The Broadband market in Italy - Research Paper Example

BUSINESS AND ECONOMIC FORECASTING ASSIGNMENT 2 The Broadband market in Italy MATTHEW SMITH & SUKHJIT SINGH DHILLON & PARVEEN DOLL Table of contents: page: i. The Broadband market in Italy. 3 ii. Estimated Level of Saturation.. 5 iii. Fitting the S Curve 7 iv. Adjusting the S Curve.. 10 v. Interpretation of results 12 vi. Evaluation in Context... 13 vii. References 15 viii. Appendixes 16 The Broadband market in Italy In this assignment we focus on the broadband market in Italy. In 2007 it was estimated that there were over 10.860 million users and this amounts to 15.8 per 100 individuals in Italy. Broadband internet in Italy was introduced in the early 1990's by telecom Italy and later more firms entered the industry. Italy Broadband internet is mainly based on Fibre to the home (FTTH) and DSL. The Telecommunication industry in Italy is dominated by Telecom Italy which offers internet services via a subsidiary telecom Italy mobile, one of the major broadband internet providers is Vodafone Italy, which had approximately 5.3million subscribers in 2007and telecom Italy mobile which had approximately 5.15 million subscribers in 2007. In the third quarter of 2008 the broadband market subscribers was approximately 18 percent of the population. Italy has consistently been ranked as one of the countries with the lowest broadband penetration in Europe, the country lacked cable and DSL infrastructure and therefore market penetration was considerably low. The entry of other firms in the industry led into an increase in penetration and in 2005 Italy recorded the highest growth of broadband in Europe. Population growth rates in Italy have been decreasing over the last several years with current rates near zero. Major cities are densely populated and most of the population lives in flats and blocks that make FTTH and DSL more appropriate to provide broadband services. Vigevano(2002) pointed out that there were some barriers to the penetration of broadband in Italy and they included digital divide and lack of competition, this means that in rural area, broadband services were less profitable and therefore majority of the providers concentrated more in urban area. Government has not typically concerned itself with encouraging more sophisticated use of the internet such as online transactions, and focus on European Computer Driving Licence(ECDL) provided to teachers, government employees, students this leads to increasing the awareness and skills of individuals. Vigevano(2002) points out that there are some barriers to the penetration of broadband in Italy, some of the factors that he highlighted include lack of capital to improve infrastructure, this hinders the growth of broadband in Italy. He also pointed out that there are regulatory issues that pose uncertainty regarding rights; another problem is digital divide whereby there is high adoption of broadband in certain area and in others low adoption and therefore firms concentrate only in large cities. According to House (2001) the shape of the S curve depends on the time period and saturation level and that the saturation level is less than 100%, he also states that saturation is driven by both the supply side and demand side, saturation level depends on the appropriateness of the technology and Economies of scale reducing prices increasing level of acceptance to peak. Estimated Level of Saturation The chart below summarises Penetration rate of Broadband in Italy from 2001-2008 Diagram 1: From the chart it is evident that broadband users(households and business) per 100 individuals has gradually increased over the years, it is also evident that in 2001 to 2002 is the introduction stage when the new product is introduced and people are not familiar about the product so the acceptance rate is very low. From 2003-2005 is the take off stage, in this period growth rate of acceptance is increase dramatically. The point of inflection (when penetration rate is 50%, in mid 2004) lies between the take off stage and the ultimate saturation rate, for the period 2005 to 2007 growth rate of market acceptance increases at a decreasing rate. House (2001) states that shape of the S curve depends on the time period and saturation level, he also states that saturation is driven by both the supply side factors and demand side factors. Demand side factors include technological appropriateness of the product which in our case it is evident that broadband services are appropriate in the provision of cheap, fast and reliable connections. According to Vigevano(2002) barriers to the penetration of broadband in Italy include lack of capital, regulatory issues and digital divide resulting to lower penetration in the future. The S curve is used to investigate the diffusion of a new product or technology. When market is made up of imitators and when the know data about the product is limited, then the S curve is used to Project the rate of development or market acceptance rate (penetration) of the new products. The S curve is used to forecast the level of saturation and rate of adoption of a product. Here we will use the given data to project the market development using a fitted S curve to the existing broadband adoption rates for the Italian market. To project the market development we will begin fitting an S curve to the given data by investigating different saturation levels and their respective RMSE values to determine which assumed saturation rate is most appropriate. When an appropriate saturation rate is determined for the Italian broadband market, we will further refine the S curve by adjusting the slope of the curve for the lowest RMSE values. On the basis of above information, the following assumptions made regarding broadband market in Italy are as follows: 1. Broadband is a new product in the market and also the market is made up of imitators of new technology. 2. Population growth will remain low in the future. 3. The market saturation level is low because of individual barriers to use such as lack of skills, affordability and perceived lack of benefit. 4. Italy has traditionally been a low adopter of new technology. For these reasons, we expect the approximate saturation level to be 20% for the broadband market in Italy based on visual examination of the existing data. Fitting the S Curve We use the following formula for logistic curve that analyses the behaviour of a market of new product or technology over time. 1 Y = S* equation(1) 1 + 1 Emt+c Where Y is the known market penetration per 100 individuals S is the level of saturation which has a maximum value of 100% but it will never achieve this level. t is time period, E is exponential, m is the slope and c is the intercept or a constant for equation Emt+c, this refers to a compound growth rate of the ratio between achieved and not achieved saturation level. Equation(1) shows non linear growth model, it cannot be estimated directly so we can transformed into linear model by taking log both side of the equation(1). Ln(Emt+c) = Ln(Y/S-Y) Ln ( Y/ S-Y) =Emt+c This equation is refers as log linear growth model. To estimate this model we need time trend to calculate the value of Y over time, time trend in life cycle period can be generated by 0,1.2 and the S depend upon the quality of the product. We calculate Ln (Y/ S-Y) assuming that saturation = 18% Using data on Ln(Y/(S-Y)) we estimate the regression model mt + c = Ln (Y/ S-Y) Using the OLS method indicates that the optimal saturation rate for the given series is 18%. The resulting equation is: Ln (Y/ S-Y) = -3.030823+.85608*t Using Eviews we estimate the model and the following are the results: Dependent Variable: LOG(Y/(18-Y)) Method: Least Squares Date: 04/22/09 Time: 21:52 Sample (adjusted): 2001 2007 Included observations: 7 after adjustments Coefficient Std. Error t-Statistic Prob. C -3.030823 0.157397 -19.25593 0.0000 T 0.856080 0.043654 19.61055 0.0000 Using the above model we now estimate values of Y for the year 2001 to 2025. The following graph demonstrates the initial fit of the curve to the existing data. Diagram:2 OLS analysis of the resulting series indicates the best saturation rate is 18%. Additional saturation rates of 10%, 15%, 17%, 18%, 19%, 30%, 40% and 50% were tested for appropriateness with 18% resulting in the lowest RMSE value. The RMSE table is included in Appendix 2. The above graph shows the fitted S curve on the Broadband market in Italy, from the diagram it is possible to identify the introduction stage, inflection point, take off stage , maturity and when the market be fully saturated. Introduction stage: This covers the period from the early 1990s until the end of 2001, during this period saturation level for this market increases at a very low rate because most of the people are not familiar with the new product. Take off stage: The period 2002-2005 indicates the take off stage, in this stage growth rate of saturation level increases at increasing rate because people are familiar about the new product and they have very keen to use the new technology. Inflection point: The inflection point is achieved in the middle of 2004 this is the point where the highest slope value is achieved shown in diagram 3 that depicts the slope, at this point strategies to increase production are appropriate as more and more individuals demand the product. Saturation: Full saturation which is 18% according to the curve is achieved in 2014 whereby the penetration level reaches 0.18, this is the period where we have minimal increase in the number of users and therefore strategies to increase production in those years would be inappropriate. Adjusting the S Curve We undertake a test to check the appropriateness of the fitted S curve to the data, the shape of the S curve will depend on a number of factors which include the saturation level, by changing the saturation level we can therefore test the level that best fits the S curve to the data, in this section we apply different saturation levels and determine the values of RMSE in each case, the following table summarizes the RMSE values for each saturation level chosen. As stated earlier, the OLS method indicated the best fit for the curve was at the 18% saturation level. This cannot be possible because data for 2008 show acceptance at 18.2%. This anomaly must be caused by the lack of smoothness of our given data which causes us to seek additional measures of appropriateness such as expected saturation level derived from the estimated inflection point. Using our previous visual determination of the inflection point would lead us to believe it is located around 10 to 11%. Because the inflection point is half the saturation rate, we would more reasonably expect the saturation rate to be approximately 20%. This estimation fits well with our given data. Selecting a new saturation level at 20% results in the following chart: The chart above shows a reasonable fit to the existing data with sufficient room for the penetration rate to slow over the next two years. The Slope: The shape of the S curve also depends on the value of the slope, we can manipulate the slope values and test the most appropriate slope that first the estimated values to the actual data. In our case our slope was determined through the estimation of a regression model, in this section we change the slope level and determine the effect of an increase and reduction of the slope on the RMSE for the fitted S curve. Slopes of 0.75, 0.77, and 0.82 were evaluated by the OLS method and 0.77 resulted in the optimum values (Appendix 3). As seen in the previous chart, the curve fitted from the equation Ln (Y/ S-Y) -3.040988 + 0.770763*t captures the anticipated saturation point as well as the estimated inflection point of the existing data. Intrepretation From the above curve it is evident that the maximum growth rate occurs in mid 2004. Applying this curve to the market means that the take off stage will last from 2002 to 2005 when the market users per 100 individuals will increase at an increasing rate, the point of inflection will be achieved in mid 2004 and the market will be fully saturated in the year 2014. After the point of inflection the growth rate increases at a decreasing rate. Evaluation in Context Broadband internet connection is a product introduced in the early 1990's in Italy, being a new product in the market the S curve will therefore be appropriate to project on future demand. Mahajan (1979) stated that the S curve is appropriate in determining the rate of diffusion of new products, therefore broadband being a new product in the market then the S curve is appropriate in the estimation of changes in market saturation and years of full saturation. It is also evident that broadband being a new product in the market the S curve is appropriate in depicting the product lifecycle which include the take off stage, the maturity period and the slow down period when the market is fully saturated. Another advantage of using the S curve in projection the broadband market is that data available is limited whereby other methods example the linear regression model requires use of a large sample in order to achieve more accurate results with lower level of mean error. For this reason therefore the S curve is appropriate in this case due to limited data to project. According to Meade (1984) the S curve is applied to imitative products whereby new products are classified into innovative and imitative, the broadband market in Italy is made up of imitators whereby the product, firms and users are likely to be imitators; this means that the S curve is appropriate for the broadband market. Another factor contributed by Meade (1984) is the fact that the S curve shows periods of demand take off and demand slow down which is not in the case with other estimation models, therefore broadband being a new product in the market period of demand peak and decline are important measures provided by the curve. The S curve is also an appropriate tool used in strategic management, having determined the changes that occur in the market strategies are aligned with the product to demand to achieve desired objectives. Some disadvantages exist in that it does provide a guideline on the effect of discontinuity of a new product introduced, these include the effect of demand of the new product when other products are introduced and other factors that may affect demand and therefore may provide unreliable projections. However despite the many disadvantages associated with the S curve it is still a reliable way to project, this curve has been used by a number of scholars such as Button (1980) who used the curve to predict future demand for cars in Britain, Bain (1963) who used the curve to project on TV demand in the UK and many other studies have utilised the S curve to project demand for new products. There were some difficulties fitting the s curve to our existing data due to lack of smoothness in our given data. These difficulties were overcome by using a common sense approach to understanding the data and the market it is taken from. References: Button, K. and etl. (1980) Disaggregate and aggregate car ownership forecasting in Great Britain, Transportation research, 14A, page 263 to 273 Mahajan, V and Peterson J. (1979) A new product growth model with a dynamic market potential, long Range planning, 12, page 51 to 58 Meade Nigel (1984) The use of growth curves in forecasting market development: a review and appraisal, Journal of Forecasting vol 3, page 429 to 451. OECD (2009) Broadband users per 100 individuals, 9th April, available at ITU(2007) Broad band in italy, users per 100, retrieved 9th April, from Paolo Vigevano (2002) Italy Broadband national strategy guidelines European Commission, Broadband Workshop Brussels, 28 January 2002 Appendix 1: Dependent Variable: LOG(Y/(20-Y)) Method: Least Squares Date: 04/22/09 Time: 14:29 Sample: 2001 2008 Included observations: 8 Coefficient Std. Error t-Statistic Prob. C -3.040988 0.173754 -17.50172 0.0000 T 0.770763 0.041535 18.55693 0.0000 Appendix 2: RMSE Y Yhat10 Yhat15 Yhat17 That18 Yhat19 Yhat20 Yhat30 Yhat40 Yhat50 0.7 0.5889 0.6946 0.7576 0.8289 0.8065 0.9122 1.135 1.18 1.1996 1.7 1.9183 1.8113 1.7981 1.8367 1.7865 1.8724 1.978 1.9809 1.9789 4.1 4.7378 4.1967 3.9225 3.7987 3.7141 3.6501 3.374 3.2797 3.231 8.1 7.735 7.8531 7.3443 6.9547 6.8891 6.5096 5.5598 5.311 5.1897 11.9 9.2833 11.349 11.196 10.748 10.852 10.21 8.699 8.3154 8.1293 13.2 9.8005 13.468 14.114 13.99 14.386 13.854 12.69 12.411 12.278 15.8 9.9466 14.42 15.732 16.046 16.711 16.594 17.047 17.416 17.651 18.2 9.9859 14.79 16.476 17.115 17.95 18.266 21.079 22.773 23.886 RMSE Calculation 0.0123 3E-05 0.0033 0.0166 0.0114 0.045 0.1892 0.2304 0.2496 0.0477 0.0124 0.0096 0.0187 0.0075 0.0297 0.0773 0.0789 0.0778 0.4068 0.0094 0.0315 0.0908 0.1489 0.2024 0.527 0.6729 0.7551 0.1332 0.061 0.571 1.3117 1.4662 2.5293 6.4524 7.7787 8.47 6.8469 0.304 0.4961 1.3266 1.0993 2.8551 10.246 12.849 14.218 11.556 0.0718 0.8361 0.6242 1.4062 0.428 0.2598 0.6218 0.8504 34.262 1.9041 0.0047 0.0606 0.8292 0.6308 1.5557 2.6116 3.4259 67.472 11.63 2.9715 1.1779 0.0627 0.0043 8.2861 20.911 32.336 MSE 15.092 1.7491 0.6155 0.5784 0.6289 0.8406 3.4492 5.7194 7.5479 RMSE 3.8849 1.3225 0.7845 0.7605 0.793 0.9168 1.8572 2.3915 2.7473 Appendix 3: Yhat Yhat Yhat Y 0.75 0.77 0.82 0.7 0.9121629 0.912163 0.912162915 1.7 1.8374417 1.872383 1.95763057 4.1 3.5278339 3.650117 3.953187551 8.1 6.2391313 6.50961 7.174105732 11.9 9.7950887 10.2103 11.18940316 13.2 13.403656 13.85424 14.85003384 15.8 16.22763 16.59425 17.34995203 18.2 18.021118 18.2656 18.73936638 (y-yhat)^2 0.0450131 0.045013 0.045013103 0.0188902 0.029716 0.066373511 0.3273741 0.202394 0.021553895 3.4628322 2.52934 0.857280196 4.4306516 2.8551 0.504947873 0.0414759 0.428029 2.72261167 0.1828677 0.630838 2.402351297 0.0319989 0.004303 0.290916095 MSE 1.067638 0.840592 0.863880955 RMSE 1.0332657 0.916838 0.929451965 Read More