Here at the National Infrastructure Commission, it’s vital that we stay on top of the latest thinking – and because we all take a genuine interest in infrastructure it’s something we enjoy as well.
So, we take it in turns to share the most interesting and thought-provoking academic papers on the economics of infrastructure we have read – and now through this blog, once a month, we’ll share the very best of them with you. Of course there’s lots more to infrastructure than economics and lots more to the evidence than academic papers, but it’s an important place to start.
They don’t have to be new, but they do have to be worth reading and relevant to the UK’s strategic infrastructure.
Since we began we’ve had papers on everything from Gibrat’s Law and the British Industrial Revolution, building codes and energy efficiency, urbanisation and public transport, to one that looks at how the costs of lighting have changed over the last seven centuries (short answer: they have fallen dramatically).
To kick things off – April’s ‘Paper of the Month’ is a paper from Pedro Bonn and Jenny Ligthart which asks what have we learned from three decades of research on the productivity of public capital?
The paper is a ‘meta-analysis’ of the literature estimating the impact on growth of public infrastructure spend. The headline finding is an elasticity of GDP with respect to public capital stock of around 0.1: roughly speaking, doubling your public infrastructure stock increases GDP by 10%. That’s a decent effect, though not dissimilar to what you would expect from capital in general in the economy. The paper summarises 578 estimates from 68 studies, which shows how widely studied this question has been over the last 30 years. One important point to remember is that the returns to infrastructure networks are typically much higher when new networks are developed – the first railway or the introduction of electricity or broadband – these very high gains are not reflected in this kind of analysis. You can’t compare GDP today to what it would be without electricity.
This is the most ‘headline’ measure of the effect of infrastructure on growth that there is, so for the NIC it’s a vital piece of research. It’s striking just how many studies there have been, since this literature was launched, essentially by a paper from Aschauer in 1989. By bringing all these estimates together in a single analysis, it’s possible to extract a reasonably coherent overall story from a wide range of differing studies, which show very different individual results. Meta-analysis offers a way of attempting to correct for the different methodologies and data sets that different studies use, and the tendency for more striking results to get published more easily, which can make comparisons between results very hard.
The paper also finds a stronger effect for capital at regional rather than central government level. It’s a little difficult to interpret this for the UK. Many of the studies are from the US, so the scale is different. But it brings out the point that the impact of infrastructure is context-specific: not all projects have the same returns and making better allocation decisions can make a big different to impacts.
The downside of such a ‘headline’ measure, however, is that is loses subtlety: which kind of projects are most effective and where? And the measure of infrastructure is a financial one: the public capital stock. A direct measure of infrastructure quality would be much more interesting. Trying to get better measures of infrastructure performance is one of the things we’re looking at in the National Infrastructure Assessment. And, of course, the impact on GDP is not the only thing that matters for infrastructure.
What there is, however, is a near inexhaustible supply of academic research into the economics and impact of infrastructure – and keeping abreast of that is partly what this blog will be about.
So each month we will publish a ‘Paper of the Month’ right here on this blog, along with the best of the rest, and we hope you will find it as interesting as we do.