Note: Network operator

End-user customers

Wholesale goods

and services

Retail goods and

services Retail goods and

services

Wholesale goods and services

Retail goods and services

Wholesale goods and services

Retail goods and services

Network operator A

Virtual operator a

Network operator

Note: Network operator

A has a wholesale agreement with virtual operator a

The dotted lines indicate the direction of revenue

The full lines indicate the direction of services and goods

Variables

Levels

First

Difference

Log of Network Operator’s Total Revenue

177.631***

(0.0005)

233.119***

(0.0000)

Network Operator’s Market Share

216.854***

(0.0000)

327.407***

(0.0000)

Number of Hosted Virtual Operators

78.8455

(0.2714)

104.674***

(0.0072)

Number of Non-hosted Virtual Operators

73.5557

(0.9954)

167.580***

(0.0001)

Log of GDP per capita

58.8446

(1.0000)

158.990***

(0.0099)

Log of Population Density

107.650*

(0.0990)

281.720***

(0.0000)

Mobile Penetration Rate

202.230***

(0.0000)

408.491***

(0.0000)

Log of Network Operator’s Size

244.122***

(0.0000)

318.313***

(0.0000)

Network Operator’s Average Revenue per Use

294.646***

(0.0000)

374.348***

(0.0000)

Number of Main Network Operators

199.498***

(0.0000)

274.812***

(0.0000)

Table 2: Augmented Dickey and Fuller unit root test (ADF test)

Note: The null hypothesis is that there is a unit root (or non-stationary series)

Dependent Variables

Description

Source

Log of Network

Operator’s Total

Revenue

Logarithm of total revenue per network operator per year, including retail revenue and

wholesale revenue

Company websites including annual report, financial statement,

presentations, etc

Network

Operator’s Market

Share

Retail market share per each network operators at the end of

each year

Merrill Lynch publication, company websites Independent Variables

Increased Number

of Hosted Virtual

Operators

Number of hosted virtual operators established during a

specific year

Takashi mobile MVNO/SP list, Telecom paper mobile MVNO/SP list,

company websites and own calculations

Increased Number

of Non-hosted

Virtual Operators

Overall number of non-hosted virtual operators established

during a specific year

Same as above

Control Variables

GDP Growth

First difference of logarithm of

GDP per capita telecommunication union) world World Bank, ITU (international telecommunication indicators

database

Log of Population

Density

Logarithm of population density Same as above

Mobile Penetration

Rate

Number of mobile user per 100 inhabitants for a country in each

year

OECD website, ITU

Number of main

Network Operators

Number of network operators with market share more than one percent in a country in each

year

Merrill Lynch publication

Log of Network

Operator’s Size

Logarithm of average number of employees in the network

operator in each year

Company websites including annual report, financial statement,

presentations, etc

Network

Operator’s Average

Revenue per User

Average revenue per user per

network operator in each year Same as above

Table 3: Variable descriptions and sources

Log of Network Operator’s Total Revenue Network Operator’s Market Share Increased Number of Hosted Virtual Operators Increased Number of Non-hosted Virtual Operators growth GDP Log of Population Density Mobile Penetration Rate Log of Network Operator’s Size Number of Main Network Operators Network Operator’s Average Revenue per User Mean 7.396823 30.93607 0.000000 0.002053 -0.000191 4.436366 70.71730 7.936046 3.668033 42.07458 Median 7.323171 31.00000 0.000000 0.000000 0.043863 4.722943 74.66000 7.821442 3.000000 40.95211 Maximum 10.73976 70.00000 19.00000 26.00000 2.283141 8.862767 116.3900 10.12435 6.000000 101.0000 Minimum 0.000000 0.000000 -43.00000 -39.00000 -2.277892 0.693147 0.930000 0.000000 2.000000 7.500000 Std. Dev. 1.372212 16.07173 15.39252 14.93623 1.329563 1.827576 23.92650 0.930422 1.021555 16.51575 Skewness -0.375923 0.201661 -5.797742 -3.673147 -1.274278 0.003772 -0.488811 -1.006163 0.709526 2.112355 Kurtosis 4.552431 2.313255 64.05154 23.20551 50.13546 3.088667 2.783044 12.65153 2.767294 19.67791 Jarque-Bera 60.49809 12.89719 78361.25 9379.431 45214.84 0.161013 20.39053 1976.428 42.04645 6018.687 Probability 0.000000 0.001583 0.000000 0.000000 0.000000 0.922649 0.000037 0.000000 0.000000 0.000000 Sum 3609.650 15096.80 1.000000 1.000000 -0.09033 2164.946 34510.04 3872.790 1790.000 20532.39 Sum Sq. Dev. 917.0048 125792.3 6768.000 38541.00 24.32578 1626.597 278796.6 421.5883 508.2213 132839.0 Observations 488 488 427 427 427 488 488 488 488 488 Cross sections 61 61 61 61 61 61 61 61 61 61

Log of Network Operator’s Total Revenue Network Operator’s Market Share Increased Number of Hosted Virtual Operators Increased Number of Non-hosted Virtual Operators GDP growth Log of Population Density Mobile Penetration Rate Log of Network Operator’s Size Network Operator’s Average Revenue per User Network Operator’s Market Share 0.053673 Increased Number of Hosted Virtual Operators 0.120090 0.027170 Increased Number of Non-hosted Virtual Operators -0.224389 -0.244378 0.437695 GDP growth 0.147297 -0.000699 0.289876 0.412383 Log of Population Density 0.004075 -0.080107 0.045558 -0.057024 0.034412 Mobile Penetration Rate 0.075922 0.049886 0.311403 0.392335 0.350343 0.259802 Log of Network Operator’s Size 0.751926 0.050134 0.001981 0.114186 0.093770 -0.194978 -0.184845 Network Operator’s Average Revenue per User 0.170053 -0.036685 -0.081631 -0.044618 -0.225878 0.110667 -0.194553 0.011281 Number of Main Network Operators -0.390788 -0.347971 -0.075890 0.131374 0.041268 0.145004 -0.030546 0.196024 -0.377149

Table 6: Colllinearity statistics

VIF TOL

Increased Number of Hosted Virtual Operators 1.171583 0.853546 Increased Number of Non-hosted Virtual Operators 1.283137 0.779340

GDP Growth 0.473342 2.112639

Log of Population Density 1.122818 0.890616 Mobile Penetration Rate 2.514926 0.602374 Log of Network Operator’s Size 0.862616 1.159264

Table 6a: Collinearity statistics for network operator’s total revenue model

Note: If VIF>10 or TOL<0.1, multicollinearity problem exist. No multicollinearity

among variables in total revenue model.

VIF TOL

Increased Number of Hosted Virtual Operators 1.117878 0.894552 Increased Number of Non-hosted Virtual Operators 1.137702 0.878965 Log of Network Operator’s Size 0.207131 4.827866 Network Operator’s Average Revenue per User 15.002028 0.066657 Number of Main Network Operators 11.064382 0.090380

Table 6b: Collinearity statistics for network operator’s market share model

Note: If VIF>10 or TOL<0.1, multicollinearity problem exist. Average Revenue per

User and Number of Main Network Operators has multicollinearity.

VIF TOL

Increased Number of Hosted Virtual Operators 1.11532954 0.896596 Increased Number of Non-hosted Virtual Operators 1.13044422 0.884608 Log of Network Operator’s Size 1.26125035 0.792864 Network Operator’s Average Revenue per User 2.48376857 0.402614

Table 6c: Collinearity statistics for network operator’s market share model after

dropping the number of main network operators.

Table 7: Model specification tests for network operator’s total revenue equation:

Sum Squared Reside Degree of Freedom F-statistic P-value Null hypothesis:

y

it=

α

+

ββββ

’

x

it+

ε

it 294.1111 481

y

it=

α

i+

ββββ

’

x

it+

ε

it 29.35149 360 26.8372944

(F0.95(121,360)=1.267) <0.05 (The null hypothesis can be rejected)

y

it=

α

t+

ββββ

’

x

it+

ε

it 257.1989 466 4.45856684

(F0.95(15,466)=1.688) <0.05 (The null hypothesis can be rejected)

y

it=

α

it+

ββββ

’

x

it+

ε

it 26.05515 343 25.5709518

(F0.95(138,343)=1.256) <0.05 (The null hypothesis can be rejected)

Table 7a Result: pooled regression model (restricted model) can be rejected by

each unrestricted model.

Sum Squared Reside

Degree of

Freedom F-statistic P-value Null hypothesis:

y

it=

α

t+

ββββ

’

x

it+

ε

it

257.1989 466

y

it=

α

it+

ββββ

’

x

it+

ε

it 26.05515 343 24.7387412

(F0.95(123,343)=1.267 <0.05 (The null hypothesis can be rejected)

Table 7b Result: cross-time period fixed effects model can be rejected by

cross-section & time-period fixed effects model

Sum Squared Reside

Degree of

Freedom F-statistic P-value Null hypothesis:

y

it=

α

i+

ββββ

’

x

it+

ε

it

29.35149 360

y

it=

α

it+

ββββ

’

x

it+

ε

it 26.05515 343 1.552606

(F0.95(17,343)=1.653) >0.05 (The null hypothesis cannot be rejected)

Table 8: Model specification tests for network operator’s market share equation:

Sum Squared Reside Degree of Freedom F-statistic P-value Null hypothesis:

y

it=

α

+

ββββ

’

x

it+

ε

it 99099.92 483

y

it=

α

i+

ββββ

’

x

it+

ε

it 5720.323 362 48.8376372

(F0.95(121,362)=1.267) <0.05 (The null hypothesis can be rejected)

y

it=

α

t+

ββββ

’

x

it+

ε

it 95685.60 468 1.1133018

(F0.95(15,468)=1.688) >0.05 (The null hypothesis cannot be rejected)

y

it=

α

it+

ββββ

’

x

it+

ε

it 5610.289 345 41.6598989

(F0.95(138,345)=1.256) <0.05 (The null hypothesis can be rejected)

Table 8a Result: pooled regression model can be rejected by cross-section fixed

effects and cross-section & time-period fixed effects model

Sum Squared Reside Degree of Freedom F-statistic P-value Null hypothesis:

y

it=

α

i+

ββββ

’

x

it+

ε

it 5720.323 362

y

it=

α

it+

ββββ

’

x

it+

ε

it 5610.289 345 0.3561297

(F0.95(17,345)=1.652) >0.05 (The null hypothesis cannot be rejected)

Dependent Variable: RESID?^2 Method: Pooled Least Squares Sample: 1999 2006

Included observations: 7 after adjustments Cross-sections included: 61

Total pool (balanced) observations: 427

Variable Coefficient Std. Error t-Statistic Prob. Constant -4.221043 3.540039 -8.132612 0.0000 Increased Number of Hosted Virtual Operators -0.745214 0.032298 -1.000911 0.4231 Increased Number of Non-hosted Virtual Operators 0.009326 0.007392 0.828343 0.6523 GDP Growth 1.232659 0.022234 9.323026 0.0000 Log of Population Density 0.192438 0.934013 0.243965 0.8322

Mobile Penetration Rate -0.725307 0.004321 -5.233951 0.0000

Log of Network

Operator’s Size -0.152333 0.118204 -3.304042 0.0293

Effects Specification Cross-section fixed (dummy variables)

R-squared 0.209780 Mean dependent var 0.265084 Adjusted R-squared 0.173804 S.D. dependent var 0.670041 S.E. of regression 0.396753 Akaike info criterion -0.067208 Sum squared resid 31.06583 Schwarz criterion 0.496732 Log likelihood 65.08570 F-statistic 10.86493 Durbin-Watson stat 0.780294 Prob(F-statistic) 0.000000

Dependent Variable: RESID?^2 Method: Pooled Least Squares Sample: 1999 2006

Included observations: 7 after adjustments Cross-sections included: 61

Total pool (balanced) observations: 427

Variable Coefficient Std. Error t-Statistic Prob. Constant 60.01153 56.60068 2.691096 0.0699 Increased Number of Hosted Virtual Operators -0.186493 0.600433 -0.08964 0.8196 Increased Number of Non-hosted Virtual Operators 0.254850 0.197450 0.899901 0.2976 Log of Network Operator’s Size -2.604729 3.032595 -1.330552 0.2780 Network Operator’s Average Revenue per

User 0.864306 0.459475 7.893224 0.0000

Effects Specification Cross-section fixed (dummy variables)

R-squared 0.299082 Mean dependent var 21.33301 Adjusted R-squared 0.257093 S.D. dependent var 45.53416 S.E. of regression 48.05066 Akaike info criterion 20.34578 Sum squared resid 78227.0 Schwarz criterion 16.73259 Log likelihood -1945.009 F-statistic 6.563083 Durbin-Watson stat 1.008623 Prob(F-statistic) 0.000000