European Journal of Scientific Research ISSN 1450-216X Vol.40 No.3 (2010), pp.428-440 © EuroJournals Publishing, Inc. 2010 http://www.eurojournals.com/ejsr.htm The Analysis of Cellular Services and Estimating Fixed to Mobile Price Elasticities-A Case Study of United Kingdom Syed Umar Farooq Professor, Kardan Institute of Higher Education, Kabul, Afghanistan Muhammad Imran Ullah Chancellor, Abasyn University, Peshawar Roeen Rahmani Chancellor, Kardan Institute of Higher Education, Kabul, Afghanistan Abstract This paper seeks to undergo the systematic economic analysis of the rapid growth in the U.K. telecom sector and demand for mobile services and the current developments in telecom sector have been thoroughly discussed. The quarterly data from Ofcom telecom, UK telecom regulatory authority of four cellular operators, and fixed line services from 1999Q2 to 2006Q4 have been analyzed in this paper. The UK Mobile demand model has been made and it has been that the price elasticity (-0.52) is negatively related to the volume of mobile. It was also found that the GDP (Gross domestic product) and Subscribers both are positively related to the minutes of mobile telephone. The paper also highlighted the estimated UK fixed line demand model. The price elasticity -0.943486 have been found in the proceedings. It was also found that the GDP (Gross Domestic Product) is positively related to the fixed line minutes. 1. Introduction Mobile telephone brought a positive change in the people’s lives throughout the world. Mobile telephone introduce in Europe and in Japan in 1980 and in U.S in 1983 and since from that date its growth is around 25-30% a year such that at the end of year 1999 about 86 million mobile telephones were in use in the U.S. (data from spring 2000 CTIA (Cellular Telecommunications Industry Association) semiannual wireless industry.) only. About 32% of all American use mobile telephone, ITU (International Telecommunication Union) estimates that in 2002 the mobile telephone line exceeds the number of fixed lines for the first time. The penetration rate in Europe, Japan and in Australia is greater than in the U.S. In Finland the penetration rate is more than 70%. In Sweden and Japan mobile telephone surpassed the regular fixed line. The rapid growth of mobile telephone put a question mark for fixed line. Mobile telephone got the edge of stylish mobile sets, which is also a decent way of attraction toward the mobile contracts. Beside that it has got valuable feature, like easy to use, quality and the most important feature of mobile telephone is that “anytime and anywhere”. The price of mobile telephone is a bit high if we compare it with fixed line but mobile due to its feature is very much popular throughout the world. So in my point of view a small difference in price will not make any difference to mobile. Fixed line got the benefit of internet, like through fixed line and DSL (Digital Subscriber Line), the internet is very The Analysis of Cellular Services and Estimating Fixed to Mobile Price Elasticities-A Case Study of United Kingdom 429 fast and most of people are using fixed line due to that facility. But now the 3G (3rd Generation) may permit sufficient access to internet and making a threat to fixed line. The FCC (Federal Communication Commission) concluded from its review of third party research that about 3% of U.S users rely on mobile as their only phone and 12% reported that they prefer mobile telephone rather than adding a supplementary fixed line. (FCC (Federal Communication Commission) 2001, 2002c). Two surveys of British residential consumers, both found that about 5-6% of individuals claims that they use mobile telephone in their home instead of fixed line. (Oftel 2001, 2002). The subscriber ship rate in the U.S is about 25-30% per year. In the end of the year 1996 about 42 million cellular telephones were in use in U.S (NBER March 1997. Paper No-W5982) From 1990-2002 the number of mobile subscribers grew from 11 million to 1.15 billion with the annual growth rate of almost 47%, and in 2002 the number of mobile subscribers surpassed the number of fixed line. The subscribers in U.S of mobile telephone spend on average of $ 41.24 per month on mobile service, which is about $ 500 per year. Thus $ 41 billion is spent on mobile services so mobile revenue is about 40% as large as long distance revenue. Mobile telephone has become a basic need of individual’s life and it also has benefited people a lot due to which telephone industry generates a lot of revenue. Mobile telephone has also made a positive impact on trade and economic growth. It reduces the transaction cost and improves marketing information. (Jussawalla and Lamberton, 1982: Antonelli 1991: Greenstien and Spiller, 1995). World telephone line increases between 1990-1995 from 519 million to 612 million with average annual growth of 5.9%, world facsimile machine increased from 14 million to 35 million with average annual growth of 19.3%. In the same period the telephone lines of Africa increased at the average of annual rate of 7.9% and facsimile machine increased at the average of annual growth of 32.2%. (International Telecommunication Union 1997) 2. Review of Literature Chris (2003) presented a model of UK mobile telephone price elasticity of demand and market penetration. He collected quarterly data from 1995Q1 from oftel, Telecoms regulatory authority for each of the four operators: Vodafone, Cellnet, One2One and Orange. Pooled time series cross-sectional analysis on the four UK operators on quarterly basis starting from 1996Q3 to 1999Q1 was used to develop a demand model. He found the price elasticity with negative sign -0.49 as per expectation. Like if the prices increase by 1% on average the minutes will decrease by 0.49%. All the coefficient had the expected sign that is, negative for price, positive for GDP and Subscribers. The adjusted R square was very high at 0.9958 and the Durbon Watson statistic was close to 2 indicating no problem with first order correlation. Hyungtaik (2004) analyzed a fixed-mobile voice traffic substitution by model of Korean telephony markets using monthly data from 1996-2000. The estimated results showed that the share of call minutes originating from fixed line was negatively related with the relative price of calls originating from fixed phone vis-à-vis calls originating from mobile phone. In estimating the model his paper specified that fixed and mobile call demands, the subscription demands and the prices consists of a simultaneous equation system. Dwenter (2004) in their paper analyzed the demand for mobile telecommunications services in Austria, using firm level data on different tariff. Both static and dynamic panel data techniques have been applied in order to estimate short and long run price elasticities and the impact of networks subscriber basis on the demand for mobile calls. 430 Syed Umar Farooq, Muhammad Imran Ullah and Roeen Rahmani He used monthly data on mobile telephone traffic in Austria from January 1998 to March 2002. The data on prices, quantities and networks subscriber bases has been provided by the three largest Austrian mobile operators: Mobilkom, One and T-Mobile. Rodhini (2002) empirically estimated the substitutability of fixed and mobile services for telecommunication access using a large, U.S. household survey conducted over the period 2000-2001. The main source of data in their analysis was the Bill Harvesting data from TNS (Taylor Nelson Sofres) Telecoms ReQuest market Monitor along with its survey responses. Taubman (1998) explores the substitution effects between traditional fixed line and mobile services across Eastern Union and Soviet Union. Their analysis was based on empirical evidence based on country data, case studies and enterprise level data. Their sample consists of more than 6000 enterprises in 26 countries of the region. They designed their survey sample in such a way so they represent the population of the firm according to their economics significant, sector, size and their geographical location within each country, subject to minimum quotas for the total sample in each country. Michael (2005) presents “ Usage Substitution between mobile telephone and fixed line in the U.S”. In their study they estimate the substitutability of fixed and mobile services for telecommunications access using a large U.S, household survey conducted over the period 1999-2001. They collect data from TNS (Taylor Nelson Sofres) telecom and from the survey responses. They used an instrumental approach to estimate predicted prices for inclusion in their demand estimation. They found that mobile telephone is a substitute for fixed line services. Madden l (2003) used a global telecommunications panel.Consisting 56 countries and annual data for the period 1995-2000 to estimate a dynamic demand model. Their study examined the substitution effect between fixed and mobile telephony while controlling for the consumption externality associated with telephone networks. They found that growth of mobile subscription is associated with the mobile network and also with the reduction in mobile service prices. They also found a substantial substation effect, that the growth in mobile services induced by the increase in the prices of fixed land lines. Christopher (2007) presents “Demand for Telecommunication service in developing Countries”. In that paper they estimate telecommunication demand model for residential mainline and mobile telephone service for developing countries for the period 1996-2003. They tested the cross price effect and found the price elasticity for residential mainline insignificant in developing countries and also found that price elasticities for mobile telephone were very large. Gruber (2001), estimated diffusion Model for mobile services and analyzed, the effects of government policies on the global evolution of mobile telephony. They analyzed how entry and standards regulations affect the diffusion of mobile services, after controlling for various other Country-specific factors. Their study was based on data for 140 countries in which mobile services were offered during the period 1981-1997 (except for Japan for which data from 1979 were available). Gary (2000) examined the substitution between fixed lines to mobile telephony while controlling for the consumption externality associated with the telephone networks. They estimate a demand model by using a telecommunication panel data of 56 countries from 1995-2000. Pavlo (2005) raised the issue that weather mobile and fixed line telephone is substitute or complements in transition countries. He used fixed effect estimation for his model. The estimation of his model was done for the panel available from 1999 to 2001 and for the panel from 1993 to 2001. He founds that the penetration rate of fixed line was statistically significant and have a negative sign which showed that mobile and fixed line are substitute. Pedro (2001) used the data from 1981 up to the fourth quarter of 1999 and divided the numbers of fixed line users over the total population for the penetration rate; they did the same for mobile, like they divide the total number of mobile telephone user over the Population and got the penetration rate for mobile. They found that the introduction of competition encourages the mobile penetration. They found that penetration has been rising over time in a diffusion process. The Analysis of Cellular Services and Estimating Fixed to Mobile Price Elasticities-A Case Study of United Kingdom 431 3. Comparison (Mobile Subscribers VS Fixed Line Subscribers) The subscription rate of mobile telephone is very fast from the last few years. In Finland the number of mobile subscribers passed the number of fixed line in June 1997 and in Japan in March 2000. From the data obtained from Ofcom telecom official web site shows that from 1999 the growth of mobile telephone was unbelievable. But that is fact the mobile telephone because of low prices, luxury and the facility of receiving and making calls from anywhere any time become most popular among their users. It is very clear that mobile telephone cross the fixed subscription line in 2002 and with continues growth now it is more than the UK population which is around 60 million where as the mobile telephone users are around 64 million. The mobile users are more than the UK population because some people keep more than 1 connection. So it’s mean that the mobile users are more than 100% in UK. Figure 3.6: Fixed line vs mobile telephone Subscribers(000) 0 20000 40000 60000 80000 1 9 … 2 0 … 2 0 … 2 0 … 2 0 … 2 0 … 2 0 … 2 0 … 2 0 … 2 0 … Total Fixed Lines Total mobile Subscribers Figure 3.7: Total volume generated by both Fixed Line and Mobile Telephone (million) 0 50000 100000 150000 200000 2000 2001 2002 2003 2004 2005 2006 Total Volume Mobile Volume Fixed Volume 432 Syed Umar Farooq, Muhammad Imran Ullah and Roeen Rahmani Figure 3.8: Total Revenue Generated by both Mobile telephone and Fixed line. (million) 0 5000 10000 15000 20000 2000 2001 2002 2003 2004 2005 2006 Total Revenue Total M obile Revenue Total Fixed Line Revenue The figure 3.8 shows that the share of Mobile telephone in call revenue is increasing year by year, in 2000 the share of fixed line and was 47% while the share of mobile telephone was 53%. In year 2001 the fixed line share decreased from 47%-40%, while on the other hand the share of mobile telephone increased from 53% to 60%. In year 2002 mobile telephone put fixed line in continues trouble by outing them on back foot, the share of mobile telephone jumped from 60% to 65 % and fixed line took the remaining 35 % share. In 2003 mobile telephone share increased from 65% to 73% and the share of fixed line decreased to 27%. With continues growth of mobile telephone the share of mobile by the end of 2006 was over 84%, while the Fixed line share was below 16%. 4. Monthly contracts and Pay as you Go 4.1. Monthly Contracts In monthly contract every one have to pay monthly line rent as per their agreement with the company, in which they have been allotted monthly usage like free minutes, SMS, video calling, MMS, and free downloading. Company provides a mobile set and then collects the charges of mobile set and services in the shape of monthly line rent. So it varies with the company intentions. In monthly contract once you made a contract then you will not be able to cancel it before its time period and if one wants to cancel his/her contract then he/she must pay the line rent for the rest of the months. Monthly contract required credit score, so if one has poor credit history will not be able to get the monthly contract. 4.2. Pay as you Go (Pre Pay) Prepaid service was first launched in Portugal in September 1995 and appeared in the UK in the summer of 1997 and since from then its growth has been very fast. In pay as you go, customer will pay in advance for his/her usage. Company does not provide any free minutes and other service, but some time company start promotional offers, like if one top up by 5 pounds in a week and finish it in the same week then he/she will be allotted by free weekends call, SMS or landline numbers. Pre pay increase very fast and the reason is that credit worthiness is not required of the subscriber and spending is easy to control. 5. Data and Empirical Specification To analyze short and long run elasticities, quarterly data on mobile telephone traffic in UK over the period 1999Q2-to-2006Q4 has been used. The main source of collection of data is Ofcom, the Telecom regulatory authority. My data consist of minutes, revenue and subscribers of each operator Vodafone, O2, T-mobile and Orange. I take the GDP and RPI from National Statistics Agency. The Analysis of Cellular Services and Estimating Fixed to Mobile Price Elasticities-A Case Study of United Kingdom 433 The variable total number of outgoing minutes measures the quarterly traffic for each tariff. The variable consists of the sum of all outgoing calls, independently of the exact type of service (except for SMS or data services). Hence the variable represents an aggregate over various services (such as on net, off net, mobile to fixed and international calls) within a specific tariff. Further more the average call rate is used, which has been constructed by dividing the total revenue for each tariff by total number of outgoing minutes of that tariff. While mobile markets are characterized by price differentiation between peak and off-peak times, ore detailed data has not been available. 5.1. Descriptive Statistics (For Mobile) Mean Variance Minimum Maximum Source Minutes 21.56 0.0925 20.852 21.92 Ofcom Telecom Price -0.32554 0.008778 -0.52873 -0.02362 UK National Statistics GDP 8.026761 0.002172 7.924072 8.086718 UK National Statistics All the values in the above table are in logs. 5.2. Demand Model for Mobile Telephone Pooled panel data Cross sectional analysis based on quarterly data on the four U.K. operators running from 1999 Q2 to 2006 Q4 is used to develop a demand model. The available data is as follows: Raw Data From Ofcom: minutes, revenue (connection and call) number of subscribers, minutes from fixed to mobile for Vodafone, O2, T-Mobile and Orange. From UK Statistical Agency: GDP and CPI. Model Data Minutes Real ARPM= (Call Revenue/Minutes)/CPI Real GDP Minutes from Fixed Line Telephone to Mobile Phone. The UK demand model has the following generic form. Minutes ίt = αί + βί*Price ίt + δί*GDP +γί Subs ίt + είt Where Minutes ίt represents minutes for each operator over time. Price ίt represent price for each operator over time. Subscribers ίt represent Subscribers for each operator over time. GDP is same for all of the four operators, but changing over time. αί, βί, δί and γί are coefficient to be estimated. εί are random error term accounting for ummodeled influences on minutes in each equation. The above UK mobile telephone demand model is consisting of four entirely separate equations, with no restriction on coefficient. In this type of equations one might expect some correlation among the error term εί, for which I used Seemingly Unrelated Regression, but I got some wrong sign like positive price and also Durban Watson d state was very low, which indicate a positive serial correlation among the errors. Then I test it for poolability, which indicates that fixed effect is better, as the test statistics show to reject the null hypothesis, which is, that poolability is justified. Then I estimate the fixed effect model, in the form below for both mobile and fixed line. Minutes ίt = αί + βί*Price ίt + δί*GDP +γί Subs ίt + είt (1) 434 Syed Umar Farooq, Muhammad Imran Ullah and Roeen Rahmani Here the common coefficient assumed for the constant term (with coefficient α), Price, GDP and Subscribers are only 4 coefficients to estimate as opposed to 16 in the unrestricted model. So there is also a problem with fully pooled model is that putting restriction on all coefficient some time gives poor results. I then restrict some of the coefficient across equation and let the others unrestricted. The constant coefficient and GDP remain unrestricted and I restrict the price coefficient (β) and Subscribers coefficient (γ). The error term εί are assumed to be hetroskedastic, i.e. different variances. Minutes vary over the four operators. After estimation equation 1 the correlogram shows that there is serial correlation among the errors so then I use white heteroskadastic-standard errors and covariance, which is robust for serial correlation. (Eviews 2004). 5.3. Demand Model for Fixed Line Pooled panel data Cross sectional analysis based on quarterly data on the two U.K. operators running from 1999 Q2 to 2006 Q4 is used to develop a demand model. The available data is as follows: Raw Data From Ofcom: minutes, revenue (connection and call) number of subscribers, minutes for BT and Ntl and Telewest.. The UK fixed line demand model has the following generic form. Minutes ίt = αί + βί*Price ίt + δί*GDP +γί Subs ίt + είt (2) Where Minutes ίt represents minutes for each operator over time Price ίt represent price for each operator over time Subscribers ίt represent Subscribers for each operator over time GDP is same for all of the two operators, but changing over time. αί, βί, δί and γί are coefficient to be estimated. εί are random error term accounting for ummodelled influences on minutes in each equation. I do the same as I did for UK mobile demand model, like I try for SUR but again get the same result, then I test it for poolability, here the test statistics highly reject the null (poolability justified), and use the fixed effect. Poolability test for Mobile Demand model SSRr-SSRu n(T-k) F((n-1)k,n(T-k)) SSRr (n-1)k The SSRr is the sum of square residual from the restricted model which is the OLS, and the SSRu is the sum of square residual from the unrestricted model which is the fixed effect. n is the number of cross section used, T is the number of quarters, k is the number of regressors. The null hypothesis is Ho: βi= β So the F test comes from the above equation 22.224, which lies in the rejection region at all significant level, so I will reject the null, thus the poolability doesn’t justified so I used the fixed effect. Poolability test for Fixed demand model. The test of mobile poolability has been repeated here and F test comes to 18.27, so here again it lies in the rejection region so the null hypothesis is rejected with the use of the fixed effect. 5.4. Brief Explanation about Panel Data and SUR Panel data is a special form of longitudinal data in which observation are collected on the same sample of respondents over a period of time. (Web based definition) The Analysis of Cellular Services and Estimating Fixed to Mobile Price Elasticities-A Case Study of United Kingdom 435 In statistics and econometrics, the term panel data refers to two-dimensional data. Panel data set is one that follows a given sample of individuals over time and provides multiple observations on each individual in the sample. Panel data relate to individuals, firm states, countries etc, over time. Panel data estimation can take heterogeneity associated with such type of data by allowing individual specific variables. By combining time series of cross section observation, panel data give more informative data, more variability, less collinearity among variables, more degree of freedom and more efficiency. Panel data can better and measure effects that simply cannot be observed in pure cross-section and time series data. In short panel data can enrich empirical analysis in ways that may not be possible if we use only cross section or time series data. It is not for sure that there will be no problem with panel data modeling, but I will not go in detail. (Baltagi, op. cit., pp. 3-6) Seemingly Unrelated Regression. (SUR) SUR is a simple model involving several equations. Suppose that we have n equations of the form yί = Xβί + uί,, ί = 1,……….,n where yί and uί are nx1 column vectors, βί is a kx1 vectors of coefficient estimates and Xί is an n x kί matrix of regression of equation ί. The set of regressors Xί can be completely different for each equation and each can have a different number of regressors kί. Every equation satisfies the classical assumption that E(uί) = 0 Var(uί) = σ²ί In These assumptions rule out heteroscedasticity and serial correlation within any individual equation but allow the variances to differ between equations. 6. Estimation Results Table 1 Below shows the result of UK mobile demand model. All the variables are in logs thus the coefficients represent elasticities. The model has log of minutes as the dependent variable versus log of Real price, log of GDP and log of Subscriber. Table 1 shows the results from a model run with the price and subscribers coefficients constrained and the other coefficient unconstrained. 436 Syed Umar Farooq, Muhammad Imran Ullah and Roeen Rahmani Table 1: Variable Coefficient Std. Error t-Statistic Prob. F?PRICE -0.525235 0.082418 -6.372838 0 F?SUBS 0.667013 0.047487 14.04631 0 1--F1GDP 2.039439 0.285694 7.138549 0 2--F2GDP 2.823916 0.502695 5.617555 0 3--F3GDP 0.410526 0.616255 0.666162 0.5067 4--F4GDP 2.198955 0.447966 4.908755 0 Fixed Effects 1--C -6.03158 2--C -12.56252 3--C 6.655985 4--C -7.51703 Weighted Statistics R-squared 0.999878 24.21423 Adjusted R-squared 0.999869 6.530065 S.E. of regression 0.074856 0.638791 Log likelihood 159.3146 103989.9 Durbin-Watson stat 0.44998 0 Unweighted Statistics R-squared 0.963089 21.86402 Adjusted R-squared 0.960175 0.379263 S.E. of regression 0.075686 0.653038 Durbin-Watson stat 0.349557 Number of cross-sections used: 4 Dependent Variable: F?MIN Method: GLS (Cross Section Weights) Date: 11/02/07 Time: 11:58 Sample: 1999:2 2006:4 Included observations: 31 Prob(F-statistic) Mean dependent var S.D. dependent var Sum squared resid Total panel (balanced) observations: 124 One-step weighting matrix Mean dependent var S.D. dependent var Sum squared resid F-statistic Where F1 represents Vodafone F2 represents O2 F3 represents T-Mobile F4 represents Orange The entire variables are in log thus the coefficients represents elasticities. The model has log of minutes (F?min) as the dependent variable versus log of Real Price, log of GDP and log of subscribers. The price elasticity is -0.52, as expected. The entire coefficients are statistically significant expect for one of the GDP. The entire coefficients have the expected sign i.e. negative for price, positive for GDP and Subscribers. The adjusted R square is very high at 0.99986 The Durban Watson d state is very low. The correlogram of the residuals shows high autocorrelation. I used White period standard errors and covariance. First of all I get the expected signs in estimations. In the Pooled fixed effect model, I get the entire coefficient statistically significant except one of the GDP. In white period standard errors and covariance I also get the entire coefficient statistically significant except one of the GDP and constants. The difference among them is that all the standard errors t-state is different from each other. The Analysis of Cellular Services and Estimating Fixed to Mobile Price Elasticities-A Case Study of United Kingdom 437 Table 2 shows results of Pooled fixed effect model with white period standard errors and covariance in which price and subscriber coefficient are restricted and constant and GDP coefficient are remain unrestricted and table 3 shows the comparison of the two model. Table 2: Dependent Variable: F?MIN Method: GLS (Cross Section Weights) Date: 11/02/07 Time: 12:00 Sample: 1999:2 2006:4 Included observations: 31 Number of cross-sections used: 4 Total panel (balanced) observations: 124 One-step weighting matrix White Heteroskedasticity-Consistent Standard Errors & Covariance Variable Coefficient Std. Error t-Statistic Prob. F?PRICE -0.525235 0.090374 -5.811806 0 F?SUBS 0.667013 0.066273 10.06463 0 1--F1GDP 2.039439 0.396969 5.137521 0 2--F2GDP 2.823916 0.646368 4.368897 0 3--F3GDP 0.410526 0.686685 0.597837 0.5511 4--F4GDP 2.198955 0.489649 4.490883 0 Fixed Effects 1--C -6.03158 2--C -12.56252 3--C 6.655985 4--C -7.51703 Weighted Statistics R-squared 0.999878 Mean dependent var 24.21423 Adjusted R-squared 0.999869 S.D. dependent var 6.530065 S.E. of regression 0.074856 Sum squared resid 0.638791 Log likelihood 159.3146 F-statistic 103989.9 Durbin-Watson stat 0.44998 Prob(F-statistic) 0 Unweighted Statistics R-squared 0.963089 Mean dependent var 21.86402 Adjusted R-squared 0.960175 S.D. dependent var 0.379263 S.E. of regression 0.075686 Sum squared resid 0.653038 Durbin-Watson stat 0.349557 Table 3: Pooled. Fixed effect white period standard errors and covariance. Variable Coefficient Std. Error t-Statistic Prob. Std. Error t-Statistic Prob. F?PRICE -0.525235 0.082418 -6.372838 0 0.09037 -5.81181 0 F?SUBS 0.667013 0.047487 14.04631 0 0.06627 10.06463 0 1--F1GDP 2.039439 0.285694 7.138549 0 0.39697 5.137521 0 2--F2GDP 2.823916 0.502695 5.617555 0 0.64637 4.368897 0 3--F3GDP 0.410526 0.616255 0.666162 0.5067 0.68669 0.597837 0.5511 4--F4GDP 2.198955 0.447966 4.908755 0 0.48965 4.490883 0 Fixed Effects 1--C -6.03158 2--C -12.56252 438 Syed Umar Farooq, Muhammad Imran Ullah and Roeen Rahmani Table 4 shows the results from a UK fixed demand model run in which all the coefficients are constrained and with white period standard errors and covariance. Table 4: Variable Coefficient Std. Error t-Statistic Prob. F?PRICE -0.943486 0.075411 -12.51119 0 F?SUBS 2.575017 0.241816 10.64865 0 F?GDP -4.403227 0.565577 7.785367 0 Fixed Effects 1--C 11.83231 2--C 15.01226 R-squared 0.966318 23.30752 Adjusted R-squared 0.963868 0.491574 S.E. of regression 0.09344 0.480211 Log likelihood 59.69992 394.4756 Durbin-Watson stat 0.73319 0 Total panel (balanced) observations: 60 Mean dependent var S.D. dependent var Sum squared resid F-statistic Prob(F-statistic) Dependent Variable: F?MIN Method: Pooled Least Squares Date: 11/05/07 Time: 08:02 Sample(adjusted): 1999:3 2006:4 Included observations: 30 after adjusting endpoints Number of cross-sections used: 2 Where F1 represents BT F2 represents Ntl & Telewest. Notes The entire variables are in logarithms so their coefficients represent elasticities. The model has log of minutes (F?min) as the dependent variable versus log of Real Price, log of GDP and log of subscribers. The price elasticity is -0.943486, Expected sign and statistically significant. All coefficients are statistically significant. All coefficients have the expected sign i.e. negative for price, positive for GDP and Subscribers. The adjusted R square is very high at 0.999 The Durban-Watson statistic is 0.733190. But again I run the model with white period standard errors and covariance, which is robust for serial correlation. 7. Summary and Conclusion The quarterly data on the U.K. mobile telephone market obtained from Ofcom from the period 1999 Q2 to 2006 Q4, to estimate the price elasticities. As replicated Chris Dineen paper “Demand Analysis and Penetration Forecasts for the Mobile Telephone Market in the U.K.” so in the same way using pooled model the price elasticities have been obtained. The Log of Minutes was put as the dependent variable versus log of price, log of GDP and log of subscribers. It was found that the expected sign for entire coefficient like negative for price and positive for GDP and Subscribers was found for the price coefficient for mobile -0.52 where as Chris Dineen found it -0.49. The interpretation is as “if price increase by 1% the volume will decrease by 0.53% on average”, so price show a negative relation with volume, which was expected. I also found the GDP positively related to the usage of mobile telephone. One thing which I would like to mention that the Durban Watson state in Dineen estimation was close to 2, which showed no problem with serial correlation. The Analysis of Cellular Services and Estimating Fixed to Mobile Price Elasticities-A Case Study of United Kingdom 439 In UK fixed mobile demand model, data was collected from the Ofcom telecom from period 1999Q2 to 2006Q4 to estimate the price elasticities for fixed line. The price elasticity was found at (- 0.943486) negatively related to the usage of fixed lines minutes. The GDP in UK fixed line is also positively related to the minutes of fixed line. From these empirical findings it can be concluded that price and competition play an important role in the development of any industry. Mobil telephone took the path of growth because of new technology and affordable price. The empirical findings also show that usage of mobile telephone or fixed line affected by the price. A small decrease in price increases the volume of both mobile telephone and fixed line. The role and policy of government should change from price regulation and network sizes and should focus on price competition and quality improvement, as both price and quality plays also an important role in improving social welfare. References [1] Christiano Antonelli, “the diffusion of advanced telecommunications in developing countries” Vol 5, July 1993, Issue 2, pp197-200 [2] Pedro Pita Barros and Nuno Cadima, “The impact of mobile phone diffusion on fixed link” 23 March, 2001. [3] Chris Dineen, “Demand Analysis and Penetration Forecasts for the Mobile Telephone market in the UK” [4] Ralf Dwenter and Justus Haucap, “Estimating Demand Elasticities for Mobile Telecommunications in Austria”, December 2004, paper No.33. [5] Madden G., G Coble-Neal and B. Dalzell (2004). A dynamic model of mobile telephony subscription incorporating a network effect. Telecommunication policy, 28, 133-144. [6] Christopher Garbacz and Herbert G. Thompson Jr. “Demand for telecommunication services in developing countries”, Telecommunications Policy, Vol 31, June 2007, issue 5, pp 276-289. [7] Ahn, Hyungtaik, Jongwha Lee and Vongkyu Kim (2004) “Estimation of a fixed-mobile substitution model in Korean voice telephony markets,” paper presented at 15th biennial conference of the International Telecommunications Society, Berlin, Germany, September 2004. [8] Ahn, H. and Lee, M, “ An Econometric analysis of the demand for access to mobile telephone networks” Information Economics and Policy, Vol 11, 1999, pp. 297-305. [9] Gruber, Harald,, “The Diffusion of Mobile Telecommunications in Central and Eastern Europe, Information Economics and Policy, 13, 2001, issue 7/8/9/10, pp 893-906. [10] Rodini, M., M. R. Ward and G. A. Woroch, "Going Mobile : Substitutability between Fixed and Mobile Access," Telecommunications Policy, 27, pp.457-476, 2002. [11] Gary Madden, Scott J. Savage “Telecommunications and economic growth”, International Journal of Social Economics, Vol 27, 2000, pp 893-906 [12] Jussawalla, M. & Lamberton, D. 1982. Communication economics and development: an economics of the information perspective in M Jussawalla & D M Lamberton (eds.). Communication economics for development (pp1-5). Sydney: Pergamon Press. [13] Shane M. Greenstien and Pablo T. Spiller, “Modern Telecommunications Infrastructure and Economic Activity: An Empirical Investigation”, Economic Developments, Vol16, August, 2002, Issue 3, pp 211-228. [14] J.M Wooldridge, Econometric Analysis of Cross Section and Panel Data, MIT Press, Combridge, 1999. [15] Madden and Coble-Neal (2003), “Economic determinants of global mobile telephony growth”, Information Economics and Policy, Vol 16, December 2004, issue4, pp 519-534. [16] Michael R Ward and Glenn A. Woroch “Fixed-Mobile Subscription Substitution in the U.S.” March 2005 440 Syed Umar Farooq, Muhammad Imran Ullah and Roeen Rahmani [17] J.P.Roos, “Postmodernity and Mobile Communications” Mobilezation – for and against Thursday 31, August. [18] Pavlo Sugolov “Are Mobile and Fixed lines Substitutes or Complements? Evidence from Transition Economies” Summer 2005 [19] Chris Taubman and Maria Vagliasindi, “Fixed and Mobile Competition in Transition Economies”. [20] World Telecommunication Development Report 1998: Universal Access [21] National Bureau of Economic Research |c 1997 490 1 |a NBER working paper series ; |v no. w5982 500