Public goods are REALLY good: thousands of years later, the Roman roads are still paying dividends

[L0ki]

@Keith_McClary

Roads and population centers are located for geographic reasons. Many of those reasons are the same today as in Roman times, so these correlations are not surprising.

@knappa

Yeah, somewhat difficult to analyze the counterfactual here.

@edrowland

I’m all in favor of roads. But one has to wonder whether they haven’t confused cause and effect. The more reasonable explanation is that nice places to live cause population concentrations. And population concentrations follow previous population concentrations for the most part, because that’s where the cities have always been, and because they’re still nice places to live. And Roman roads just threaded between the nice places to live because that’s where the people were.

@fuzzyfungus

I wonder if a suitable comparison with regions that never fell under enough imperial influence to get the Roman infrastructure would do?

Might fall prey to the “never fell under imperial influence because it was a hostile environment” problem in a similar manner; but might work.

@zot23

I’m going to call (a small amount) of BS on this “study”. I don’t doubt anything they found, but the Romans likely built roads between places that already had good conditions and largish populations. [snip…]

I appreciate that everyone’s available time for reading stuff is limited but it does often help to read the linked article rather than just the BB summary - the authors (amazingly for a bunch of academics in the field of the reflected sound of underground spirits) did manage to think of that point:

A key challenge in identifying the effects of ancient infrastructure is filtering out the effect of underlying (geographic) factors that may have influenced both Roman road density and modern-day outcomes. Another concern is how to separate the effect of ancient infrastructure on economic development from the potential broader influence of the Empire (e.g. via institutional or cultural channels).

We address these issues in several ways. First, we focus exclusively on observations (i) within the boundaries of the Roman Empire, and (ii) that were ‘treated’ by at least one road. By focusing on areas connected to the network, we reduce the risk of omitted variables bias, and the risk that our results convolute the impact from the legacy of Roman rule more broadly (e.g. Landes 1998). To further partial out a potential Roman legacy on contemporary outcomes, we control for country fixed effects as well as language fixed effects, which help us to deal with institutional and within-country cultural variation (e.g. Andersen et al. 2016). Second, based on the literature on Roman road construction and our own formal tests thereof, we control extensively for potential geographic confounders throughout the entire analysis. Third, we exploit a natural experiment to which we return below.

There is then also a whole long section on how they compared the results in Europe with results from the Middle East and North Africa (MENA) where use of wheeled transport was apparently largely abandoned “during the second half of the first millennium CE”.

They posit that:

Consequently, in the MENA region, Roman roads should be a weak predictor of contemporary roads density and, by extension, Roman roads should also be a weak predictor of contemporary comparative development. In contrast, within the European region where wheeled carriages were in use throughout the period, one would expect more maintenance and therefore more persistence in road density and, by extension, Roman roads should be a stronger predictor of contemporaneous comparative development.

And come to the conclusion:

When we split the data and estimate our baseline model on the MENA and the European samples separately, this differentiated influence from Roman roads indeed emerges. Roman road density is not a statistically strong predictor of current road density nor economic activity within MENA, despite the fact that Roman roads do predict comparative economic activity prior to the abandonment of the wheel. Within the European region, Roman roads not only predict current infrastructure, but also ancient and current economic activity.

Whether one agrees with their conclusion, one can’t simply say they didn’t think about this obvious point.