Empirical Analysis of Factors Promoting Broadband Deployment in OECD 30 Countries
University of Hyogo - Graduate School of Applied Informatics
KDDI Research Institute
Osaka University - Graduate School of Economics
March 31, 2012
Currently it is said that investment in broadband deployment has already reached a ceiling (Atkinson, Noam, Shultz; 2010). However, the average diffusion rate of OECD is about 60%, and policy measures are required for further deployment including the National Broadband Plan of the US, and Digital Agenda for Europe. This paper attempts to analyze factors promoting the diffusion of three broadband services in OECD 30 member countries from 2000 to 2010 by panel data analysis based on comprehensive data including prices and speeds. OECD countries are categorized into the following three types depending upon their larger share than the OECD average: (i) FTTH (e.g. Japan and Korea); (ii) DSL (Germany and France); and (iii) Cable modem diffusion type (US and the Netherlands).
The analysis particularly focuses on the following hypotheses related to FTTH, which are commonly believed, but not proved yet: (i) DSL is promoted by local loop unbundling (LLU), while LLU of optical fiber suppresses FTTH diffusion; (ii) faster speed of FTTH promotes its diffusion; and (iii) investment strategy of carriers such as the termination of copper local loop promotes FTTH.
Methodology and model:
The estimation model is formulated as follows.
D_T=a0 a1*P_CATV a2*P_DSL a3*P_FTTH a4*H_Intra a5*H_Inter a6*S_CATV a7*S_DSL a8*S_FTTH a9*C a10*U a11*I
D_T: number of subscribers of technology T
P_T and S_T: monthly subscription prices (USD) and the connection speed of technology T
i and t: country and time (quarterly)
H_Intra and H_Inter: HHI of intra-platform and inter-platform
C: number of cable TV subscriber as of 2000
U: dummy variable for unbundling of copper and optical fiber local-loop
I: period of announcement to terminate copper local loop by Japan, Korea and Australia.
An instrumental variable estimation method is used, since prices and HHIs are endogenous.
The result of estimation is summarized in Table 1. The price elasticity in all the equations satisfied the sign conditions, and the cross-price elasticity shows that three technologies are substitutes. The coefficients of connection speed satisfy the sign conditions in all models. As for intra-HHI, Cable modem is promoted by competition among carriers, while DSL and FTTH are promoted by less competition. Inter-HHIs have negative signs indicating that competition among three technologies promotes broadband diffusion.
LLU of DEL is positively, whereas that of FTTH is negatively significant. Thus LLU has different effect on the diffusion of DSL and FTTH. A coefficient of the connection speed of FTTH is negatively significant for DSL implying that the faster connection speed of FTTH induces migration from DSL to FTTH. Finally, the carriers’ decision of investment for FTTH is positively significant.
Using comprehensive data and a rigorous estimation method, this analysis proves for the first time that (i) three technologies satisfy sign conditions of prices and speed, (ii) three technologies are substitutes, (iii) LLU of optical fiber suppresses FTTH, and (iv) positive attitudes of carriers toward FTTH investment promotes FTTH. This analysis provides an important basis for policy makers and carriers.
Atkinson, B., E. Noam and I. Schultz  “Has Telecom Investment Peaked?” TPRC Proceedings.
Distaso, W., P. Lupi and F. M. Manenti  “Platform Competition and Broadband Uptake” Information Economics and Policy, 18 (1).
Lee, S.  “A Cross-Country Analysis of Ubiquitous Broadband Deployment: Examination of Adoption Factors,” mimeo.
Table 1 Result of estimation:
(1) Cable modem (2) DSL (3) FTTH
Log (price (Cable modem)) -2.290*** 0.809* 1.129*
[0.644] [0.443] [0.644]
Log (price (DSL)) 1.439* -2.910*** 2.199**
[0.816] [0.641] [1.032]
Log (price (FTTH)) 0.999* 3.272*** -1.979***
[0.529] [0.522] [0.627]
Log (speed (Cable modem)) 0.424*** 0.086 0.1
[0.124] [0.111] [0.204]
Log (speed (DSL)) -0.27 0.340* 0.469
[0.235] [0.204] [0.367]
Log (speed (FTTH)) 0.122 -0.308** 1.244***
[0.149] [0.136] [0.246]
Log (HHI (intra-platform)) -2.228*** 0.509* 0.920***
[0.432] [0.300] [0.270]
Log (HHI (inter-platform)) -1.312** -1.537* -1.765**
[0.563] [0.892] [0.727]
Log (number of CATV subscribers (the year 2000)) 0.695*** -0.079*** -0.049*
[0.112] [0.022] [0.026]
Unbundling of dry copper -0.22 0.533*** -1.836***
[0.179] [0.182] [0.359]
Unbundling of fiber local loop -0.098 -1.188*** -1.223*
[0.304] [0.316] [0.649]
Investment decision on optical fiber -1.827*** -0.189 6.407***
[0.500] [0.434] [1.423]
Fiscal year dummy Included. Included. Included.
Constant term 31.727*** -0.381 10.898
[4.938] [9.314] [8.053]
Number of observations 231 234 184
F-test (test of models) 32.84*** 5.08*** 24.81***
Coefficient of determination (within a group) 0.1995 0.1608 0.2875
Coefficient of determination (among groups) 0.6628 0.2196 0.8145
Coefficient of determination (total) 0.7347 0.2521 0.727
Note 1: *, **, and *** indicate significance at the 10%, 5%, and 1% level.
Number of Pages in PDF File: 21
Date posted: April 3, 2012 ; Last revised: April 2, 2013