ESTIMATE DEMAND FUNCTION &

FORECAST DEMAND
FOR INTERNET & EMAIL SERVICE
 Contents

1 Introduction

2 Specifying the Demand Function

3 Regression & Estimation

4 Demand Forecasting

5 Conclusion & Discussion

 Introduction

Improvement in ICT literacy

Enables a Telco operator

Declining Cost of Computing • To plan the future activities
• To make decisions on investments

Changing Culture
 Demand Function

Telephone
subscribers
T • Dependent variable
• Quantity demand

Demand
• Independent variables
for • Telephone subscribers
Internet
& • CPI
Email
service • Government initiatives

Consumer’s
Government D I
Price Index
initiatives (CPI)
 Specifying the Demand Function

 Linear
Q=a+bT+cI+eD
 Log Linear
ln Q =ln f + g ln T + h ln I + j D
 Legend
 Q: Quantity demand
 T: Telephone subscribers
 I: Consumer’s Price Index
 D:
• 0: Year 1996 to Year 2001 and 1: Year 2002 to Year
2009
 Data
Number of Internet Number of Telephone Consumer Government ICT
Year
& Email subscribers line subscribers Price Index Initiatives
1996 2,504 255,049 1,907 0
1997 10,195 341,622 2,089 0
1998 18,984 523,529 2,284 0
1999 25,535 669,113 2,392 0
2000 40,497 767,411 2,540 0
2001 61,532 827,195 2,900 0
2002 73,468 883,108 3,176 1
2003 85,500 939,013 3,377 1
2004 93,444 991,239 3,632 1
2005 115,000 1,243,994 4,055 1
2006 130,000 1,884,076 4,356 1
2007 202,348 2,742,059 4,896 1
2008 234,000 3,446,411 5,747 1
2009 250,000 3,391,484 6,541 1

Sources:
 Regression Results
Linear
Model Summary
Model R R Square Adjusted R Std. Error of Durbin-
Square the Estimate Watson
1 .997a .993 .991 7727.634 2.732

• 99% of the variation in the dependent variable (quantity demand) is

explained by the regression

ANOVA
Model Sum of Squares df Mean Square F Sig.
1 Regression 8.688E10 3 2.896E10 484.979 .000a
Residual 5.972E8 10 5.972E7
Total 8.748E10 13
COEFFICIENTS

Unstandardized Standardized 95.0% Confidence

Coefficients Coefficients Interval for B
Model t Sig.
Lower Upper
B Std. Error Beta
Bound Bound
1 - -
(Constant) -51515.728 15393.024 -3.347 .007 85813.523 17217.932

ACCESS .039 .009 .514 4.185 .002 .018 .060

CPI 23.869 8.513 .411 2.804 .019 4.901 42.836
ICT 17333.013 7983.213 .109 2.171 .050 -454.694 35120.720

Q
Q == -51515.728
-51515.728 ++ 0.039
0.039 TT ++ 23.869
23.869 II ++ 17333.013
17333.013 D
D
 Log Linear
COEFFICIENTS

Unstandardized Standardiz 95.0% Confidence

ed
Coefficients Interval for B
Model Coefficients t Sig.
Lower Upper
B Std. Error Beta Bound Bound
1 (Constant) -2.241 6.164 -.364 .724 -15.974 11.493
ACCESS 2.013 .786 1.226 2.562 .028 .262 3.763
CPI -1.865 1.952 -.546 -.956 .362 -6.213 2.484
ICT .824 .502 .324 1.642 .132 -.294 1.943

• t-statistics
• R square value : < 90 %
 Demand Function

Q
Q == -51515.728
-51515.728 ++ 0.039
0.039 TT ++ 23.869
23.869 II ++ 17333.013
17333.013 D
D

 Fitted by 99%
 Significance
 t statistic for all independent variables > 2
 p value < 0.05
 Forecasting Demand
Q
Q == -51515.728
-51515.728 ++ 0.039
0.039 TT ++ 23.869
23.869 II ++ 17333.013
17333.013 D
D

 Telephone Subscribers  Consumer’s Price Index

4000000 7000
f(x) = 252677.07 exp( 0.19 x )
3500000 f(x) = 324.63 x + 1128.97
f(x) = 233689.66 x − 402293.69 6000

3000000 5000
2500000
4000
2000000
3000
1500000
2000
1000000
1000
500000
0
0

 Forecasted using the ‘forecast’ function of Excel

 D = 1 (for 2010, 2011, 2012)
 Forecasted demand for the next periods

20

20

20
230

246

263
10

12
11
,

,
,87
014

739
6
Telephone subscribers Elasticity of Demand CPI Elasticity of Demand

2010 0.526 2010 0.622

2011 0.527 2011 0.611
2012 0.527 2012 0.601
 Conclusion

 Model is satisfactory

 Limitations
 Quality of Data

 Amount of ICT penetration

 This is use full to
 A Telco planning to invest more in this model

 A good product development will do it!!!