Connecting Economies

Connecting Economies

I was inspired by the impressive blog run by DiViNetworks, to investigate the developmental impact of telecoms. It’s a question that’s regularly asked in the industry: do economies grow when their telecoms infrastructure improves?

I was inspired by the impressive blog run by DiViNetworks, to investigate the developmental impact of telecoms. It’s a question that’s regularly asked in the industry: do economies grow when their telecoms infrastructure improves?

DiVi inspired me by doing a simple regression: average download speed in a country against GDP per capita (download speed is provided by netindex.com, and the output figures came from the World Bank). DiViNetworks’ analysis, as most similar plots do, shows a strong correlation between output and connection speed - either connected household speed or average speed per house.

The blog raises the possibility that the relationship may be causal – fast internet may help a developing country grow. This is a strong claim, but it would be important if it were true. Correlation is not causation, though. When you have a relationship on a graph, it might mean that A causes B, B causes A, or that A and B both happen when C exists. I downloaded the World Bank and netindex.com statistics to have another look.

The first point: the relationship between output per capita and average download is strong: in a sample of 154 countries, the correlation was about 0.7, with significance at better than 1%. It’s about the same level of correlation as if we plotted download speed against the percentage of GDP spent on research and development in the country, which you would imagine would have a clear impact on economic growth.

But we seek to measure the relationship between telecoms quality and development in a country. Development is a difficult property to measure, because it is more than wealth. Money gives a country the means to escape the poverty trap, but measuring GDP alone ignores at least two important things: how the money is spent, and which people in the country get the benefit. We generally use GDP per capita as a convenient proxy for development, but in the words of economist Amartya Sen, this confuses well-being with being well off.

The most common alternative measure of development is the Human Development Index (HDI) calculated by the UN, which aggregates indices of GDP per capita, education and health. So it measures not just whether a country’s economy is growing, but whether it uses that wealth to improve human development (HD). When categorising quality of life, the UN divides its member countries into four quartiles: very high HD, high HD, medium and low.

It’s relatively simple to do a regression of download speed against the level of the HDI to establish the strength of this relationship. First the good news. The results are even stronger than for output per capita. In words: the higher the average download speed, the higher the position we expect the country to occupy in the global HDI.

Clearly all three possible models of causation are still possible. Fast download speed – a proxy for good data communications infrastructure – may be important in helping countries to develop, so telecoms lifts people out of poverty. On the other hand, developed countries may have more money to spend on telecommunications infrastructure (or had it in the past – there’s almost as strong a relationship between current download speed and the level of development in 1990 and 2000, when a lot of fibre was laid in rich countries). Therefore, development means more people and companies pay for broadband. Or the third interpretation, both are caused by an unmeasured external factor. For example, countries with strong financial and legal systems are likely to be high development countries. They also encourage entrepreneurs and FDI, who can get credit and protect their investments. Those entrepreneurs and investors create broadband.

In reality, finding a correlation starts the discussion rather than settling it. There’s a bit of all three, and many more factors besides – in statistical terms, this analysis explains only about half the relationship.

But there’s one further point in this simple regression: there’s no evidence of any relationship between these properties for poor countries.

On a scatter diagram of download speed against development (or GDP per capita) you see a cluster of dots near the origin, and a fan out to the top right. The relationship between download speed and development (or GDP per capita) is strong and positive over the whole sample. But if you live in a country with less than median HDI – as seven out of 10 people on the planet do – there is no statistically significant relationship between either your country’s level of development (GDP per capita) and the download speeds you’re getting. The whole-sample correlation is entirely based on what happens in rich countries, where moving from “quite rich” to “stinking rich” (note: the World Bank does not use these descriptors) goes hand in hand with getting much faster downloads. This data tells us nothing about sub-Saharan Africa or the poorer economies of Asia or Latin America at the moment, where B doesn’t cause A, and A doesn’t cause B; probably because they have to deal with problems C to Z first.

So while the DiViNetworks offer of half price capacity must be interesting to operators in countries with little purchasing power and huge potential for rapid development in the future, there’s no evidence from this data that it will lift those countries out of poverty.

More from across our site
Gift this article