Tuesday, September 17, 2013

Does Broadband "Cause" Economic Growth and Higher Household Income?

Correlation is not causation. In other words, even when there is a relationship between two events, trends or occurrences, that does not mean one causes the other in a direct sense. Still, a study sponsored by Ericsson claims higher speeds "cause" higher household income.

With all due respect for Arthur D. Little and Chalmers University of Technology, which conducted the study, that’s a logical fallacy. No doubt, household income and use of broadband access are correlated.

But it is just as likely that higher income “causes” purchases of faster-speed broadband access, than that broadband access “causes” higher income. In truth, there is clear correlation, not “causation.”

Correlations are different from the classic method of making predictions: trying to find out "why" or "what caused something." That's known as "causation." But while correlation is not causation, correlation still is useful.

A correlation does not mean that one thing caused the other, but “big data” can yield important insights. For instance, after analyzing billions of records, MasterCard learned that when someone buys gas at around 4 p.m., the person will likely then spend $35 to $50 on groceries or at a restaurant. The gas purchase didn't cause the food purchase, but the two events are correlated.

That doesn’t invalidate the importance of the insight. In this instance, correlation is as important as causation, one might argue.

But it arguably is incorrect to see a causal relationship when there is clear correlation, but no clear evidence of causation. In fact, one might plausibly argue the inverse case, that is is higher income that causes purchasing of faster and more expensive broadband, not broadband causing higher income.

The study argues that, in OECD countries, the greatest measured increase in income was for households that upgrade from 0.5 Mbps to 4 Mbps, at around US$322 per household per month.

On average, upgrading from 4 Mbps to 8 Mbps yielded US$122 per household, per month. For Brazil, India and China, upgrading from 0.5 Mbps to 4 Mbps is projected to yield US$46 per month, per household, the report claims.

The Ericsson-sponsored study does attempt to control for factors known to influence income (age, sex/gender, education, household size, skills and type of occupation).

That there is a correlation between gross domestic product and Internet access speed does not seem out of place. Whether there is causation is the questionable assumption.

"Results are in line with our previous study that quantified the impact of broadband speed increases on the gross domestic product of 33 countries, as well as a slew of other studies we reviewed,” said Sebastian Tolstoy, VP Radio Business development and Strategy, Ericsson.

Few might disagree that “broadband access has a positive effect on the economy.” What is contestable is that broadband causes economic growth. Some would argue it is more likely economic growth causes faster-speed broadband to be deployed and purchased.

One might argue that mobile services likewise cause economic development. Others would argue mobile services are most plentiful and most advanced in densely populated urban areas where there are high concentrations of customers because economic growth is robust in those regions, compared to rural regions, for example.

That is why new mobile networks are built first in urban areas, not rural areas, and why most mobile service revenue is generated in urbanized areas. That is where the economic activity, higher incomes and wealth tend to be located. In other words, it makes as much sense to argue that economic growth “causes” mobile adoption as to argue that mobile “causes” economic growth. They are correlated, clearly, but not causal in a clear way.

The evidence is building for broadband speed as a driver of economic growth,” said Martin Glaumann, Partner at Arthur D. Little.

The study can show that increased broadband speed increases income,” said Erik Bohlin, professor at Chalmers University of Technology.

No comments:

"Tokens" are the New "FLOPS," "MIPS" or "Gbps"

Modern computing has some virtually-universal reference metrics. For Gemini 1.5 and other large language models, tokens are a basic measure...