Keywords

1 Introduction

Diffusion of Mobile Communication Technology (MCT)Footnote 1 - a distinct type of Information and Communication Technology (ICT) – remains an important research theme for social scientists the world over. This is evident in the substantial body of literature dedicated to exploring the various facets of such diffusion of the MCT innovation [14]. To date, there have been five successive generations (Gs) of MCTs, commercialized by the Mobile Network Operators (MNOs) globally, with each generation separated roughly by a decade in time from the other (please refer to Annexure A, Fig. 3 for details). Each subsequent generation has demonstrated enhanced capabilities in terms of data exchange speed (data rate), real-time delivery (latency), and traffic carrying capacity (network bandwidth), to name a few. It is worth noting that the successive generations of the MCT innovation, namely 2G, 3G, 4G, and 5G, have independently attracted significant research attention, which is not surprising considering the increasing scale and scope of socio-economic implications associated with the diffusion of these MCT generations [12].

The wider literature focused on analyzing the diffusion of “newer products and services in the market” primarily originates from the discipline of marketing [1, 11, 22]. These studies are predominantly quantitative in nature - the methodology adopted being that of the estimation of model parameters using non-linear regression technique. While there are several models of diffusion used in the aforementioned literature, the Bass model and its extensions have garnered the highest popularity so far [2, 18]. An important advantage of using Bass model (and the various extensions) is the relaxation of the criteria concerning the need for external variables to model the diffusion phenomena, notwithstanding the simplicity of these models. Notably, most of these diffusion models, including the Bass model have drawn inspiration from Evertt Rogers’ Diffusion of Innovations (DOI) theory, by way of their model parameterisation [25].

The extant literature on the diffusion of MCTs largely follows the approach undertaken in the wider marketing literature, as mentioned above. Previous studies regarding the diffusion of MCTs have largely focused on analysing specific MCT generations, emphasizing quantitative assessment of the various diffusion characteristics, such as the “rate of diffusion”, “time to market saturation”, and the “ultimate market potential”, etc. [29]. Such prior literature has also analysed the impacts due to various macro-level factors and drivers of such MCT diffusion at the national level, proving very helpful to policymakers, especially in the developing and underdeveloped regions of the world [15, 16]. An extensive review of the extant literature, however, points to the following limitations and possibilities for further research endeavours. Firstly, we find that, to the best of our knowledge the extant literature lacks in the diffusion analysis of recent MCT generations, namely 4G and 5G diffusion, especially in the context of the emerging market economy. This could be owing to the lack of secondary data, more so given that the rollout of 5G services have only begun recently for many countries. Secondly, extant literature on the MCT diffusion have taken a comprehensive view of the demand-side, without accounting for the existing heterogeneities owing to variations in economic conditions (aka willingness-to-pay for mobile services), demography (rural vs. urban), geography, e-literacy levels, etc. [30]. Finally, prior literature is silent on the interplay regarding the effectiveness of “communication channels” (formal and informal) in the diffusion process, and the socio-economic conditions prevailing in the market.

In light of the gaps in prior literature and the opportunity provided thereof, we posit the following research question for investigation in the current study: To analyse the dynamics regarding the diffusion of 4G and 5G MCTs in an emerging market economy, by accounting for the demand-side heterogeneities. We explain further below.

For our investigation, we take the case of the mobile services market in India. The rationale is explained as follows. The telecom market structure in India is unique, as India is divided into twenty-two administrative zones - with each zone referred to as a Telecom Circle (TC) - for the targeted implementation of telecom policies in the country. These twenty-two TCs are further grouped into four categories, namely “Metros, “A”. “B”, and “C” categories. These respective categories indicate (in decreasing order) the potential for revenue generation for MNOs, which is in turn used for valuation of the reserve prices (base prices) of the radio spectrum. Therefore, Metro TCs signify greater potential for financial returns compared to category-A TCs and so forth. Such hierarchy in the “attractiveness” of TCs mimics the order of socio-economic development in these TCs, with category-A TCs faring better on the development indicators than category-B, and so forth. For MNOs, such heterogeneity in TCs translates into the varying potential for financial returns, which has considerable bearing on the decision to undertake capital investments and deploy Mobile Network Infrastructures in the TC. In addition, due to such heterogeneity, the potential revenue to the government resulting from the allocation of critical resources, such as radio spectrum, is also affected. This necessitates differentiated policies and regulatory imperatives for different regions [30]. To the best of our knowledge, studies on the diffusion dynamics of 4G and 5G MCTs, in the presence of such heterogeneous market conditions have not been conducted in the literature and could be useful.

We realise that analysing the diffusion of specific MCT generation is challenging due to various reasons. For example, there is lack of availability of longitudinal data regarding number of subscribers for the specific MCT generation. Moreover, MNOs rarely provision services using a single MCT generation, combining instead multiple MCT generations together. For example, 2G, 3G and 4G MCTs could all coexist in the market at a given point in time. Therefore, independent diffusion analyses of specific MCT generation would pose problems of accuracy. Furthermore, granularity of data is an important concern in such scenarios, as too few data separated widely across time, could lead to erroneous conclusions. Furthermore, the number of data points concerning the diffusion of 5G may not sufficient, in the statistical sense, especially in the emerging market context. We overcome these challenges through the following approach.

For investigating the stated research objective, we take the unique case of the market entry of the “4G only” MNO, Reliance Jio Inc (henceforth Jio) in the year 2016 in India. A new entrant in the Indian telecom market, Jio launched its 4G services simultaneously, across the twenty-two TCs in India. This could be argued to be a unique market intervention, which presents the opportunity to analyse the diffusion dynamics of an important innovation (viz., 4G MCT based on the Voice-over-Long-Term-Evolution (VoLTE) standard), in the “pure” greenfield scenario. We restrict our focus on analysing the “early-diffusion” of Jio’s 4G services, by making use of a micro time-series (monthly) dataset, which comprises the total number of 4G subscribers, across the twenty-two circles in India, during September 2016 to March 2019. We further restrict our focus on the pre-Covid era, to rule out the confounding impacts due to the pandemic on the diffusion process. In line with the literature, we follow the non-linear least-squares (NLS) regression methodology for estimating the diffusion model parameters, using four different diffusion models, namely Bass, Gompertz, Logistic, and Weibull, in a comparative sense. To analyze our findings, we make use of two theoretical strands – Rogers’ DOI theory, and the Heterogenous Market Hypothesis (HMH) theory from the domain of finance [32]. Considering the space constraints, we have briefly explained these theoretical underpinnings in later sections.

The rest of the paper is organized as follows. Section 2 presents the background and literature review on the diffusion of 4G and 5G MCTs. Section 3 explains the various diffusion models chosen for analyses. Section 4 and 5, respectively, highlight the research framework, and data and methodology pertaining to our analysis. Section 6 provides the results and discussions from the analysis. Finally, Sect. 7 implications and concludes the work.

2 Background and Literature Review

2.1 Evolution in 4G and 5G MCT Generations

The standards for 4G MCT were specified by the International Telecommunication Union-Radio communications sector (ITU-R), in the form of the International Mobile Telecommunications Advanced (IMT-Advanced) specifications [13]. The key features included a peak theoretical data rate of 100 Mbps (megabits per second) and 1 Gbps (gigabits per second) for high and low mobility situations, respectively [4]. Long-Term Evolution (LTE) is the predominant mobile network standard that powers the 4G MCT, so much so that 4G is synonymous with LTE networks and gets commonly referred to as 4G-LTE. The demand for higher data rates and ubiquitous internet connectivity, driven by an increase in the use of social media, e-commerce, and over-the-top (OTT) services, has led to a steady migration of existing 2G and 3G subscribers to 4G the world over. As a result, we are observing a rapid diffusion of 4G MCT worldwide, although different countries are at different stages of such diffusion. Figure 3 provides a snapshot of the evolution of various MCT generations,

A variant of LTE, the Voice-over-LTE (the technology used by Jio, the case in point), was first launched during the year 2012 in the United States (US). Since then, the VoLTE network has become quite popular amongst MNOs the world over. For example, over 253 MNOs across 113 countries have either already deployed VoLTE or are on their way to such deployment (as of February 2019) [8]. VoLTE deployments have also proven quite successful in terms of their market adoption by mobile subscribers the world over. This is especially true for the case of growth markets in the developing world, such as India and China. For example, within three years (2016–2019) of the launch of VoLTE in India by the new-entrant MNO, Jio, a total of 300 million mobile subscribers were added. This also resulted in Jio acquiring the major portion of the overall market share for 4G subscribers in India, which amounted to approximately 400 million subscribers as of February 2019 (the cutoff point of our analysis) [19]. Much of the subscription acquired by Jio is posited to be a result of the loss in the mobile subscriber base of the incumbent MNOs in India. Similarly, in China, the VoLTE mobile subscriber base of MNOs, such as China Unicom and China Mobile, is witnessing a steady rise reportedly; China Mobile has deployed VoLTE across 313 cities in the country, investing over $66 billion in the process [24]. Similar examples can be cited for various other countries as well.

The fifth generation (5G) of MCT, which is the latest addition to the MCT family, is a significant leap on all the important parameters of MCT performance, namely, data rate, latency, and network bandwidth, and is being hailed as the “lifeblood for the business of the future”, especially Industry 4.0 [3]. Such “transformative leap” is posited to have large scale implications for the growth of several emerging technologies, such as the Internet of things (IoT), Augmented and Virtual Reality (AR and VR) and Edge Computing, to name a few. 5G’s low latency and high bandwidth could also have significant impacts on healthcare through telemedicine, remote surgery, and real-time patient monitoring, etc. Currently, 5G comprises three main business use cases, namely Enhanced Mobile Broadband (eMBB), Ultra-Reliable Low-Latency Communications (URLLC), and Massive Machine-Type Communications (mMTC): eMBB leverages 5G’s high data speeds and capacity to deliver superior mobile internet experiences, enabling seamless high-definition video streaming, immersive augmented and virtual reality applications, and bandwidth-intensive content consumption; URLLC focuses on providing ultra-low latency and high reliability for critical applications, such as autonomous vehicles, remote surgery, and industrial automation; mMTC enables the connectivity of a massive number of IoT devices, facilitating smart city infrastructure, industrial IoT, and remote monitoring [23].

2.2 Prior Studies on Diffusion of MCT Generations

The quantitative perspective on diffusion is concerned with the dynamics of cumulative market adoption of innovation over time in a particular market. Thus, diffusion is a group-level phenomenon, which slowly manifests over time as the information about the innovation spreads in the market (e.g., newer network generations have been found to appear in the telecom market after every ten years). Diffusion studies are, therefore, temporal and rely on the longitudinal data of market adoption (total subscribers in our case) of the innovation. As mentioned earlier, prior literature on the diffusion of mobile innovations uses growth models, which parameterize the various characteristics of the diffusion process in non-linear functional forms [1, 17, 25]. To estimate the growth model parameters corresponding to various innovations, researchers take the help of suitable non-linear regression techniques for fitting the functional expressions of these growth models with the historical data of sales of the innovation [28]. Bass, Gompertz, Logistic, and Weibull models are examples of the most widely used growth models, especially for mobile innovations. Such growth models have found their application in studies investigating the diffusion of 2G, 3G, and 4G networks the world over [9, 10, 31]. In addition to analyzing the diffusion of various network generations, the above-mentioned growth models have also found applications to innovations in several other domains – ranging from agricultural sciences (e.g. hybrid corn), corporate finance (e.g. financial investments), marketing (e.g. consumer durable goods), to several industrial innovations (e.g. IBM Mainframes and IPTV) [5].

2.3 Research Context

The research context in this paper comprises the market for mobile services in India. India is currently the world’s second-largest market for mobile services in terms of volume, having over 1.2 billion mobile subscribers as of June 2018 [30]. The mobile economy in India also contributes substantially to its overall Gross Domestic Product (GDP). India is also the world’s second-largest market in terms of mobile Internet subscribers. However, this was not always the case. At the end of the year 2016, the total Mobile Broadband (MoBro) subscriptions (3G + 4G) in the country stood at a little over 230 million (approximately 20% penetration). In September 2016, Jio, the new entrant, launched VoLTE services across the twenty-two circles in the country. Jio was the first 4G-only MNO, which achieved pan-India 4G coverage, enabled by an extensive VoLTE network deployment drive across the country. Notably, Jio rolled out its services for free during the initial six months, beginning to charge its subscribers only after March 2017, adopting a low-price-based “growth-hacking” strategy all the while [26]. Jio acquired 16 million subscribers within the first month of its launch, creating a world record for the fastest ramp-up by any MNO in the world [7]. As of March 2019, the new entrant had already become the second-largest MNO in India and sixth largest MNO in the world, having over 306 million VoLTE subscribers. Notably, MoBro connectivity in the form of 3G and 4G was launched in India many years ago - 3G in 2009 and 4G in 2012, however, the incumbent MNOs were finding it difficult to increase their base of MoBro subscribers during these years. The entry of Jio changed this scenario. The incumbent MNOs witnessed significant attrition in their subscriber base, which led to the MNOs reducing tariffs for mobile data services. The resulting “price wars” led to an overall decline in the average revenue per user (ARPU) figures for both voice and mobile data services, resulting in reduced revenue margins for the incumbent MNOs. The disruption in the telecom market, brought in due to Jio’s entry, also precipitated the ongoing business restructuring and consolidation trends in India. Consequently, there remained only three private MNOs, namely Bharti Airtel (Airtel), Vodafone Idea (Vi), and Jio, in India. Bharat Sanchar Nigam Limited (BSNL) and Mahanagar Telephone Nigam Limited (MTNL) are the other two state-owned MNOs in the country. Figure 4 in Annexure highlights the trend of mobile subscriptions for the majority of MNOs in India.

3 Research Framework

Our research framework is summarized in Fig. 1. Notably, we are motivated by two main theoretical strands for explaining our analysis: (a) the Diffusion of Innovations (DOI) of Rogers, and (b) the Heterogenous Market Hypothesis (HMH) theory from the discipline of Finance. We explain the details below.

The DOI theory was first proposed by Everett Rogers in 1962, and it has since become one of the most widely used theories in management [25]. The theory seeks to explain how new ideas, products, or technologies spread through a social system over time. In addition to identifying various stages in the diffusion process, Rogers also identified five categories of adopters: innovators, early adopters, early majority, late majority, and laggards [25]. Innovators are the first to adopt an innovation, while laggards are the last. Early adopters are opinion leaders who adopt the innovation early on and help to spread it to others. The early and late majority adopt the innovation after it has been tested and proven successful by the innovators and early adopters. The diffusion of innovation theory has been applied to a wide range of fields, including technology adoption, marketing, and health. By understanding the factors that influence the adoption of new ideas or technologies, researchers and practitioners can develop strategies to promote their adoption and accelerate the diffusion process.

Fig. 1.
figure 1

Research Framework

Full size image

The Heterogenous Market Hypothesis (HMH) theory suggests that market participants may have different beliefs and expectations about future asset prices, which can lead to price fluctuations and volatility in the market [32]. According to HEH, investors have varying levels of information, analysis, and interpretation of market data, leading to different expectations about future events and outcomes [32]. With specific reference to HMH, while the theoretical underpinnings do not directly apply to the domain/phenomena under consideration in this study, the rationale and the core elements of the theory seem to be useful in explaining the interplay regarding the perceptions’ of the twenty-two TCs for the supply-side stakeholders (MNOs and government); considering the heterogeneity across these TCs across various factors, such as the distribution of adopter categories, socio-economic conditions of general populace, market structures, geographic conditions, demography, usage patterns and the overall E-literacy levels.

4 Theoretical Overview of Diffusion Models

4.1 Bass Model

The probability density and the cumulative distribution function equations for the Bass model can be expressed, respectively, as:

$$n \left(t\right)=(p+q\, N(t))(1-N(t))$$
(1)
$$N\left(t\right)=M\left[\frac{1-{e}^{-\left(p+q\right)t}}{1+\left(\frac{q}{p}\right){e}^{-\left(p+q\right)t}}\right]$$
(2)

where \(n\left(t\right)\) is the noncumulative number of adopters at time t, \(N(t)\) is the cumulative number of adopters till time t, \(p\) is the coefficient of innovation, \(q\) is the coefficient of imitation, and \(M\) is the ultimate market potential of the innovation. The coefficient of innovation also corresponds to the probability of an initial purchase during the beginning of the product’s lifecycle and has direct relationships with the initial critical mass of the adopters, i.e., the innovators. This factor influences significantly the rest of the diffusion process involving the imitators, which constitutes the remaining population yet to adopt. Essentially, innovators are the earliest adopters who gather information about the new innovation through formal channels of communication, whereas imitators rely on informal sources, such as word-of-mouth, interpersonal interactions and direct observations of product use, in order to make the decision to adopt [21].

4.2 Gompertz Model

The probability density and the cumulative distribution function equations for the Gompertz model can be expressed, respectively, as:

$$n\left(t\right)= {b}_{1}\, ln\frac{K}{N(t)}$$
(3)
$$N\left(t\right)= K{e}^{-{e}^{-{b}_{1}(t-{b}_{2})}}$$
(4)

where \(n\left(t\right)\) and \(N\left(t\right)\) are the same as in the Bass model, \(K\) is the asymptote, which also corresponds to the ultimate market potential (similar to \(M\) in Bass model), \({b}_{1}>0\) is a scaling factor indicating the “intrinsic growth rate,” i.e., the “rate of diffusion,” \({b}_{2}\) is the offset in time scale, and the product \({b}_{1}{b}_{2}\) is related to the “point of inflection” [20]. Two main characteristics of Gompertz curve are the occurrence of the point of inflection before the point of saturation, and the rate of growth being always non-negative, even though it may exhibit a decrease over time [6]. Gompertz model was originally proposed by the mathematician Benjamin Gompertz for modeling human mortality in society and since then, has been used for fitting and forecasting the diffusion of technological innovations.

4.3 Logistic Model

The probability density and the cumulative distribution function equations for the Logistic model can be expressed, respectively, as:

$${n\left(t\right)= b}_{1}\left(1-\frac{N(t)}{K}\right)$$
(5)
$$N\left(t\right)=\left[\frac{K}{1+{e}^{-{b}_{1}(t-{b}_{2})}}\right]$$
(6)

where the notations have similar meanings as in the Gompertz model. Similar in nature to the Gompertz model, the Logistic model was proposed by the Belgian mathematician Pierre Francois Verhulst in 1838 and was also originally meant for demographic studies. The rationale behind the Logistic model is that the growth (referring to human population) slows down as the population approaches its uppermost limit, essentially due to the feedback limits on the system [3]. The point of inflection of the Logistic model-based diffusion curve lies about midway between the asymptotes (where it differs from the Gompertz model).

4.4 Weibull Model

Weibull model is found to effectively model the technological diffusion successfully in a variety of situations [27]. Unlike other models, Weibull provides flexibility in the skewness of the diffusion curve. The probability distribution function (cumulative p.d.f.) equation of the Weibull model can be expressed as:

$$ n\left( t \right) = \left( {\frac{\beta }{\alpha }} \right)\left( {\frac{t}{\alpha }} \right)^{\beta - 1} e^{ - ({t / \alpha })^\beta } $$
(7)
$$N\left(t\right)=K(1-{e}^{-{\left(\frac{t}{\alpha }\right)}^{\beta }})$$
(8)

where \(t\) is time; \(n(t)\), \(N(t)\), and \(K\) have similar interpretations as in the Gompertz model; and \(\alpha \) and \(\beta \) are constants. While both \(\alpha \) and \(\beta \) determine the steepness of the curve, \(\beta \) alone determines the shape of the curve. For estimating the model parameters, we use the transformed equation of the Weibull model [27].

5 Data and Methodology

We use twenty-two unique time-series micro-datasets (monthly), one corresponding to each circle in India, which comprises the 4G-VoLTE subscription in the respective TCs. We extract our datasets from the publicly available “Telecom Subscriptions Reports,” which is published monthly by the Telecom Regulatory Authority of India (TRAI). Such monthly data corresponds to the period from September 2016 to March 2019 (a total of 31 data points). We use R and Tableau applications for our data analysis and visualizations, respectively. Non-linear models, such as Bass model, which are non-linear in parameters, require a different approach for fitting the observational data unlike the linear models, where simple linear-regression techniques may serve the purpose. For non-linear models, the observational data are fitted to the model expression by following a method of “successive approximations,” which requires the use of a suitable non-linear regression technique, such as Non-linear Least-Squares (NLS) regression, Maximum Likelihood Estimation (MLE), or Bayesian approximation techniques. Of these available techniques, we use NLS regression for estimating our model parameters, which matches with the approach highlighted by [28]. Notably, when compared to the other techniques, the key advantage of NLS regression lies in obtaining valid standard error estimates of the model parameters.

6 Results and Discussion

Table 1 summarizes the results of the NLS regression-based estimation of diffusion parameters for the considered Bass model. The remaining results from other diffusion models are summarized in Table 2, Annexure A. We can infer the following from the diffusion analyses results summarized in Table 1.

6.1 The Formal vs. Informal Channels of Communication in MCT Diffusion

We find that, the Bass model parameter estimates for the coefficient of imitation (q) are much higher than those for the coefficient of innovation (p), across the twenty-two TCs. This could potentially indicate the relatively higher impact of informal channels of communication, such as word-of-mouth and interpersonal signalling, towards the subscribers’ decision to adopt the 4G MCT across the twenty-two TCs. This also signifies an overbearing presence of the potential adopters belonging to the “imitators” category, when compared to the “innovators” category, amongst the larger pool of 4G subscribers in India. However, given the market heterogeneities prevailing across the TCs, further analysis is warranted to understand the q vs. p scenarios in these TCs. Therefore, we further clustered the TCs using an unsupervised clustering technique, based on the p-q variations. Such analyses resulted in two main clusters displaying distinct characteristics in their “innovation-imitation dynamics”. Figure 2 highlights the clusters.

Table 1. Bass model parameter estimates of VoLTE diffusion in India’s Telecom Circles
Full size table

The Y- and X- axes in Fig. 2 represent the ‘coefficient of innovation (p)’ and the ‘coefficient of imitation (q)’ from the Bass model. The four distinct bubble sizes correspond to the four categories of TCs, as summarised in the legend in Fig. 2. An unsupervised clustering analysis clearly reveals two distinct clusters (Clusters 1 and 2), as shown in different colors, and on both sides of the solid line. The majority of TCs belonging to Cluster 1 are ‘Metros’ and ‘A’ category TCs, whereas those in Cluster 2 are ‘B’ and ‘C’ category TCs (except one TC in Cluster 2, namely Maharashtra, that is borderline and belongs to ‘A’ category).

Interestingly, the inferences from Fig. 2 are very counterintuitive. For example, we find that the majority of B and C category TCs, which are perceived to yield lower financial returns for MNOs, as reflected in the relatively lower base prices of the radio spectrum in these TCs, comprises high values of ‘coefficient of imitation’. This signals that the impacts due to informal channels of communications are much greater in the overall diffusion process. Notably, these TCs are also characterized by relatively lower per capital income of their populace, as compared to Metros and A categories. The same is true for other socioeconomic indicators of development, such as Health and Education. On the other hand, the power of innovation (in this case 4G MCT) is significant towards catalyzing the adoption of 4G MCT by the Innovators and Early adopters. Notably, such TCs fare better in terms of the various socioeconomic indicators of development. Thus, market heterogeneity seems to have a strong association with how the twin dimensions of diffusion, namely innovation power and interpersonal influences, are mobilized and activated. While we do not intend to ascertain causality amongst variables through our analyses, we could formulate propositions, based on the results from the early diffusion dynamics of 4G MCT in various parts of India. The propositions warrant investigation for the diffusion of 5G MCTs too.

Fig. 2.
figure 2

Coefficient of Innovation vs. Coefficient of Imitation plot for TCs

Full size image

We formulate our propositions as follows.

Proposition 1:

The early diffusion of 4G MCT is primarily propelled by informal communication channels, including word of mouth and interpersonal signaling, among others. The diffusion of 5G MCT is likely to follow a similar trend.

Proposition 2:

The impact of informal communication channels on the diffusion of 4G MCT is more notable in markets where the purchasing power is lower, in contrast to those with higher purchasing power. Such impacts are likely to persist for the diffusion of 5G MCT too.

Proposition 3:

In markets where purchasing power is higher, the effects of formal communication channels, such as advertising and mass media, on the diffusion of 4G MCT are more pronounced compared to those with lower purchasing power. Such differentiation in the effects due to formal communication channels are likely to persist for 5G MCT diffusion as well.

6.2 The Speed of MCT Diffusion

We find that the speed of 4G MCT diffusion (as assessed through the Logistic model parameters) is much higher in TCs that are otherwise lower in their attractiveness quotient from the point of view of the MNOs, considering the lower levels of socio-economic development. This is counterintuitive to the general assumption that Metros and category-A TCs, which comprises the large cities and metropolis, are characteristic of much greater demand for 4G MCT. Interestingly, such assumptions have led to the neglect of 4G mobile network infrastructures deployment activities in these regions in the past, given the prevalent assumption regarding such regions having very low revenue potential. This belief also shapes the valuation of the base prices in these TCs, and, therefore, could lead to a loss for the government during the spectrum auctions. While these are initial findings from a relatively smaller dataset, nevertheless, these findings need to be further investigated for understanding the changed service usage patterns of 4G MCT subscribers across various markets. Notably, such studies are also scant globally. Future research may cater to this aspect of 4G and 5G diffusion the world over.

We state our final proposition below.

Proposition 5:

In markets where purchasing power is relatively lower, the speed of diffusion of 4G MCTs could be substantially higher than the earlier MCT generations, considering the shift in demand and usage patterns. With the launch of 5G MCT, this trend could continue to persist.

We hold a strong belief that these propositions could offer valuable avenues for exploration in future research, especially when extended to global data sets. By gathering enough data, such investigations could offer greater understanding of the rural-urban dynamics in the spread of the next generation MCTs, viz., 4G and 5G. This could provide useful perspectives for policymakers and other stakeholders to fine-tune their approaches towards infrastructure implementation, pricing, and other regulatory actions.

7 Conclusion

This research delves into the initial diffusion patterns in the growth of 4G MCT within an emerging economy, namely India, offering valuable insights for the impending 5G MCT diffusion. Utilizing Everett Rogers’ Diffusion of Innovations theory and the Heterogeneous Market Hypotheses, the study examines the early 4G MCT diffusion by a “4G only” mobile network operator (MNO) across India’s twenty-two telecom circles (TCs). Employing nonlinear regression-based quantitative analysis, the study estimates diffusion model parameters across these TCs. Through a comprehensive exploration of diffusion dynamics amid socio-economic diversity across markets, the study presents several counterintuitive propositions based on its findings.