The importance of roads - the academic debate

I have blogged earlier expressing my disappointment with researchers who question the economic value of investments in basic infrastructure like roads and electricity. These studies are often held up as evidence against channelling scarce resources, especially aid money, into infrastructure. Instead proponents argue in favour of investing in education, health, nutrition, skilling etc. Two samples below.

This paper on rural electrification program in Kenya finds,

We do not find meaningful medium-term impacts on economic, health, and educational outcomes nor evidence of spillovers to unconnected households. These results suggest that current efforts to increase residential electrification in rural Kenya may reduce social welfare.

This paper on India's rural roads construction program finds,

There are no major changes in consumption, assets or agricultural outcomes, and nonfarm employment in the village expands only slightly. Even with better market connections, remote areas may continue to lag in economic opportunities.

This argument is based on the methodological biases among academic researchers. George Akerlof had pointed to the same, drawing the distinction between hard (quantitative and rigorous) and soft (qualitative) methodologies to examine development phenomena. He said that the former, which is the current preferred approach, generates 'sins of omission' which leads to under-estimation of economic impacts. 

In this context, The Economist points to a new work by Ana Paula Franco et al which examined the long-term development impact of the 30,000 km long Inca Road which stretched across several countries and which was built to collect taxes and deploy troops across the 10 m sqkm empire. 

To test if the Inca road, the Incas’ main thoroughfare, has boosted modern living standards, the authors split the map into small squares. For four indicators of welfare—wages, nutrition, maths-test scores and years of schooling—they compared levels from 2007-17 in squares crossed by the road with those in neighbouring squares not on its route. On every measure, residents of roadside squares fared better than those in adjacent ones, even after controlling for differences in such factors as the slope of terrain and the presence of rivers. Women gained more than men.

How did the road grant such long-lived blessings? The Spaniards used it to ship silver and turned the warehouses into profitmaking shops, often staffed by women (possibly inculcating more equal gender roles). This made land near the road unusually valuable, encouraging colonisers to settle there. The authors argue that Spaniards who moved in claimed legal title to their landholdings and built schools and new roads in the vicinity, creating enduring property rights and public goods. Today, the presence of the Inca road alone accounts for a third of the observed difference in levels of formal land ownership between dwellers on the road and those in nearby areas. It explains half the difference in the number of schools.

From the original paper itself,

We estimate the long-term effects of the Inca Road and find a positive and statistically significant relationship between the Inca Road and hourly wages, with residence within 20 km of the Inca Road increasing hourly wages by around 10.5% as compared to wages in other areas during the period 2007-2017. This effect is as large as that of an additional year of schooling. Along the same lines, we find a significant negative relationship between the Inca Road and child malnutrition, as attendance at a school located within 20 km of the Inca Road reduced the probability of being malnourished by 3.4 percentage points (a reduction of around 8%) in 2005. In addition, we find a positive effect between the Inca Road and educational outcomes, with residence within 20 km of the Inca Road increasing the length of schooling by 1.64 years, which equals 22% of the average years of schooling of our sample (7.41 years).

With respect to the channels of persistence of the effects of the Inca Road, there is a significant and positive association between the Inca Road and the current provision of two main public goods, primary schools and roads. We report that 20-km grid cells crossed by the Inca Road have an extra 18 km of road density, which equals 82% of the road density in our sample. Similarly, we find that grid cells crossed by the Inca Road also have an additional 48 primary schools, for an increase of 79% over the sample average. What is more, we find that there is a positive association between the Inca Road and current property rights. The size of this impact can be appreciated by taking into consideration the fact that only 26% of the members of our sample own their homes, as this makes the increase attributable to the Inca Road equivalent to 31%.

Finally, we find a positive association between the Inca Road and female labor outcomes. Women who live within 20 km of the Inca Road average 1.63 more years of schooling than women who live farther from the road system. This is a sizeable increase (22%) over our sample mean (7.4 years), and one which outpaces the increase for men. Furthermore, there is also a positive association between the Inca Road and different measures of female intra-household bargaining power. Residence within 20 km of the Inca Road reduces the likelihood of being a teenage mother by 1.4 percentage points. It also increases the likelihood that women are making health-related decisions for the household by 5.2 percentage points and the likelihood that they are making decisions about high-value purchases, such as buying a home, by 5.4 percentage points.

It would be great if someone could do a study of the Grand Trunk Road or numerous other major roads built by old Indian empires, like this study of the impact of the Indian Railways by Dave Donaldson.

This is yet more reminder, if they were needed, to the western academic world engaged in development about the importance of basic infrastructure like roads and electricity.

Update 1 (07.11.2021)

Santanu Chatterjee and co-authors find differential impacts on formal and informal firms from the development of national highways,

We find that formal firms that are located in districts along the planned route of the GQ/NS-EW corridor are, on average, 9-10 percent more productive than firms that are not on the planned route. By contrast, informal firms on the highway corridor are no more productive relative to their off-route counterparts. Further, we also find that formal firms located in districts 0-30 miles from an upgraded or completed section of the corridor produce 2-4 percent more for every additional year after the completion of the project. The corresponding benefit for informal firms is much smaller, at around 1 percent. Further, using quantile regressions we find that while the productivity benefits of public investment are spread evenly across the size distribution of formal sector firms, they are strictly increasing in firm size for the informal sector. Additionally, smaller informal firms are significantly disadvantaged by the highway upgrades: public investment benefits not only the larger firms in each sector, but also formal firms much more than informal ones.

As to possibly why this happens, they write,

We use quantile regressions to examine whether public investment has a differential impact along the size distribution of firms in each sector. Here, we find evidence that the complementarities generated by an increase in public investment lead to large firms crowding out the output of smaller firms, both within and across sectors: large firms in the informal sector tend to crowd out smaller firms within that sector and, formal sector firms also tend to crowd out small informal firms. Intuitively, large informal firms tend to have a higher capital intensity in production than their smaller counterparts, and formal sector firms also tend to have higher capital intensity than their informal counterparts over all. As such, public investment benefits not only larger firms in each sector, but also formal firms much more than informal ones. This can help explain why we are unable to find any positive and significant association between public investment and informal production for the average firm in our sample.