“Forecasting and analysing the characteristics of 3G and 4G mobile broadband diffusion in India: A comparative evaluation of Bass, Norton-Bass, Gompertz, and logistic growth models”
Introduction
The diffusion process of an innovation helps us understand the dynamics of spread and adoption of the innovation by the members, known as the potential adopters, of a given social system (Rogers, 2010). The study of diffusion phenomenon, specifically, seeks to explain how, why, and at what rate innovations spread. An initial theoretical conceptualization of the diffusion process was popularized during the 1960s by Everett Rogers in his seminal work titled Diffusion of Innovations (DOI) (Rogers, 2010). The theoretical foundations put forth by Rogers have proven very useful to several mathematical models – known popularly as the growth models – attempting to quantify the diffusion characteristics of innovations. Given sufficient historical data of adoption (where diffusion implies cumulative adoption), the growth models can estimate various parameters belonging to the individual aspects of the diffusion process with the help of linear and non-linear regression techniques (Bass, 1969; Franses, 1994a,b; Meade and Islam, 1995). Such estimated parameters indicate various technological and behavioral undercurrents driving the overall diffusion process; these include word-of-mouth, consumer interactions, signaling and interpersonal communications, social networks, positive externalities and the role of advertising and marketing (Peres et al., 2010). The same parameters are also very much useful to help forecast the future adoption of the innovation, thereby determining its ultimate market potential and time to saturation. The speed of diffusion, as determined by the growth models, can also be utilized to uncover the influence of various environmental factors on the diffusion process. Additionally, extensions of some of these models can explain the inter-generational dynamics of diffusion of multiple generations of the same innovation that overlap in time (Norton and Bass, 1987; Tseng et al., 2014).
Over the years, the growth models have proven their applicability across innovations belonging to several domains, ranging from agricultural sciences (e.g., hybrid corn), corporate finance (e.g., financial investments), marketing (e.g., consumer durable goods) to those in the Information and Communication Technologies (ICT) arena (e.g., IBM Mainframes, IPTV etc.) (Fareena et al., 1990; Lee et al., 2015). Similarly, in case of the innovations in the wireless communications domain, a vast majority of the literature has remained focused around the diffusion and forecasting of early mobile telephony in various nations (Gruber, 2001; Gruber and Verboven, 2001; Frank, 2004; Wareham et al., 2004; Massini, 2004; Koski and Kretschmer, 2005; Rouvinen, 2006; Lee and Cho, 2007; Gamboa and Otero, 2009; Hwang et al., 2009; Liu et al., 2012; Gupta and Jain, 2012; Yamakawa et al., 2013; Sultanov et al., 2016). However, the early mobile telephony systems, known today as the second generation (2G) of mobile services, have undergone multiple paradigm shifts resulting in the third (3G) and subsequently the fourth (4G) generations of Mobile Broadband Services (MBSs). In MBSs, 3G and 4G represent the evolution in access technologies and cellular architectures, resulting in higher data rates, lower latencies and the availability of much higher transmission bandwidths (ITU 2016b; ITU and UNESCO, 2014; Jha and Saha, 2018). The MBSs have also been termed as “the fastest growing ICT in human history” (ITU, 2016a), with the total number of worldwide subscriptions likely to surpass 3 billion by the end of the year 2019 (GSMA, 2018).
There are many socio-economic implications of MBSs diffusion for a country. Already, previous literature has established that enabling access to broadband services leads to numerous direct and indirect benefits on a country's Gross Domestic Product (GDP) through various positive externalities, the creation of consumer surplus and improvement in firm efficiencies (ITU 2016a; Katz, 2012; Qiang, 2010). The use of broadband also helps in achieving goals that are inclusive and sustainable – provided there be greater accessibility – especially in the developing world that comprises the major share of the world's overall population (ITU, 2016a). The provisioning of such accessibility is also crucial for governments in order to strengthen their public-sector capabilities and promote inclusivity among its citizens. Through accessibility and requisite capability1 for the subsequent usage of broadband, the citizens can reap the benefits of digital dividend generated across the world (World Bank, 2016). Therefore, a deeper understanding of various aspects of the diffusion process of the MBSs, including scenarios of their future adoption as well as the role of various environmental and policy variables, will help inform the stakeholders, namely the mobile network operators (MNOs) and the policymakers, in formulating their future strategies. Specifically, studies involving multiple generations of MBSs will help the MNOs estimate a priori the future demand, pertaining to each mobile service generation, and accordingly undertake the required network-rollouts/phase-outs based on the demand. For policymakers, this will translate into developing newer regulatory approaches that can help the MNOs transition to latest mobile service generations seamlessly, while ensuring healthy competition amongst them. The diffusion analysis can also assess the relative success/failure of prior policy initiatives targeted at promoting the adoption of MBSs, apart from assessing the impact of other environmental variables on the overall diffusion process.
The studies pertaining to the dynamics of diffusion and forecasting of MBSs (typically 3G and 4G) in different parts of the world are, therefore, getting traction among the researchers (Chu and Pan, 2008; Yates et al., 2013; Lim et al., 2012; Shin et al., 2015). Our study attempts to bolster the said initiative, by taking the case of India, for analyzing various aspects of MBSs diffusion in the country. India has the second largest telecommunications network in the world (in terms of volume of telephone users), as well as one of the lowest call tariffs in the world owing to a very high level of competition amongst the MNOs (TRAI, 2016). However, what is intriguing is that the overall (i.e., wired and wireless combined) broadband2 penetration in the country stands at a meager 13.7% of its total population (as of the year 2016) – much lower than its overall teledensity of 86% (TRAI, 2016). This is notwithstanding the fact that the 3G services were already in place by the year 2008 in India while 4G services were deployed in the year 2012. Also, the Department of Telecommunications (DoT) of the Government of India had introduced an updated National Telecom Policy (NTP) in 2012, with a view to accentuate the uptake of broadband in the country. The two available studies on the mobile telephony diffusion in India have each catered to the period 1995–2006 (Singh, 2008) and 1998–2009 (Gupta and Jain, 2012). Singh (Singh, 2008) in his study explained the implications likely to arise due to the change in the overall mobile density in India during the period 1995–2006. Gupta (Gupta and Jain, 2012) analyzed the diffusion of 2G services during the period 1998–2009 and evaluated the effects of competition, regulation, and legacy technology (wired landline) on the speed of diffusion. To the best of our knowledge, no existing literature has undertaken the diffusion analysis of 3G and 4G services in India.
The prior studies on diffusion have established that, given the same innovation, the best-fit growth model may still vary across countries due to the unique country-specific variables affecting the diffusion process (Meade and Islam, 2002; Meade and Islam, 2006). Therefore, for a given dataset of adoption of an innovation, multiple growth models need to be tested first for estimating the respective diffusion parameters, and then the best model needs to be determined by a statistical comparison of the model-fitness indicators. In the majority of the works mentioned above on diffusion of mobile services, three common growth models, namely Bass, Gompertz and Logistic, were utilized for fitting the historical data of adoption for each country. The best fit was then determined through evaluating and comparing various statistical measures of model robustness and goodness-of-fit. Similarly, for forecasting too, the most accurate model needs to be determined through a comparative evaluation of the statistical measures of forecasting accuracy. The parameter estimates of the best-suited model were then used to forecast the future adoption of the innovation. Given this backdrop, we have also followed an identical approach. Thus, firstly we analyze the diffusion process of MBSs (i.e., 3G and 4G) services in India with a view to forecasting the future adoption of these services and determining the time to saturation and the ultimate market potential of these services. Following the methodology adopted by prior studies, we too test the three commonly used growth models, namely Bass, Gompertz and Logistic, in a comparative manner, to achieve the above-stated objectives. Secondly, we analyze the inter-generational dynamics of diffusion of 2G, 3G and 4G services with the help of the multigeneration diffusion model of Norton and Bass (NB) (Norton and Bass, 1987). This is especially unique for our work because no previous works on India have ever used this model in MBS adoption and forecasting. Thirdly and finally, we evaluate the relative impact of governmental policies, prevailing economic environment, technological landscape and other market dynamics, on the diffusion of MBSs in India. Although our present study focuses on analyzing the factors of MBS adoption in India only, the findings may apply to many other developing nations in South-East Asia, Africa, and Latin America, wherever the regulatory and market scenario is similar to that in India. Our results may also be useful to the policymakers in those countries too. In that sense, our work bears a generic flavor too.
The remainder of this paper is structured as follows. Section 2 provides an overview of the background and related literature and therefrom draws upon our research objectives. In Section 3, we provide the theoretical overview of the chosen models. In Section 4, we highlight our research methodology, providing detailed explanations of various steps undertaken. Section 5 summarizes the results obtained from our analyses. In Section 6, we discuss the implications of the results for both theory and praxis. In Section 7, we summarize the main conclusions of the work. Finally, Section 8 provides the limitations of the study and the scope for future works in this area.
Section snippets
A brief overview of the history of mobile broadband services (MBSs)
The story of MBSs in the form we understand it today began with the first pre-commercial launch of 3G services by NTT DoCoMo in Japan in 1998. The first commercial launch of 3G services was also by NTT DoCoMo in October 2001. In Europe, the first 3G services were UMTS based, and the pre-commercial and commercial launches were by British Telecom and Telenor respectively in the year 2001. The 3G services were subsequently launched by SK Telecom and KT, during the year 2002, in South Korea. In the
Growth models of diffusion of innovations (DOI)
In this section, we briefly describe the theoretical underpinnings of the four chosen models of diffusion, namely Bass, Gompertz, Logistic and Norton Bass (NB) models. As already mentioned, Bass, Gompertz and Logistic are the most frequently applied models for the diffusion analysis of ICT innovations (Sultanov et al., 2016; Ovando et al., 2015; Naseri and Elliott, 2013; Turk and Trkman, 2012; Liu et al., 2012; Gupta and Jain, 2012). Apart from those, the literature on multigeneration diffusion
Methodology
Our analysis methodology consists of five stages as shown in Fig. 3. The detailed steps under each stage have been mentioned in the sections below.
Analysis
We use the historical data (quarterly) of 2G, 3G and 4G adoption in India for all the subsequent analyses pertaining to the diffusion model-based estimation and forecasting. The data is collected from multiple sources; namely the BMI Research database, the quarterly Performance Indicator reports and the Wireless Subscription reports released by TRAI.4 The
Diffusion analysis and adoption forecasts of MBSs in India
Tables 15 and 16 summarizes the results of the model parameter estimation and forecasting related analysis for all the three generations of mobile services (2G, 3G, and 4G) as well as MBSs (i.e., 3G+4G) in India. The parameter estimates for 2G (Bass, Gompertz, and Logistic) have been sourced from the previous study on the diffusion of mobile telephony in India (Gupta and Jain, 2012), for comparative purposes only.
Based on our extensive analyses and the results generated thereby in the previous
Conclusions
This paper utilizes four different growth models, namely Bass, Gompertz, Logistic and Norton-Bass (NB), in order to: a) determine the best-suited model for explaining the countrywide diffusion of 3G and 4G services in India, b) analyze the multigeneration diffusion aspects of 2G, 3G and 4G services in India, and c) determine the best-suited model for forecasting the future (years 2016–2026) adoption of 3G and 4G services in India. Additionally, the study also evaluates the relationships between
Limitations and future work
Considering that this is one of the initial studies on the diffusion of MBSs in India, there are several limitations which can be addressed in the future diffusion studies. Firstly, owing to the short history of MBS adoption in India, especially 4G services, some of the models could not yield the estimates of diffusion parameters. Also, since the growth models are highly sensitive to the innovation type and the size of available data, it remains to be seen whether the findings of this study
Acknowledgments
We would like to thank the Editor-in-Chief, the Associate Editor and the anonymous reviewers, who all provided many valuable comments that helped us in improving the paper.
Ashutosh Jha is a doctoral candidate in the Management Information Systems (MIS) group at the Indian Institute of Management (IIM) Calcutta. He currently teaches at the S.P. Jain Institute of Management and Research (SPJIMR), Mumbai. His research interests are primarily on the adoption and diffusion of next-generation mobile networks and services, techno-economics of mobile networks, radio spectrum management and telecommunication policy. He has completed his B.E. (Hons.) from BITS-Pilani, Goa
References (64)
Technological innovations and 3G mobile phone diffusion: lessons learned from Japan
Telemat. Informatics
(2010)- et al.
A flexible logistic growth model with applications in telecommunications
International Journal of Forecasting
(1988) - et al.
The diffusion of cellular phones in Portugal
Telecomm. Policy
(2004) - et al.
On the combination of exponential smoothing and diffusion forecasts: An application to broadband diffusion in the OECD area
Technological Forecasting and Social Change
(2011) - et al.
The forecasting of the mobile internet in Taiwan by diffusion model
Technol. Forecast. Soc. Change
(2008) An analysis of the effect of the economic situation on modeling and forecasting the diffusion of wireless communications in Finland
Technol. Forecast. Soc. Change
(2004)A method to select between Gompertz and logistic trend curves
Technol. Forecast. Soc. Change
(1994)- et al.
An estimation of the pattern of diffusion of mobile phones: the case of Colombia
Telecomm. Policy
(2009) Competition and innovation: The diffusion of mobile telecommunications in Central and Eastern Europe
Information Economics and Policy
(2001)- et al.
The di!usion of mobile telecommunications services in the European Union
European Economic Review
(2001)
Diffusion of mobile telephony in India: an empirical study
Technol. Forecast. Soc. Change
Investigation of factors affecting the diffusion of mobile telephone services: an empirical analysis for Vietnam
Telecomm. Policy
Determinants of IPTV diffusion
Telemat. Informatics
An empirical analysis of fixed and mobile broadband diffusion
Inf. Econ. Policy
Forecasting 3G mobile subscription in China : a study based on stochastic frontier analysis and a Bass diffusion model
Telecomm. Policy
Modelling and forecasting the diffusion of innovation - A 25-year review
Int. J. Forecast.
Forecasting with growth curves: an empirical comparison
Int. J. Forecast.
Diffusion models of mobile telephony in Greece
Telecomm. Policy
LTE techno-economic assessment: The case of rural areas in Spain
Telecommunications Policy
Innovation diffusion and new product growth models: a critical review and research directions
Int. J. Res. Mark.
Diffusion of digital mobile telephony: are developing countries different?
Telecomm. Policy
The diffusion of mobile phones in India
Telecomm. Policy
The diffusion of mobile telephony in Kazakhstan: an empirical analysis
Technol. Forecast. Soc. Change
An integrated model for analyzing the development of the 4G telecommunications market in Taiwan
Telecomm. Policy
Bass model estimates for broadband diffusion in European countries
Technological Forecasting and Social Change
DRAMs, fiber and energy compared with three, models of market penetration
Technol. Forecast. Soc. Change
DRAMs as model organisms for study of technological evolution
Technol. Forecast. Soc. Change
Wireless diffusion and mobile computing: implications for the digital divide
Telecomm. Policy
Reprint of: competition and regulation in China’s 3G/4G mobile communications industry - Institutions, governance, and telecom SOEs
Telecomm. Policy
The diffusion of mobile telephones: an empirical analysis for Peru
Telecomm. Policy
A new product growth for model consumer durables
Manage. Sci.
Cited by (35)
Urban wireless traffic evolution: The role of new devices and the effect of policy
2023, Telecommunications PolicyEvaluating and forecasting the niche fitness of regional innovation ecosystems: A comparative evaluation of different optimized grey models
2023, Technological Forecasting and Social ChangeHow to promote the diffusion of green behavior among contractors? Analysis and simulation using the SIR model
2023, Journal of Environmental Management
Ashutosh Jha is a doctoral candidate in the Management Information Systems (MIS) group at the Indian Institute of Management (IIM) Calcutta. He currently teaches at the S.P. Jain Institute of Management and Research (SPJIMR), Mumbai. His research interests are primarily on the adoption and diffusion of next-generation mobile networks and services, techno-economics of mobile networks, radio spectrum management and telecommunication policy. He has completed his B.E. (Hons.) from BITS-Pilani, Goa Campus in Electrical and Electronics Engineering (EEE) discipline, and has also served in the IT/ITES industry for six years in various capacities. His publications have appeared in IIMB Management Review (IMR) Journal, Decision Journal, International Conference on Information Systems (ICIS), European Conference on Information Systems (ECIS), Hawaii International Conference on System Sciences (HICSS), International Association for Management of Technology (IAMOT), and IEEE, among others.
Debashis Saha is a Full Professor in the MIS Group of Indian Institute of Management (IIM) Calcutta. Previously, he was with CSE Dept. of Jadavpur University, Kolkata. He has co-supervised 14 doctoral theses and published about 280 research papers, and directed four funded projects on telecom. He has coauthored five books including Networking Infrastructure for Pervasive Computing: Enabling Technologies and Systems (Norwell, MA: Kluwer, 2002) and Location Management and Routing in Mobile Wireless Networks (Boston, MA: Artech House, 2003). He was the Co-Editor-in-Chief of the International Journal of Business Data Communications & Networking (IJBDCN) [2009–2013].