Technology Diffusion and Long Term Forecasting: Application to Growth of Wireless Mobile Services AT&T Symposium August 3-4 2006 M. Hosein Fallah, Ph.D. Elias Aravantinos Wesley J. Howe School of Technology Management Stevens Institute of Technology Hoboken, NJ 07030 MHF/EA August 2006 Slide No. 1 Wesley J. Howe School of Technology Management
Outline Forecasting Technology Diffusion Modeling growth of Wireless services Problems with Mobile Diffusion Forecasts Case Analysis for Greece Issues and observations Implications of the study Future Research MHF/EA August 2006 Slide No. 2 Wesley J. Howe School of Technology Management
Forecasting People do forecasts all the time o Manufacturing o Sales o Financial performance o Natural phenomena o Short term forecast Daily weather o Long term Global warming MHF/EA August 2006 Slide No. 3 Wesley J. Howe School of Technology Management
What is Diffusion? Diffuse means spread out, scatter, pour in different directions Webster Dictionary Diffusion of an innovation is the process by which innovation is communicated among and adopted by the members of a social system MHF/EA August 2006 Slide No. 4 Wesley J. Howe School of Technology Management
Technology Diffusion People naturally resist change The adoption behavior for a new technology tends to follow an S curve Common diffusion models: Bass, Gompertz, Fisher-Pry The growth pattern of a new technology can be represented in general by a function of the form f ( e g ( t ) ) % Adoption, Y Time, t The S Curve MHF/EA August 2006 Slide No. 5 Wesley J. Howe School of Technology Management
The Gompertz Model The Gompertz model is asymmetric, with the adoption rate slowing down as it progresses The formula for the Gompertz model can be written as : y(t) = e -e-b(t-a) Where a is the year the growth reaches the inflection point on the curve. This point normally correspond to 37% of the saturation level b measures the speed of diffusion MHF/EA August 2006 Slide No. 6 Wesley J. Howe School of Technology Management
Diffusion of Some Past Innovations MHF/EA August 2006 Slide No. 7 Wesley J. Howe School of Technology Management
Wireless Saturation Level Diffusion of most technologies in the past was bounded by the total population. Saturation levels were below 1. Wireless mobile has changed the paradigm Everyone above age 10 can carry a cell phone. Some people may have multiple phones or multiple SIM Cards. In some countries Wireless mobile penetration has already past 100% So, where is the ceiling? MHF/EA August 2006 Slide No. 8 Wesley J. Howe School of Technology Management
Wireless Growth for Selected Countries (Actual) 1.2 1 0.8 0.6 0.4 UK Italy Greece USA Wireless Growth (Density) 0.2 0 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Year MHF/EA August 2006 Slide No. 9 Wesley J. Howe School of Technology Management
Wireless Growth for Selected Countries (Forecast) 1.2 1 0.8 0.6 0.4 0.2 Actual Forecast with Gompertz Italy UK Greece USA Wireless Growth (Density 0 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Year MHF/EA August 2006 Slide No. 10 Wesley J. Howe School of Technology Management
Forecasts with Traditional Models Short term forecasts could be very reliable Accuracy of long term forecasts varies with o Amount of the historical data o Where we are on the S curve o Potential external factors Technological, social, economic factors Methods for improving long term forecasts o Delphi Method o Analogy * and interpolation from similar observations * Yongil Jeon, Kwang Hyun, and Clive Granger (2004) Long-term Technological Forecasting,, Telektronikk MHF/EA August 2006 Slide No. 11 Wesley J. Howe School of Technology Management
Approach to Improving Longer Term Forecasts Short Term Historical Data Forecasting Model New Forecast Longer Term Historical Data Forecasting Model New Forecast Other Information Relevant to the Future MHF/EA August 2006 Slide No. 12 Wesley J. Howe School of Technology Management
Similarities between Greece and Italy Similarities Greece Italy GDP/capita $23,000 $28,000 APRU $27 $30.2 UMTS services launch 2004 2004 Population group 15-64 years: 66.7% 15-64 years: 66.8% Total median age 40.5 41.7 MHF/EA August 2006 Slide No. 13 Wesley J. Howe School of Technology Management
Wireless Mobile Growth- Italy vs. Greece 1.2 1 Growth (Density) 0.8 0.6 0.4 Italy Greece- Actual 0.2 0 1994 1995 1996 1997 1998 1999 2000 Year 2001 2002 2003 2004 2005 MHF/EA August 2006 Slide No. 14 Wesley J. Howe School of Technology Management
Actual vs. Forecasts for Greece- Application of Analogy Wireless Growth (Density) 1.2 1 0.8 0.6 0.4 0.2 Greece Greece from Italy Greece from Gompertz 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Year MHF/EA August 2006 Slide No. 15 Wesley J. Howe School of Technology Management
Issues and Observations Wireless technology is evolving very rapidly Countries are also going through policy changes that affects growth of wireless services Existing diffusion models can not realistically predict growth more than a year or two, particularly for rapidly evolving technologies While the notion of S curve is fundamentally sound, the current models have significant limitations for longer term forecasting because they look only backward and not forward. The logistic models need to be augmented with other forward looking information from the lead markets using analogy to improve long term forecasting MHF/EA August 2006 Slide No. 16 Wesley J. Howe School of Technology Management
Implications of the Study Improved models for diffusion of communications technologies will help service providers with better planning for o Infrastructure o Substitution of traditional services with advanced services o Resource management o Global expansions MHF/EA August 2006 Slide No. 17 Wesley J. Howe School of Technology Management
Future Work Further assess application of analogy to long term forecasting Improve and generalize the approach as a modified Gompertz model Validate model application and forecast reliability Explore applications to broadband services MHF/EA August 2006 Slide No. 18 Wesley J. Howe School of Technology Management