Roman roads and infrastructure's impact on development

How persistent is economic development? When you look back over history, on average the places that are most developed today seem to be the places that were most developed a century ago. Or two centuries ago. Or a millennium ago. It's an interesting question. Another interesting question is, does infrastructure contribute to development? These two questions are, of course, closely related. Infrastructure is (usually) built to last a long time, and new infrastructure (such as roads, railways, etc.) are often built over top of, or next to, existing infrastructure.

This 2018 CEPR discussion paper by Carl-Johan Dalgaard (University of Copenhagen), Nicolai Kaarsen (Danish Economic Council), Ola Olsson (University of Gothenburg), and Pablo Selaya (University of Copenhagen), provides us with some guidance towards the answer to both questions (and the authors wrote an interesting article at Vox on their paper as well). The paper looks at the persistence of Roman roads, and whether Roman roads in 117 CE are related to modern-day road networks, as well as modern day development (as measured by night lights intensity, and population density, with both data sources from 2010). To give you some idea of the results, here's Figure 2 from the paper, which shows the Roman roads (red lines) and modern night lights for the area surrounding the Roman town of Lutetia (modern-day Paris):

Notice how the night lights cluster around the Roman roads? That demonstrates the persistence of development in the areas immediately surrounding the location of the ancient roading network. Dalgaard et al. analyse data across the whole Roman Empire (as it was in 117 CE). At that time, there was an estimated 80,000 km of paved roads. Controlling for geographic features, proximity to water bodies, distance to Rome and to the borders of the Empire, and the location of historical mines (as well as country and language fixed effects), they find that:

...historical road networks and geography account for about one fifth of the variation in contemporaneous road network across grid cells.

A one percentage point increase in Roman road density is associated with a statistically significant 0.15 percent increase in modern road density. In addition:

...Roman road density appears strongly associated with economic activity, both in the past and in the present... In our full specification we find that economic activity during antiquity rises by about 0.6 percent for every percentage point increase in road density; in the modern day context we find elasticities in the range 0.5 - 1 depending on the indicator.

Those effects are large. However, you would be right to be cautious at this point, because these are correlations, rather than causation. However, Dalgaard et al. are not done. They have an interesting natural experiment that provides some indication that the results may be causal. I hadn't realised this, but the parts of the Roman Empire in the Middle East and North Africa abandoned wheeled transport (in favour of caravans of camels) in the first millennium CE. That meant that the Roman roads were not maintained in the MENA region, unlike in Europe. So, if Roman roads cause development, and the observed relationship between modern development and Roman roads isn't instead driven by something else like the roads being placed where development would have happened anyway, then the roads should have an impact in Europe, but not in the MENA region. And indeed, that's what they find:

We find that in the MENA region, Roman roads lose predictive power vis-a-vis modern day roads. Moreover, Roman road density does not predict current day economic activity within the MENA region. In contrast, in the European region, where the roads were maintained, our baseline results carry over.

What can we take away from this? Infrastructure is persistent - it lasts a long time. And, it can have positive impacts on development.

[HT: Marginal Revolution, last year]