A conjoint analysis of a Next Generation Network (NGN) in Japan

Transcription

1 A conjoint analysis of a Next Generation Network (NGN) in Japan Naoki TAKANO 1 NTT DoCoMo Ventures Inc. takano.desu@tenor.ocn.ne.jp Received: 05 November 2013 / Accepted: Day Month 20XX Springer 2013 Abstract. A conjoint analysis on a Web survey in 2010 with 160 participants and 8 questions shows that a Marginal Willingness To Pay (MWTP) on a Next Generation Network (NGN) of Flet s Hikari Next is lower than its actual price as provided by NTT East. Conjoint analysis is one application of conditional logit. MWTP is WTP when one unit of property is increased, and MWTP is used to verify how much each person evaluates each property. Higher transmission speed, digital TV availability, and higher Internet security affect the MWTP positively, and the amount of the monthly payment of NGN affects the MWTP negatively. Therefore, it is assumed that the NGN of Flet s Hikari Next by NTT East will be accepted by potential customers and diffused as a major communications service of NTT East and NTT West. NGN providers may be able to maximize their benefits by adjusting the NGN of Flet s Hikari Next property to meet customer demands, especially in digital TV availability, higher transmission speed, and communications security. Keywords: Applied Microeconomics, Industrial Organization, Telecommunications, NGN, ICT 1 The views expressed herein are my own and do not necessarily represent the views of the company or the company group to which the author belongs.

2 2 1 Analysis background In this paper, the author analyzes the potential demand of a Next Generation Network (NGN), which will create a new telecommunications market and the future development of NGN. According to ITU-T (International Telecommunication Union Telecommunication Standardization Sector), the definition of NGN is stated as follows: A Next Generation Network (NGN) is a packet-based network able to provide services including telecommunication services and able to make use of multiple broadband, QoS-enabled transport technologies, in which service-related functions are independent from underlying transport-related technologies. However, the interpretation of the definition of NGN differs by respective telecommunication operating companies. For example, NGN in British Telecom is provided through ADSL, while NTT s NGN is provided only through high speed FTTH. These are totally and technically different services. The following hypothesis will be proved in this paper: NGN (Flet s Hikari Next) developed in 2010 will be accepted by potential customers and diffused because of its high security, high speed, and TV availability. Flet s Hikari Next is a part of the NGN provided by NTT East and NTT West. The current Japanese telecommunication market faces a revolution caused by the unification of existing telecommunications networks to NGN. NGN is a new network unifying the current telephone network (PSTN: Public Switched Telephone Network), the data network, and the mobile network by IP (Internet Protocol) technology. NGN has both the reliability and stability of PSTN and the flexibility and economy of an IP network. BT launched the first NGN to the world in In Japan, NGN was commercialized in 2008 and in August 2010, NTT East and NTT West provided Flet s Hikari Next as a B to C service, in Hikari Denwa (Telephone by fiber optic), Business Ether Wide, Flet s VPN Gate, and in Flet s Cast. In the future the diffusion and maturity of NGN, which will include PSTN, the data network, and the mobile network, may change the structure of the telecommunications industry, thus impacting existing telecommunications networks. Consequently, it is necessary to analyze which type of telecommunications industry will be developed as a result of competition among telecommunication companies. Therefore, this paper will focus on the determinants of subscription and utilization of NGN on the demand side. Because NGN was in the development phase in 2010, only a few people actually used NGN. Accordingly, utilizing RP (Revealed Preference), which requires the

3 3 actual data of services is difficult. For this reason conjoint analysis as SP (Stated Preference) is used in this paper. 2 The Japanese telecommunications market FTTH (Fiber To The Home) is becoming the major access network because the number of subscribers of FTTH is more than DSL (Digital Subscriber Line). As of March 2010, the diffusion rate of FTTH was 28.7% in Japan, and the number of subscribers was 15,017,316. Prefectural diffusion rates were 42.5% in Shiga, 41.9% in Tokyo, 35.5% in Osaka, and 35.5% in Kanagawa. The author of this paper originally thought that the ADSL (Asymmetric Digital Subscriber Line) would be a temporary technology until FTTH prevails generally; however, there are still many ADSL subscribers even though the number is diminishing because of the increased use of FTTH. The number of combined subscribers to both ADSL, whose speed is different in uplink and downlink, and to DSL, whose speed is equal in uplink and downlink are 11,184,265, and their average diffusion rate is 21.4% in Japan. ADSL and DSL diffusion rates are the following in different prefectures: 31.0% in Shizuoka, 26.8% in Ibaraki, 26.8% in Niigata, and 25.2% in Yamagata. (Press release material of The Ministry of Internal Affairs and Communications, June 19, 2010) 2 The number of subscribers of PSTN continues to diminish because of the diffusion of personal mobile phones and IP telephones by FTTH and ADSL. The mainstream of radio broadband services is the 3G 3 mobile phone and new radio communication such as WiMAX 4 that has appeared in the market. FWA (Fixed Wireless Access) and BWA (Broadband Wireless Access) are efficient radio data subscriber services in depopulated areas, and do not depend on FTTH as their last one mile. The number of telecommunication carriers is increasing and the variety of services is accelerated so that a diversity of choices from the view of users is increasing. The number of subscribers of NGN was 1,642,000 as of March 2010, reflecting a change in the telecommunications market Third generation mobile phone conformable to IMT 2000 standard by ITU-T (International Telecommunication Union-Telecommunication Standardization Sector). 4 Worldwide Interoperability for Microwave Access: 40 Mbps in downlink, 10 Mbps in uplink.

4 4 3 Previous research and the position of this paper The research in the paper is new because the subject of the research refers to NGN, which is one of the newest telecommunication services. The analytical method applied in this research is conjoint analysis, which is used in Kuriyama and Oono [1], Ida and Sato [2] and Ida [3]. We also refer to various studies of broadband services that have researched the IP telephone or other services in Japan by conjoint analysis such as in: Ida, Kinoshita and Sato [4], Ida and Horiguchi [5], Kondo, Nakamura and Mitomo [6]. In Ida [3], conjoint analysis is conducted with SP data divided into two groups: potential subscribers who live in a FTTH available area and potential customers who live in unavailable areas. As a result, those who live in non-ftth available areas show a higher WTP (Willingness To Pay) for higher speed than the group who live in available areas. 4 Conjoint analysis Conjoint analysis was born in psychometrics in the 1960 s and was developed in marketing research. Conjoint analysis was applied in economics from the mid-1990 s and utilized in Washida [7], Kuriyama and Ishii [8], Kuriyama [9] in environmental economics, and in Ida [3] in network economics. In this paper, the research design, the estimation method, and the measurement method of economic benefit were adopted from Kuriyama and Oono [1]. Also, the method of creating alternatives in questions was referenced from Ida [3], in which conjoint analysis was used to examine the substitute IP telephone and PSTN. The CJ2 method of Kuriyama [9] was used in the actual analysis in this research. Using conjoint analysis, it is possible to evaluate the value of property constituting targeted services or materials by repeatedly asking a consumer s preferences to plural alternative sets. The respondents can easily interpret the hypothetical services, because they are stated as concrete service sets. It is easy for the respondents to compare the actual services that they prefer or use with the hypothetical services. In conjoint analysis, goods are deemed to consist of a variety of bundles of properties (profiles) and the value of every property can be calculated by presuming the statistical relationship between the profiles and answers. Conditional logit is utilized in conjoint analysis. The choice probability of the conditional logit is shown by equation (1) below. The numerator represents the

5 5 consumer utility and the denominator represents the sum of the utility of alternatives. (Train [10]: 34-36) = (1) Table 1. Example of questions regarding conjoint analysis SPEED: Transmission speed (Mbps) TV: Digital TV availability SEC: Degree of Internet security NTT: If provider is in NTT group or not COST: Monthly charge (1,000 JPY) Service A Service B Service C Service D Service E 200Mbps 100Mbps 50Mbps 1,000Mbps 50Mbps Available Available N/A Available N/A Very high High Normal High Low By NTT Not by NTT By NTT Not by NTT Not by NTT Question: Which service do you most prefer? Please choose one service you like best. Table 1 shows an example of a conjoint analysis question. In answer to the question, Services A-E are shown in the Table and a respondent chooses one alternative as the most preferred. By statistically analyzing the relationship between presented alternatives and the answers of respondents, the value of the property, which consists of alternative choices, can be evaluated. Because choosing one alternative is similar to choosing behavior in a store, respondents find it easy to answer the questions. The conditional logit model, a discrete model, is used to estimate values from data. Because the log-likelihood function has to be maximized, particular statistical applications need to be handled, or a new program must be developed. In this paper, Kuriyama s CJ2 program [9] is used instead of a new model. 5 Determining properties and profile design To begin conjoint analysis, first, properties that constitute an object or a service are determined. Each property has several levels. After the properties and their levels are

6 6 determined, profiles of combinations of properties and levels are made. Generally, an orthogonal table as a profile design is used. Table 2 is an example of an orthogonal table of two properties and three levels. For example, profile <2> is made by property A of level 1, property B of level 2, and property C of level 2. Avoiding multi-co linearity is important because the correlation between each property is 0 in an orthogonal table during estimation. Table 2. Example of an orthogonal table L 4 (2 3 ) Profile Property A Property B Property C <1> <2> <3> <4> In choice type conjoint analysis, several profiles are presented to a respondent and the respondent is asked to choose the profile he/she likes best. Table 3 is an example of a question. In this case, the respondent is requested to choose one alternative (profile) from four alternatives (<1>~<4>). Table 3. Example of an orthogonal table L 4 (2 3 ) Profile Speed Monthly payment TV <1> 100Mbps 6,000 yen <2> 100Mbps 3,400 yen <3> 50Mbps 6,000 yen <4> 50Mbps 3,400 yen Question: Which service (profile) do you most prefer? Please choose one service you like best. 6 Determining properties and profile design Table 4 shows what properties and levels were determined, and the author analyzed Flet s Hikari Next FTTH service, which represents the NGN of NTT East and NTT West. Flet s Hikari Next FTTH service was chosen because, first, it is easy to

7 7 compare to other FTTHs such as like B Flet s of NTT East and NTT West; secondly, Flet s Hikari Next FTTH is the most diffused NGN service. The variables to each question are based on the following: COST: Monthly charge (unit: Japanese thousand yen) SPEED: Transmission speed (unit: Mega bit per second<mbps>) TV: Digital TV availability (Dummy variable) SEC: Degree of Internet security NTT: Whether the service provider is NTT East or West or not (Dummy variable) Table 4. Properties and levels Level SPEED: Transmission speed (Mbps) TV: Digital TV availability N/A(0) or Available(1) SEC: Degree of Internet security Low(1), Average(2), High(3) or Very High(4) NTT: If provider is in NTT group or not Not by NTT East/West(0) or By NTT East/West(1) COST: Monthly charge (1,000 JPY) SPEED has four levels because the transmission speeds of Internet data access are 50 Mbps in typical ADSL, 100 Mbps in FTTH, 200 Mbps in NGN (Flet s Hikari Next), and 1000 Mbps (1Gbps) is available in business data services. Digital TV availability (TV) is a dummy variable and it has two alternatives: available (1) and not available (0). Digital TV in this paper means that the consumer can receive the transmission of terrestrial digital broadcasting. Digital TV service is an option of FTTH such as B Flet s or Flet s Hikari Next (NGN), which are provided mainly for consumers. Generally, more than a 6 Mbps stable speed is necessary to transmit digital TV. Digital TV is not available in ADSL because the distance from a telephone office or status of subscriber lines easily affects transmission speed in ADSL. Also, digital TV is not available in business data lines because digital TV may affect data transmission and there is little demand for digital TV in business. The degree of security (SEC) is regarded as very high (4), high (3), average (2), and low (1) by converting to the possibility of masquerading or converting to the vulnerability of a computer virus. SEC of NGN (Flet s Hikari Next) is very high (4) because it is furnished with a subscriber line ID check function to prevent masquerading. It is difficult for the consumers to understand the degree of security because the author sets the attributes of SEC on a proportional scale for the better recognition of the respondents. NTT is a dummy variable because NGN in Japan has only been provided by NTT East and NTT West as of August 2010 and there have not been many subscribers. The

8 8 question about which telecommunications service the respondent most prefers uses the brand preferences of consumers. Monthly payments (COST) are leveled reasonably as 2,000 yen (2), 4,000 yen (4), 6,000 yen (6), and 8,000 yen (8) because actual monthly payment of NGN (Flet s Hikari Next) by NTT East and NTT West is 3,400 yen for an apartment and 5,200 yen for a house. The levels of each profile in a typical NGN (Flet s Hikari Next) are 200 Mbps in SPEED, available in TV, very high in SEC, provided by NTT, 3,400 yen in COST in case of apartment plan 2 5 which uses a fiber optic subscriber line. 7 Making profile design and executing research Table 5 shows how the author formulated the profile design. One respondent is asked eight different questions. Each question contains four alternatives (profiles) and the respondent answers the alternative he/she most prefers. The last alternative of the four alternatives is always the same in order to compare NGN (Flet s Hikari Next) with other FTTHs. In concrete terms, the respondent was asked the following questions and these questions are an example of card 1-8, Ver. a, Table 5 Question 1: Which telecommunications service do you most prefer? Please choose one service. 1. 2,000 yen monthly payment, 100 Mbps speed, digital TV available, low security, provided by a carrier other than the NTT group. 2. 6,000 yen monthly payment, 200 Mbps speed, digital TV available, average security, provided by a carrier other than the NTT group. 3. 2,000 yen monthly payment, 50 Mbps speed, digital TV not available, very high security, provided by a carrier other than the NTT group. 4. 3,400 yen monthly payment, 200 Mbps speed, digital TV available, very high security, provided by NTT East or West. By utilizing an orthogonal table, the eight questions were made by eight sets and 20 respondents answered for each set. Consequently, the number of data is = 1 5 Apartment plan 2 is the FTTH service provided for apartments.

9 9 Table 5. Profile design # Ver. Card C_1 Alternative 1 Alternative 2 Alternative 3 Alternative 4 (same) SPEED_ 1 TV_1 SEC_1 NTT_1 C_2 SPEED_ 2 TV_2 SEC_2 NTT_2 C_3 SPEED_ 3 TV_3 SEC_3 NTT_3 C_4 SPEED_ 4 TV_4 SEC_4 NTT_4 1 a a a a a a a a b b b b b b b b c c c c c c c c d d d d d d d d e e e e e e e e f f f f f f f f g g g g g g g g h h h h h h h h

10 10 8 Estimated result The estimated result is shown in Table 6. *** is to the right of the p value and represents significance at a 1% level. Table 6. Estimated result Coefficient t value p value MWTP COST *** SPEED *** SPEED TV *** TV SEC *** SEC NTT *** NTT N 1280 log likelihood The coefficient is the value estimated as each property s variable. Because the coefficient of COST is minus (-), the utility of a respondent becomes lower as the monthly payment is raised so that probability to choose the alternative decreases. Other coefficients of properties are plus (+) as the author expected so that the utility of the potential customer is raised by increasing transmission speed (SPEED), by having an availability of digital TV (TV), and by being provided by NTT East or West. The choice probability of the three alternatives is raised as the utilities are raised. All t values are more than three and all p values, which represent a significant level, are significant at the 1% level. To examine if the IIA (Independent of Irrelevant Alternatives) condition is effective, the Hausman test is executed to Hausman statistics and the result is χ 2 = Therefore, the IIA condition is not denied and the IIA condition is valid. The IIA condition is not effective if the Hausman statistics is more than an χ 2 threshold. 9 Marginal Willingness To Pay (MWTP) Marginal Willingness To Pay (MWTP) is the willingness to pay when a property increases its utility by one unit. MWTP is used to verify which function of a telecommunications service a person evaluates highly. MWTP is calculated by dividing the estimated coefficient by a monetary property coefficient. In this survey, the unit of the monthly charge is 1,000 yen so that the unit of MWTP is also 1,000

11 11 yen. However, TV (digital TV availability) and NTT (if provider is NTT or not) are dummy variables. Because a unit of SPEED is 1 Mbps, the MWTP is 1.16 yen per 1 Mbps. Also, because MWTP is only 116 yen per 100 Mbps, it can be said that consumers do not evaluate transmission speed as expected. Similarly, the MWTP of TV is 1,711.4 yen and the MWTP of the SEC (security level) is yen per one level. Consumers evaluate digital TV availability (TV) very highly. Evaluation of Internet security (SEC) is almost half the digital TV availability (TV), and, as a fundamental property of telecommunications, the attractive application of TV is evaluated more highly than Internet security (SEC). The MWTP of NTT (if provider is NTT or not) is 1,013 yen, which is very high, even though many other telecommunications companies exist in Japan. This means that consumers feel a great value in NGN (Flet s Hikari Next) provided by the NTT group. As shown above, a service can be evaluated by properties, when conjoint analysis is utilized. 10 Evaluating alternative services The monetary value of alternative services can be estimated by multiplying the levels of alternative service properties by MWTP, and then by taking their sum. For example, if the monthly charge of a fictional telecommunications service as shown in Table 7 is 4,017 yen, its properties are 500 Mbps in SPEED, available in TV, average in SEC, and not available in NTT. Table 7 Example of the evaluation of an alternative service Level MWTP Level MWTP SPEED: Transmission speed (Mbps) TV: Digital TV availability SEC: Degree of Internet security NTT: If provider is in the NTT group or not Amount (thousand yen) 11 A comparison of MWTP to actual service The MWTP of NGN (Flet s Hikari Next) actually provided by NTT East and NTT West is calculated as an alternative service in Table 8. Because actual monthly

12 12 charges are 2,500-4,400 yen for an apartment and 5,200 yen for a home, the theoretical MWTP, 6,416 yen, in this research is higher than the actual monthly charge. The reasons why the actual monthly charge of NGN (Flet s Hikari Next) is lower than the theoretical MWTP are first, NTT East and NTT West have a marketing strategy to penetrate the Japanese FTTH market, especially in apartments, at a lower price, and secondly, the installation cost of NTT East and NTT West is relatively inexpensive in apartments because of common fiber optic subscriber lines and related equipment. In this study, the MWTP may have an upward bias because the choice set of this survey does not contain the alternative of Don t subscribe. If consumers could choose the Don t subscribe alternative, a significant number of respondents might choose that alternative. This might reduce the MWTP of services. Table 8. MWTP of the actual telecommunications of NGN services Level MWTP Level MWTP SPEED: Transmission speed (Mbps) TV: Digital TV availability SEC: Degree of Internet security NTT: If provider is in the NTT group or not Amount (thousand yen) 12 Conclusion and solution The author uses a theoretical interpretation for the subject of demand analysis in NGN (Flet s Hikari Next) as a conclusion. 1. The hypothesis in this survey is that NGN (Flet s Hikari Next) of 2010 will be accepted by potential customers and is expected to be diffused because of its high security, high speed, and TV availability. The author has tested this hypothesis successfully in this paper. The first reason for the statistical significance of the hypothesis is that higher transmission speed will be available for digital TV, and will be provided by NTT East or West, which will bring a higher demand of NGN (Flet s Hikari Next) according to the conjoint analysis of this survey. However, demand will be diminished as monthly charges are raised. The second reason for the significance is that the actual monthly payment of NGN (Flet s Hikari Next) is lower than the theoretical MWTP of this research. Therefore, potential customers in the future will accept NGN (Flet s Hikari Next) from the

13 13 perspective of its properties and the monthly charges of NGN (Flet s Hikari Next). 2. According to the MWTP in this research, availability of digital TV and the consumer s brand preference to a telecommunications service provider are more highly evaluated by customers than Internet security and transmission speed, which are basic properties of telecommunications services. 3. Telecommunications companies may be able to provide a more attractive NGN (Flet s Hikari Next) to customers by adjusting properties such as digital TV availability, transmission speed, and Internet security level. From the point of view of price, potential users will accept NGN (Flet s Hikari Next) as mentioned above in 2. Consequently, NGN (Flet s Hikari Next) will be diffused in the future with the current monthly charge and adjusted attractive properties so that telecommunications companies will obtain more customers and profit through NGN (Flet s Hikari Next). A problem to be solved is that differences between the properties of the Web questionnaire respondents and the properties of the general consumers could not be adjusted before the actual Web survey began. In concrete terms, FTTH users in this research of the Web questionnaire of August 2010 were 51.3%, but the average FTTH diffusion rate in Japan was 28.7 % according to the Ministry of Internal Affairs and Communications in March The difference in diffusion rates of FTTH between the two types of research can be adjusted by a preliminary investigation. Also, the ratio of male to female users, the marriage rate, and residence area in terms of prefecture can be corrected by a preliminary investigation. In addition, the sample collection number of 160 may not be robust enough as a result for conjoint analysis in general. Appendix The data in this research was obtained by outsourcing a Web survey to a research company in Japan. The outline of the Web survey is as follows. The number of monitors registered in the Web research company was approximately 4.25 million as of August respondents were selected randomly so that a certain level of generality and objectivity could be confirmed. There can be little bias because 92.7% of the Japanese population are connected to the Internet and 100% of the 160 respondents in this research are connected to the Internet because this is Web research through the Internet.

14 14 Research area: Japan Respondents: Randomly sampled from Internet users who are registered as Web survey respondents Research date: August 9-11, 2010 Research agent: Net Mile Inc. The number of samples: 1,280 = 160 respondents 8 questions Preliminary investigation: No Research cost: Paid by Naoki TAKANO Copyright holder of data: Naoki TAKANO Acknowledgements I wish to thank two anonymous referees and the editorial board for very helpful comments on an earlier version of this paper. References 1. Kuriyama, K. and Oono, E.: Conjoint Analysis, and Practical Environmental Economic Evaluation. Keiso Shobo, , 2000 (in Japanese) 2. Ida, T., Sato, M.: Conjoint Analysis of Consumer Preferences for Broadband Services in Japan. The Kyoto Economic Review. 75(2), , Ida, T.: Broadband Economics. Nihon Keizai Shinbun Sha, 2007 (in Japanese) 4. Ida, T., Kinoshita, S. and Sato, M.: Conjoint analysis of demand for IP telephony: The case of Japan. CAEA Discussion paper series, Kyoto University, Ida, T., Horiguchi, Y.: Consumer Benefits of Public Services Over FTTH in Japan: Comparative Analysis of Provincial and Urban Areas by Using Discrete Choice Experiment. The Information Society: An International Journal. 24(1), 1-17, Kondo, M., Nakamura, A., Mitomo, H.: Quantifying the Benefits of the Internet and Its Applications. Keio Communication Review. 31, 37-50, Washida, T.: An Introduction of Environmental Evaluation. Keiso Shobo, 1999 (in Japanese) 8. Kuriyama, K. and Ishii, Y.: Environmental Value of Recycled Goods and Market Share: Conjoint Analysis Study. Environmental Science, 12(1), 17-26, 1999 (in Japanese) 9. Kuriyama, K.: Conjoint Analysis by EXCEL, Environmental Economics working paper #302, Waseda University, 2003 (in Japanese) 10. Train, K.: Discrete Choice Methods with Simulation, Second Edition. Cambridge University Press, 2009