Roads go in both directions

Yesterday I wrote about declining regions, and noted that Japan could teach policy makers a lot about what to do, or what not to do, if they want to mitigate population decline in rural and peripheral regions. When it comes to public policy and affecting population change, a common refrain is that infrastructure (such as roads, rail, etc.) will increase population growth. The idea is that road facilitate movement of goods and people, increasing inter-regional trade and economic growth, creating more jobs and attracting more people. For instance, see this recent New Zealand Herald article about people commuting to Auckland from up to two hours (or more) away. However, those arguments miss a key point: roads go in both directions. So, while some people might live outside Auckland and commute in, others may well be going in the other direction. In fact, commuting against the flow of traffic isn't that uncommon.

So, is there evidence about the impact of roads on population and economic growth? Unsurprisingly, there is lots, including this recent article by Bishal Bakhta Kasu (South Dakota State) and Guangqing Chi (Penn State), published in the journal Population Research and Policy Review (it looks like it's ungated, but in case it isn't, the authors provide a good summary here). Kasu and Chi looked at the relationship between population growth and economic growth, and each of road density, railroad terminal density and the number of airports. Their data covered the county level in the US, over the period from 1970 to 2010. They found that, after controlling for various demographic, socioeconomic, and geographical differences between counties:

[t]he impact of railroads on population change is negative across the models, except for the 2000s. The impact of highways is not significant, but airports are significant in the 1980s and 2000s and from 1970 to 2010. The impact is positive, indicating that the number of airports contributes to population growth...

The impact of railroads on employment growth is negative across all periods, although the impact is not statistically significant in the 2000s. Highways have a statistically significant positive impact on employment change in the 1980s but a negative impact in the 1990s. Similarly, airports have a statistically significant [and positive] impact on employment change in the periods of the 1970s, the 2000s, and 1970–2010.

In other words, counties with more road density had no greater population or employment growth than counties with less road density. Railroads were generally negative for growth, and airports positive for growth. Of course, one of the problems with this study is that it shows correlation, not causation. That means that there could be a bunch of other variables (which weren't controlled for in the models) that affect both infrastructure and population or employment. However, this study adds to the existing evidence (reviewed in the Kasu and Chi paper) that shows inconsistent effects of roads on population change. As Kasu and Chi note:

...highway infrastructure does not produce any significant population or employment change; rather, it plays a facilitator role for the flow of people, materials, and raw and finished products.

Roads go in both directions.