To produce our forecasts we can deploy explanatory variables from amongst 26 set of time series statistics. There are 16 set of time series variables which provide purely background information and therefore do not appear in our published reports and briefs.

We also use historical data from as far back as 1980. What differentiates us from other providers of statistical data is our consistency in application of models by including the same explanatory variables across all countries; this of course is subject to the nature of the analysis. This enables us to be more confident about our market estimates. Quantum-Web produces like-for-like forecasts. This gives the organisations the confidence to easily compare country-specific forecasts against each other. Our data also provide easy access to a one-stop source of harmonised worldwide forecasts.

We usually adopt non-linear methods such as Gompertz, Logit, Tobit and Probit for long-term technology adoption based on the available information and model selection criteria.

For Short-term forecasts, on the other hand, we deploy Holts Winter methods.

To illustrate by an example, in trying to quantify the growth or potential size of broadband penetration per household in a country we start with an economic theory based on facts. These variables could be:

• Computers-in-use at home.
• Online households.
• Internet penetration.
• National telecommunication infrastructure expenditure.
• Internet tariffs.
• Competition level.
• Broadband penetration.
• Average household income.
• Socioeconomic indicators.

Based on certain restrictive assumptions we extrapolate data from our in-house data base and build an econometric model from it. In case of broadband penetration per households we apply a non-linear model and use the method of Least Squares.

Another example is to estimate the Internet access growth. The Internet access in a country is correlated to a series of variables. These could include computers-in-use at home/work, online households, dial-up and broadband fees, level of literacy, age band of population, geographical characteristics and distribution of population in urban and rural areas. Based on these characteristics we build a techno-economic model for groups of countries, which fall in the same category. For example countries cluster around certain level of market maturity or GDP per capita.