing/disconnection while speaking and surfing the net on the move). We als
attribute for unrestrained use in Turkey and South Cyprus into two attributes, because
bilateral roaming regulation is required for each.
Table 1. Attributes and attribute levels
Attribute Number of levelsLevels
Mobile internet speed 3 Present speed (3 Mbps), 10 times faster
(30 Mbps), 100 times faster (300 Mbps)
Mobile internet limit 2 Limited/meter-rate, Unlimited
Quality (speaking and internet) 2 Present level, Better quality
Unrestrained use in Turkey 2 No, Yes
Unrestrained use in South Cyprus 2 No, Yes
Additional monthly subscription costa 4 20 TL, 40 TL, 60 TL, 80 TL
aAt the time of the survey (March 2015), 2.52 TL (Turkish lira) was approximately equal to 1 U
3.3 Experiment design and choice sets
There are various classes of design that can be used in a CE. The most general c
full factorial design, in which all possible combinations of the attribute levels are used

12
the decision-makers reveal their preferences for all possible combinations and the rese
able to fully observe their behaviour. However, as the number of attributes to be used
the size of a full factorial design grows exponentially. Using all possible treatment com
in this particular study would require us to ask (3×24×4)2 choice questions in total. Since this
would be impractical, we opt for an orthogonal main effects design, which reduces the
experiment and also maintains zero correlations between the attributes (Hensher et al
We generate our experiment design using the SPSS v20 software package
consists of 32 treatment combinations, or profiles, which we divide into 8 blocks so tha
survey respondent receives 4 choice questions. Although many previous studies in the
employed eight choice questions per respondent (Paulrud and Laitila, 2010; Kwa
2012), we believe we can eliminate effects of fatigue by limiting the survey to four que
also duplicate each version of the survey for the purposes of randomizing, by reversing
of the first two choice questions and the last two choice questions in each duplicate. T
minimize the effects of order bias in the study. We end up with 16 versions of the surv
with 4 choice questions.
Every choice question in the survey presents the respondent with two new
services (A and B) with varying levels of the attributes, and the respondent‘s e
service (C) to choose from. The respondent has the option to transfer to a new service
with the existing service. Because current technology in North Cyprus is 3G, an
local operators do not offer plans with unlimited internet or unrestrained use a
attribute levels for service C are fixed in every choice question. Table 2 displays a sam
question from the survey.