Evan Soltas
Mar 23, 2015

Where Is the Internet Boom?

Facebook Connections
The Internet is arguably the most important technology to emerge since 1990. Has it done anything for economic growth?

It seems it should have. Take a look at the growth of e-commerce. Or the rise of mobile banking in the developing world. Or the market capitalizations of Internet stocks in the Nasdaq. Everywhere you look, you see the Internet.

Except, as I will show, in economic data.

Now, skepticism about the economic effects of digital-age technology has quite a track record. A bad one. "You can see the computer age everywhere but in the productivity statistics," the economist Robert Solow famously said in 1987. Erik Brynjolfsson called it the "productivity paradox" in 1993.

And then, in the late 1990s, the IT-driven productivity boom finally arrived. Output per man-hour soared in the U.S. Also, economists who looked at investment in computers at the firm level began to find real, substantial productivity gains. Those findings have been supported by a later generation of research as the pace of IT gains slowed.

Just as Zvi Griliches found with the use of hybrid corn in the American Midwest, the adoption of the Internet has followed a logistic curve -- slow in the beginning, fast in the middle, and then slow at the end. The technological diffusion of the Internet, also like hybrid corn, did not happen evenly across countries. Some moved online faster than the others.

In the graph below, I've plotted the share of population with Internet access in the median country for the years 1990 to 2013, and then the 25th-percentile (i.e., low-Internet) and 75th-percentile (i.e., high-Internet) countries. The median country, as of 2013, has just over 40 percent of its population with Internet access. The gap between the 25th and 75th percentiles, called the interquartile range, is about 50 percentage points in 2013. The data come from the World Bank.

Here's the problem. Countries embraced the Internet at widely different paces, as the graph shows. If the Internet has a meaningful effect on economic growth, then the relative rates of adoption of the Internet should show up in relative rates of growth. Countries that moved online quickly should have grown more quickly than similar countries that moved online more slowly. Yet they did not, at least upon a first glance at the data.

The graph shows the average annual growth rates in GDP and Internet access by country between 1990 and 2013. There is no relationship.

There are more sophisticated ways of doing this analysis, but so far, they have ended up in the same place. (For example, if you predict GDP growth rates based on the level of GDP and past growth, the level of Internet access and the rate of Internet-access growth usually have no additional effect.) I cannot seem to find any effect of Internet access on GDP.

That's why I am making my dataset freely available here. It includes about 5,000 observations on Internet access and GDP for the roughly 200 countries for which the World Bank has data. (If you're interested in working on this, please do send me an email.)

Maybe it's the case that the Internet has an effect on productivity but not the total level of output. In a recent paper, Daron Acemoglu, David Autor, and other top economists showed that the IT boom of the 1990s resulted in higher productivity -- but only because IT-intensive manufacturing firms cut employment faster than they cut output. There was no IT boom in terms of output.

The next step, then, is to find reliable data on productivity that are consistent across countries and span the relevant period. I am currently looking at getting these from the ILO, but they will be iffy at best.

Alternatively, maybe we're just in the Internet's own "productivity paradox" moment. Those who doubted of the economic importance of personal computing in 1993 had to change their tune only a few years later. (I should note here that research at the firm and local level and for the US has found positive effects of the Internet on output.)

Yet, even if we do end up finding some positive relationship between Internet access and GDP at th country level, it would be premature to call it causal. Intuitively, a country that is growing rapidly might choose to invest in Internet access, and richer citizens may buy Internet-enabled phones. The process of investment and economic growth is importantly circular.

It will take more careful approaches to measurement to move from "countries that went online faster grew faster" to "the Internet causes economic growth." Yet, right now, we can't even say the former. So much for the Internet's effect on growth.