Principal investigator: Eduarda Miller de Figueiredo
Article title: RURAL ROADS AND LOCAL ECONOMIC DEVELOPMENT
Article authors: Sam Asher and Paul Novosad
Location of the intervention: India
Sample size: 11.432 villages
Sector: Labor market
Type of intervention: Effects of road construction
Variables of interest: variable created by subtracting the population limit from the village's treatment limit
Evaluation method: Discontinuous Regression – fuzzy
Policy problem
Worldwide, nearly one billion people live more than 2 kilometers away from a paved road, with one-third of these people in India (World Bank Group, 2016). Seeking a solution to this problem, the Indian government launched a road program to finance road construction, the Prime Minister's Village Road Program (PMGSY).[1].
Literature suggests that road construction is associated with increased growth in both agricultural and non-agricultural sectors, as well as poverty reduction. This is because paved roads would lower input costs, leading to higher production costs and increased production of non-agricultural goods, which would be offset by higher wages. The price variation between these markets is the main argument for how rural roads help develop the rural economy, both domestically and internationally.
However, if the villages have few exports, this may result in low demand for transportation, meaning operators would not be willing to pay the fixed cost to reach the village. Therefore, the problem of high transportation costs would persist even after the construction of paved roads.
Assessment context
The PMGSY is based on the idea that poor roads are the biggest obstacle to rapid rural development (National Rural Roads Development Agency, 2005) and, by 2015, had benefited from the construction of roads linking nearly 200.000 villages, at a cost of nearly $40 billion.
The program was implemented in locations that met the population limits: villages with more than 1.000 inhabitants in 2003, with more than 500 people in 2007, and with a population greater than 250 after 2007, according to data from the 2001 population census. By 2015, more than 400.000 kilometers had been built at a cost of nearly 40 billion dollars.
Figure 1: Data on road construction through PMGSY by year.
Figure 1 presents data on road construction by year, where it is possible to see that construction is insignificant until 2001, increasing in subsequent years until reaching a peak of 11.107 villages receiving roads in 2008, after which the number of roads decreases again.
Policy Details
For the research, two population limits were used in the villages: 500 and 1.000. Villages exceeding these limits had a 22 percentage point higher probability of receiving a paved road. Thus, the authors overcame the obstacle of the correlation between road construction and the economic and political characteristics of the locality, allowing them to estimate the causal impact through the regression discontinuity (RD) method.
A comprehensive database was built containing information on all businesses and families in rural India, with geographic and socioeconomic data on individuals, village identities and characteristics, and firm characteristics.[1]Using the data, he built a proxy of consumption, which allowed testing the impacts of roads on the average projected per capita consumption and the distributional effects.
The impacts of infrastructure investment are complex to analyze because they have a high cost and a large potential return, so allocation is not random on the part of policymakers. Therefore, there is not a sufficient number of treaties and controls, but the authors managed to overcome this obstacle by combining near-random variations in program rules with georeferenced village-level data.
Since the rules for population boundaries do not change discontinuously, the probability of treatment will increase discontinuously within these boundaries, making it possible to estimate the effect of new roads through regression discontinuity. fuzzy, in which the dependent variable will be the subtraction of the treatment limit from the village population. Thus, the rd estimator fuzzy calculated the average local effect of the (late) treatment of receiving a new road for a village with a population equal to the threshold (+500, +1.000).
The indicators used as control variables were the presence of a primary school, a medical center, and electrification in the village. log of the total area of agricultural land, the proportion of agricultural land that is irrigated, the distance from the nearest densely populated city, the proportion of workers in agriculture, the literacy rate, the proportion of inhabitants belonging to the regular caste, the proportion of families with their own agricultural land, the proportion of families who are subsistence farmers, and the proportion of families who earn more than US$4 per month.[2].
Figure 2: Effect of the probability of new roads in 2012
Figure 2 shows the proportion of villages that received new roads before 2012 in each population range. A discontinuous increase in the probability of treatment is observed at the threshold, where, upon crossing this threshold, the probability of treatment increases by 21-22%.
Results
The results show a large positive effect on the availability of transportation services, a significant reallocation of labor outside of agriculture, and a smaller positive effect on employment growth in small businesses. These results can be observed in Figure 5, which presents graphical representations for each variable, demonstrating that significant effects appear in transportation and labor outflow, but little impact on businesses.
Figure 3: Effect of roads on the main results indices.
A new road causes a significant increase of 12,9 percentage points in the availability of public bus services. More expensive modes of transport, such as taxis and vans, did not experience significant growth, however, cheaper private motorized transport did (autorickshaws) showed growth. Therefore, it was possible to observe that the new roads significantly affect connections with the external market.
The results for occupational choice suggested that new roads cause a 9,2 percentage point reduction in agricultural workers and a 7,2 percentage point increase in non-agricultural workers. Since land is the main input for agricultural production, analyzing the results for this index revealed that a new road does not significantly change the proportion of landless families.
Regarding the characteristics of the individuals, the results suggest that men are more likely to leave agriculture, as are younger people. This result could exemplify the physical advantage men have in non-agricultural work or an attitude against women working away from home, as Goldin (1995) states; however, the estimates for the average control group for male and female workers are very close.
It was also estimated that there was a 27% increase in employment in non-agricultural firms and a significant 33% growth in retail in response to a new road, which creates an average of 4,2 new jobs in the village during road construction. Thus, the authors believe this is evidence suggesting that roads facilitate access to the external labor market more than job growth in village businesses.
The authors found no evidence of increases in ownership of mechanized farms or irrigation equipment, nor did they observe any effect on earnings, assets, or consumption. In short, according to the authors, the results demonstrate that there is no effect on the structure of agricultural production with the new roads.
Lessons in Public Policy
The construction of new roads has substantial impacts on economic activity, facilitating the reallocation of labor from agriculture; however, it does not generate major economic changes. In other words, rural roads increase transportation services and reallocate labor away from agriculture, but do not bring about significant changes in village businesses.
Reference
ASHER, Sam; NOVOSAD, Paul. Rural roads and local economic development. American Economic Review, vol. 110, no. 3, p. 797-823, 2020.
[1] Pradhan Mantri Gram Sadak Yojana.
[2] Names of the databases used: Socioeconomic High-resolution Rural-Urban Geographic Dataset (SHRUG), PMGSY, Socioeconomic and Caste Census (SECC) and Below Poverty Line (BPL).
[3] Approximately 250 INR (Indian rupees).
