THE DIFFUSION OF

MOBILE PHONES IN INDIA

-
Dr. Sanjay K. Singh
Department of Humanities and Social Sciences
Indian Institute of Technology Kanpur
INDIA
Growth in telephone subscriber base in India

Mobile is becoming the dominant means for accessing

communications primarily because deploying mobile
network is not only more cost-efficient but also mobile
provides greater flexibility and convenience to its
subscribers than landline telephone.
Teledensity in India from 1995-96 to 2005-06

Growth in mobile-density has been phenomenal during the

last 5 years or so. Mobile-density in the country has
increased more than 23-fold from 0.35 in 2000-01 to 8.12 in
2005-06.
Growth in Mobile Subscriber Base in India

There has been 25-fold increase in mobile subscriber base in

a span of just five years from 2000-01 to 2005-06. During the
same period, mobile-density has increased more than 23-fold
from 0.35 in 2000-01 to 8.12 in 2005-06.
An effective management of mobile services requires an
understanding of the factors that underlie the evolution of
the market. Factors such as market potential and timing
and speed of adoption are of great importance for telecom
operators for capacity planning. Understanding the
evolution of mobile phone market and its likely future
trend is equally important for policy makers.

The main objective of this study is to analyze the

diffusion of mobile phones in India to inform the larger
discussion of managing the communication services as
well as to assist analysts concerned about assessing the
impact of public policies in the evolution of telecom
sector.
Estimation of the future trend and analysis of the
pattern and rate of adoption of mobile phones in India.
Spread of a successful innovation over time typically follows a
sigmoid or S-shaped curve. During an early phase of diffusion only a
few members of the social system adopt the innovation whereas, over
time, due to network consumption externality and dissemination of
information, many people opt for innovation as the diffusion process
unfold. Finally, during the maturing phase, the rate of diffusion goes
down when diffusion curve approaches a saturation level.

Therefore, it is hypothesized that the growth in mobile-density

over time follows a sigmoid curve. Among various functional
forms that can describe sigmoid curves (the logistic, Gompertz,
logarithmic logistic, log reciprocal, simple modified exponential,
etc.), the first two are the most widely used ones. Therefore, it
is decided to use these two functions to model and forecast the
development of mobile-density in India.
The logistic model can be written as:

Md t   t (1)
1   exp(  (time) t )

where Mdt is mobile-density (no. of mobile phones per 100

inhabitants), (time)t is value assigned to time at period t, 
is the saturation level and t is an error term.

All the parameters: ,  and  are positive.

Mdt ranges from a lower asymptote of 0 to the upper bound
 as time ranges from - to +. Maximum growth rate (=
/4) occurs when Mdt = /2 (i.e., at half of the saturation
level). Thus, the logistic curve is rotationally symmetric
about its inflection point (the point at which maximum rate
of diffusion takes place).
Similarly, the Gompertz model can be written as:

Md t   exp( exp(  (time) t ))   t (2)

where all the variables and parameters have their previous

meaning and t is an error term.

Again, all the parameters: ,  and  are positive. In this case,

maximum growth rate (= /e) occurs when Mdt = /e (i.e.,
at 37% of the saturation level).

These two models are estimated using non-linear least

square method once by assuming no restriction on the
saturation level and then by imposing restrictions on the
same. This is because there is no guarantee that the final
estimate of the saturation level, , will be close to the global
optimum (Heij C. et al., 2004).
The shape of logistic and Gompertz curves
120

100

Logistic (a=120, b=0.6, g=8000)

80
Logistic (a=120, b=0.6, g=5000)
Mobile-density

Logistic (a=120, b=0.5, g=8000)

60
Logistic (a=120, b=0.5, g=5000)

Gompertz (a=120, b=0.15, g=15)

40
Gompertz (a=120, b=0.15, g=20)

Gompertz (a=120, b=0.20, g=15)

20
Gompertz (a=120, b=0.20, g=20)

0
Time
15

22

29

36

43

50
1

8
 The saturation level of mobile-density for a country is likely to depend
on whether it is an early adopter or a late adopter of telephones. Early
adopters (developed countries) are expected to have lesser reliance on
mobile phones (due to high switching cost) whereas late adopters
(developing countries) are expected to have lesser reliance on main
line telephones (due to high infrastructure cost).
Teledensity and Percentage Share of Mobile in Selected Developed Countries
Analysis reveals that the saturation level of mobile share in developed
countries could be anywhere between 50% and 70% whereas the same
would be between 80% and 90% for the developing countries.
Assuming that the saturation level of teledensity could be anywhere
between 120 and 150 telephones per 100 inhabitants, the saturation
level of mobile-density in developing countries is likely to be between
100 and 120 mobile phones per 100 inhabitants.
Teledensity and Percentage Share of Mobile in Selected Developing Countries
Model estimation
Since India is a late adopter of telephones, its saturation
level of mobile-density is likely to be between 100 and 120
mobile phones per 100 inhabitants.
However, both logistic and Gompertz models are estimated
for six different saturation levels (70, 80, 90, 100, 110 and
120 mobile phones per 100 inhabitants) along with without
imposing any restriction on the same. The mean absolute
percentage error (MAPE) for the last three observations is
used to find out the most appropriate model and the
saturation level.
Annual data of mobile-density from 1995-96 to 2005-06 is
used for the estimation of the models.
Data on mobile subscriber base and mobile-density is taken
from Telecom Regulatory Authority of India (TRAI)
publications (www.trai.gov.in) and telecom sector database
from www.infraline.com.
Estimation results (with t-statistic in parentheses ): According to both R2 and
MAPE, the Gompertz models fit the data better than the logistic ones. As
expected, final estimate of the saturation level in the no restriction model does
not seem to be globally optimal. It seems that the Gompertz model with the
saturation level of 120 mobile phones per 100 inhabitants is the best model to
depict the diffusion of mobile phones in India.
Model Estimate

No restriction on the saturation level

Logistic (1)  = 18.9 (3.8),  = 0.7347 (11.0),  = 4331.2 (2.9); R2 = 0.997; Adj. R2 = 0.997; MAPE = 5.43
Gompertz (2)  = 217.9 (0.7),  = 0.1392 (3.1),  = 15.2 (8.7); R2 = 0.997; Adj. R2 = 0.997; MAPE = 4.70
Saturation level,  = 70
Logistic (1)  = 0.5970 (26.9),  = 5361.6 (4.3); R2 = 0.995; Adj. R2 = 0.996; MAPE = 5.12
Gompertz (2)  = 0.1954 (32.1),  = 18.6 (16.0); R2 = 0.997; Adj. R2 = 0.997; MAPE = 5.10
Saturation level,  = 80
Logistic (1)  = 0.5921 (26.6),  = 5895.6 (4.3); R2 = 0.995; Adj. R2 = 0.995; MAPE = 5.22
Gompertz (2)  = 0.1865 (32.6),  = 17.9 (17.0); R2 = 0.997; Adj. R2 = 0.997; MAPE = 5.04
Saturation level,  = 90
Logistic (1)  = 0.5883 (26.3),  = 6440.0 (4.3); R2 = 0.995; Adj. R2 = 0.995; MAPE = 5.30
Gompertz (2)  = 0.1793 (33.0),  = 17.4 (17.9); R2 = 0.997; Adj. R2 = 0.997; MAPE = 4.99
Saturation level,  = 100
Logistic (1)  = 0.5853 (26.1),  = 6991.4 (4.2); R2 = 0.995; Adj. R2 = 0.995; MAPE = 5.37
Gompertz (2)  = 0.1733 (33.2),  = 17.0 (18.6); R2 = 0.997; Adj. R2 = 0.997; MAPE = 4.94
Saturation level,  = 110
Logistic (1)  = 0.5828 (26.0),  = 7547.6 (4.2); R2 = 0.995; Adj. R2 = 0.995; MAPE = 5.42
Gompertz (2)  = 0.1683 (33.4),  = 16.6 (19.3); R2 = 0.997; Adj. R2 = 0.997; MAPE = 4.91
Saturation level,  = 120
Logistic (1)  = 0.5808 (25.8),  = 8107.2 (4.2); R2 = 0.995; Adj. R2 = 0.995; MAPE = 5.46
Gompertz (2)  = 0.1639 (33.6),  = 16.4 (19.9); R2 = 0.997; Adj. R2 = 0.997; MAPE = 4.88
Assumptions and Projections of Mobile-density in India
Further analysis will primarily be based on the estimated Gompertz
model at saturation level of 120 mobile phones per 100 inhabitants:
0.1639(time)
16.4e
Md  120e
Rate of growth of mobile-density
The analysis reveals that the inflection point (the maximum growth
rate point) of the curve will occur between 2011-12 and 2012-13
(when mobile-density will be around 45). During the year 2015-16,
there will be 71 mobile phones for 100 people in the country. Analysis
show that the no. of mobile phones will exceed the no. of people in
the country by 2022-23.
80

70
71.0
No. of mobile phones per 100 inhabitants

60 64.7
57.9
50
50.9
40 43.7
36.5
30

20

10
0.4 8.1
0
2000-01 2005-06 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16
Future Mobile Subscriber Base in India
It is projected that almost 350 million new mobile subscribers will
be added between 2005-06 and 2010-11 and more than 450 million
will be added between 2010-11 and 2015-16.

1000
899
900
808
No. of mobile subscribers (in million)

800
715
700
620
600 526
500 433
400

300

200
90
100
4
0
2000-01 2005-06 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16

Note: Future population of India is taken from the United Nations Population
Division publication.
Estimates of revenues collected by mobile operators
and the government
•Mobile operators’ revenue depends on ARPU and no. of subscribers
•Assuming that the ARPU will stabilize at around Rs. 300 per month
by the year 2010-11, mobile operators’ revenues during the year
2010-11 and 2015-16 have been estimated
Average Revenue per Mobile User per Month in India
Estimates of Mobile Operators’ Revenue
No. of Mobile Revenues GDP Mobile
mobile ARPU per from (Rs. in billion at revenue as a
subscribers year mobile factor cost at percentage of
(in million) (Rs.) services current prices) GDP
(Rs. in
billion)
2005-06 90 4500 405 32000 1.3
2010-11 433 3600 1559 57600 2.7
2015-16 899 3600 3236 103680 3.1

Rapid increase in mobile subscriber base and mobile spending

will have equally important implications for the government revenue
particularly in the form of regulatory charges (license fee including
universal service obligation and spectrum charges) and service tax.
Estimates of the Government’s Revenue
Presently, on an average, annual direct regulatory charges
faced by the operators in India is around 13% [far more than
that in Pakistan (4.5%), Sri Lanka (0.3%), Malaysia (6.5%),
and South Africa (5%)]. If we include the education cess of 2%
(of 12%), service tax burden on the sector would be 12.24%
from the financial year 2006-07 onwards.

Estimates of the Government’s Revenue

Rate of Revenue from Rate of Revenue from Total revenue
regulatory regulatory service tax service tax (Rs. in billion
charges charges (excluding (Rs. in billion at at current
(%) (Rs. in billion education current prices) prices)
at current cess of 2%)
prices) (%)
2005-06 13 53 10 41 94
2010-11 10 156 12 187 343
2015-16 10 324 12 388 712
Concluding Remarks
In this study, the growth of the mobile phone and mobile-
density in India has been analyzed using S-shaped growth curve
models.

The result shows that the Gompertz model adequately describes

the path of mobile phone diffusion in India.

The analysis shows that the high growth phase of the diffusion
of mobile phones will continue till 2012-13.

It is estimated that there will be 71 mobile phones per 100

inhabitants in India at the end of year 2015-16. The number of
mobile phones will exceed the number of people in the country
by 2022-23.

Total mobile phone demand is projected to increase from 90

million in 2005-06 to 433 million in 2010-11 and nearly 900
million in 2015-16.
Concluding Remarks ….
Rapid growth in mobile subscriber base in the India will have
important implications for revenues collected by the operators
and the government.

Revenue collected by the mobile operators is projected to

increase from Rs. 405 billion (1.3% of GDP) in 2005-06 to Rs.
1559 billion (2.7% of GDP) in 2010-11 and Rs. 3236 billion
(3.1% of GDP) in 2015-16.

The government’s revenue from regulatory charges and service

tax will increase substantially due to rapid increase in
operators’ revenue.

Revenue from regulatory charges is expected to increase from

Rs. 53 billion in 2005-06 to Rs. 156 billion in 2010-11 and Rs.
324 billion in 2015-16. Revenue from service tax is projected
to increase from Rs. 41 billion in 2005-06 to Rs. 187 billion in
2010-11 and Rs. 388 billion in 2015-16.
Concluding Remarks ….
It is quite likely that the rapid expansion of mobile services will
provide economic, logistic and strategic challenges to the
operators.
As operators expand coverage into urban, semi-urban, and rural
areas, they will be confronted with the daunting tasks of
developing a countrywide infrastructure and improving and
maintaining the quality of service.
Mobile operators should be ready with contingency plans to
deploy and operate infrastructure including customer care,
billing, applications, etc., faster than that they might have
initially planned.
Infrastructure providers, handset suppliers, and vendors should
also be geared up to respond to such plans.
THANKS