Telecoms statistics: more than just a numbers game?

11 September 2012 | Tim Phillips

It has never been easier to find the numbers we need, or harder to make sense of them. This month, the ITU makes its statistics available through Google Public Data Explorer.

One of the exciting aspects of the penetration of mobile phones has been the mess it has made of our simple and satisfying measures of teledensity.

In May 2011, according to research firm The Mobile World, mobile teledensity globally exceeded the number of people for the first time. The ITU is more conservative, reporting that mobile subscriptions are at 86% of the world’s population at the end of 2011.

Statisticians must pine for the good old days when teledensity really meant something. Or, to be more rigorous, when it correlated with our best efforts to measure a country’s state of development.

In the 1960s, the first decade in which the UN defined the concept of “development” as something that should be planned, encouraged and measured, the number of telephone lines per 100 inhabitants was a useful statistic, not least because it correlated closely with measures of national wealth such as GDP per capita.

The correlation, first discovered by German engineer AGW Jipp in 1962, led to one of those convenient semi-truisms of economics: if you stick a lot of telephones in a country, people will, on average, get richer.

Mobile teledensity (or, as the jargon terms it, mobidensity) is – even in an industry where many statistics don’t quite measure what they seem to measure – a problematic measurement. It couldn’t be anything else, now we’re over 100% globally. Editing the numbers to account for phones that haven’t been used recently cleans up the data a little, but it fails to capture so much that’s relevant. Surely we should add it to fixed lines, or we overestimate the potential of developing countries?

Most importantly, what are we using the connections to do? There’s clearly a diminishing return to connectivity in terms of national accounting – when mobidensity increases from 30% to 40% it makes sense that it will have more impact on the welfare of the population than when it changes from 120% to 130%.

The statisticians at the ITU must long for the old days, when collecting figures involved compiling a set of numbers from incumbents and drawing a graph with a straight line through the middle of the dots.

The ITU’s “Core ICT Indicators” (current edition: 2010 revision) suggests that local surveys use a core list of 46 variables. We now measure who uses internet banking, how many people per 100 play computer games online, and the proportion of businesses using an extranet.

If the impact of a sector correlates with the number of statistics used to measure it, this is terrific news for telecoms. In 2005, the previous year in which the ITU reported its core indicators, there were only 41.

Demand ebbs and flows – but the opportunity to create really complicated graphs is growing, wherever you are.

Tim Phillips can be contacted at: