For this publication, regional and global aggregates up to 2020 were calculated using data supplied by Member States to ITU, supplemented by ITU estimates. Aggregates can differ from those produced for previous editions of Facts and figures, because of new or revised data submitted by Member States. All 2021 aggregates are estimates computed by ITU, based on the methodology described below.
The percentage of the population covered by a mobile signal (2G/3G/4G and above) refers to the percentage of inhabitants who have such coverage, regardless of whether they use the service. The indicator thus measures the physical availability of mobile cellular services, not the actual level of use or subscriptions. It is differentiated by urban and rural areas.
The data for this indicator are generally provided in aggregate form (urban and rural). As with many indicators, ITU collects the data from telecommunication operators, telecommunication/ICT regulators and national ministries. This information is widely available for both developed and developing countries.
Since most countries provide data without differentiating between urban and rural coverage, it is necessary to perform disaggregation. It is known that mobile cellular coverage is virtually ubiquitous in urban areas. Furthermore, country data on the proportion of the population that lives in rural areas is published by the World Bank. Subtracting the urban population from the total population with mobile cellular coverage therefore gives the number of rural inhabitants who have such coverage in rural areas.
The percentage of the rural population covered by a mobile cellular signal (2G/3G/4G and above) is then obtained by dividing the number of such rural inhabitants by the total rural population and multiplying by 100.
Statistics on Internet use and household access to computers and the Internet can be derived from household surveys. However, relatively few countries administer such surveys, owing to their cost and the complexity of their implementation; accordingly, there are large data gaps.
In addition, the delay between the collection of household survey data and their publication can be as much as two years or more, limiting their usefulness for ICT statistics given the rapid pace of technological change.
These shortcomings make it necessary to rely on data modelling tools to estimate missing values and forecasting techniques to estimate current Internet use and access.
For 2021, forecasting was used to estimate the proportion of individuals using the Internet and the proportion of households with Internet access. Estimates were used to fill gaps in the data so as to produce regional and global aggregates (up to 2020) for Internet use—total, by gender, for young people (aged 15-24), and in urban and rural areas—and for households with Internet.
In addition to official ITU data collected from the membership, data was obtained from other sources, including Multiple Indicator Cluster Surveys, Demographic and Health Surveys, and the surveys of Research ICT Africa and LIRNEAsia. Additional data on socio-demographic characteristics were obtained from the World Bank and the United Nations Population Division.
Based on the data points or estimates, multiple-country aggregate values were calculated based on a weighted average of the values for individual countries. Internet use aggregates were weighted by the population of each economy, while aggregates for the proportion of households with Internet access were weighted by the number of households.
Disaggregation of overall values was performed separately. For instance, where country data on the number of Internet users were only available in aggregate form (urban and rural), comparable economies for which disaggregated data are available were used to estimate that ratio for the country in question, filling the gap in real data. Existing data on the country’s population size and urbanization were then used to produce separate estimates of the proportion of the population using the Internet in both urban and rural areas. Global and regional figures were calculated by weighting the figures for individual countries by the rural and urban population in each country.
A similar procedure was used to estimate Internet use separately for men and women, and for young people and the rest of the population.
June 2021 data on subscriptions were compiled from publicly available data from regulators and ministries, as well as subscription information published by each country’s main operators. When the data from the main operator of the country was used, the operator-reported number of subscriptions was divided by its market share to obtain the total number of subscriptions in the country for a particular service. In the absence of annual reports, subscription data were estimated from industry analyses, authoritative news articles and operator press releases.
Data from these sources include the absolute number of subscriptions, market shares, penetration and growth rates, which were used to derive the country estimates using the same method as with operator data. In the case of countries for which data were not available either from the national administration or from annual and industry reports, subscriptions data were estimated using univariate time series analyses applied to the data from the last 10 years.
The univariate time series analyses were done by decomposing the time series of penetration data of a particular service to its trend and residual component so as to obtain the autoregressive integrated moving average (ARIMA) models. The resulting ARIMA models were used to make the 2021 point prediction for each country and service.
The basic assumption here is that international bandwidth usage is a function of demand for total bandwidth capacity in a country, which can be obtained by multiplying the number of Internet users by their average bandwidth use. Since very few countries publish monthly or quarterly statistics on international bandwidth usage and given the extraordinary nature of Internet activity in 2020 and 2021 due to the COVID-19 pandemic, estimates were performed using proxy indicators.
ITU statistics on fixed broadband subscriptions were combined with crowd-sourced statistics on average download speeds published by Ookla [1] for 180 economies. Smoothed speed change ratios from December 2020-January 2021 to June-July 2021 served as the basis for extrapolating growth rates for 2021. Estimates were validated against reports by submarine cable operators.
For economies with missing or unreliable fixed broadband download speed data, 2021 values were estimated with exponential smoothing relying on historical figures since 2016. It should be noted that, due to variations in the quality of speed statistics, the explanatory power of the model is stronger in countries where fixed broadband Internet is dominant, and in middle-income economies in general. Results are potentially sensitive to significant changes in broadband subscriptions during the year.
[1] Ookla URL: https://www.speedtest.net/global-index/ (2020-2021 monthly editions; retrieved January to September 2021)