Economics 23/01/2010: Knowledge economy infrastructure

Some interesting data from a study "Broadband Infrastructure and Economic Growth" by Nina Czernich, Oliver Falck, Tobias Kretschmer and Ludger Woessmann, CESIfo Working Paper 2861 published in December last year that provides good comparatives for Ireland relative to the peers in terms of what I would call rudimentary 'Knowledge Economy' infrastructure -

• the relationship between physical capital and knowledge-related capital (broadband penetration and education); and
• the relationship between GDP per capita and the above

all taken over a long period of time (1996-2008).

First, broadly, the findings of the study itself: "We estimate the effect of broadband infrastructure, which enables high-speed internet, on economic growth in the panel of OECD countries in 1996-2007. Our instrumental-variable model ... [shows] voice-telephony and cable-TV networks predict maximum broadband penetration. We find that a 10 percentage-point increase in broadband penetration raises annual per-capita growth by 0.9-1.5 percentage points. ...We verify that our instruments predict broadband penetration but not diffusion of contemporaneous technologies like mobile telephony and computers."

Interesting - a 10% increase in broadband penetration ups the growth rate by 0.9-1.5%. In other words, to get a 10% increase in GDP per capita out of a 10% rise in broadband penetration requires 6.4-10.7 years. Not a bad return. The problem here is that, of course, the starting levels from which this effect is measured are low, so the law of diminishing marginal returns has to kick in somewhere.

I took their data and run through some of it in a very crude way to see if I can glimpse other interesting aspects. Here are the results.
Maximal (for the period GDP per capita, PPP-adjusted) with 2 standard-deviation 'candles' around it. Notice two broadly defined groups of countries: Overperformers (including Ireland) and Underperformers. Now, I know - I shouldn't be using GDP here, but I am not about to make a statement about the actual 'wealth' or 'riches' of Ireland, so GDP will do.

Next, take a look at scatter plot relating GDP per capita to two measures of communications sector performance: broadband penetration for 2008 (the end score, if you want) and starting point measure (voice telephony penetration back in 1996).
It looks like GDP per capita in the end is much more responsive to increases in broadband penetration than to the starting position. In other words, economies with low legacy stock of communications seem to be catching up through broadband penetration improvements. Is this suggesting that a country can leapfrog weak communications sector legacy by jumping straight into broadband age? Well, sort of. The problem here is that we do not separate out the twin effects of growth in broadband penetration (much higher for countries doing leapfrogging) and simultaneous growth in voice telephony penetration (also likely to be much higher for countries doing leapfrogging).

A very revealing chart next:
Let us take physical capital as a share of GDP and compare its effects on overall GDP per capita, against the same effect being induced by education. What is unambiguous is that countries with higher physical capital base share of GDP tend to have lower GDP per capita. How come? Because they are physical capital-intensive, i.e their production is stuck in the late industrial age. Countries with higher education are more labour-intensive and especially skilled-labour intensive, and thus have higher GDP per capita.

Note Ireland. It is relatively poor in physical capital per GDP and yet relatively rich in GDP per capita. Why? Because we do have a modern economy - an economy where value is added through human capital side (of course this happens much more on the side of MNCs, where transfer pricing is used to import, artificially, human capital-intensive value-added, but it also happens in services economy, in our IFSC, etc). And yet, our education measure is far from being impressive.

The gap between our unimpressive levels of education and the levels of education consistent with the 'average' OECD pattern of relationship between education and GDP per capita, to me, clearly shows the importance of transfer pricing in our GDP figures. This gap is captured here by, in effect, showing that our capital and human capital stocks cannot support our GDP fully!

Here is more detailed view of our physical capital stock relative to our education (human capital stock).
Ouch. We are an outlier precisely in the direction suggested by the gap identified above. Note that moving to a 'Sweet Spot' of highly productive economies with significant rates of utilization of human capital requires both - more physical capital formation and even more education. Also note just how inefficient is the stock of education in the upper 'bubble' group of countries that includes all Nordics, Japan and France. These countries are simply not being able to derive the same returns to education in terms of GDP per capita as the 'Sweet Spot' nations.

So here is a question no one is asking - is there such a thing too much of education? Is there an inverted U-curve for the relationship between education and income, whereby too smart for its onw good society leads to suboptimal levels of growth? After all, since the 1990s we are seeing an emerging trend in the developed world whereby the new generations of slackers are increasingly composed of highly educated people...

This is not an argument out of the blue - take example of a potentially 'too livable city' concepts discussed in a brilliant article here. Can the same happen to the 'too-knowledgeable-economy'?

Ok, couple more charts on the same point. Broadband penetration is positively correlated with capital formation... Hmmm. This might reflect the fact that higher stock of capital imply better infrastructure through which broadband can be delivered. The relationship is not very strong, though.

And there is an even weaker, and negative, relationship between education and physical capital. This negative coefficient of correlation does suggest, though, that we are in the early stages of the process whereby physical capital takes second seat to human capital in characterising modern economy. If so - good news for 'Knowledge' economists out there - machines do not possess knowledge. People do. But it is also bad news to all social engineers out there who still think technocratic management of economy/society is possible. Knowledge requires heterogeneity and creativity. And these are antitheses of planning and policy-driven controls and incentives.

Far from being dead, the age of Friedmans' Freedom to Choose is only dawning!

And the final point: education and broadband infrastructure are much more strongly (almost 4:1) positively correlated with each other than they are with physical capital.
This, of course, can be interpreted as a warning to the folks interested in restricting the freedom of people to communicate. If China, and other countries that impose controls on internet, want to have a 'Knowledge'-intensive, modern economy, they will have to deliver real (i.e. free of political ideologies and biases) education and meaningful (i.e. free of political 'bottlenecks') knowledge infrastructure (in this case, broadband).

If they don't, the risk is they will end up being physical capital giants - countries where the world does its 'dirty work' of mass manufacturing widgets...