Anabella Abadi
Anabella Abadi es ejecutiva de la Dirección de Evaluación de Impacto y Aprendizaje de Políticas de CAF
Before building or remodeling our homes, we often seek expert advice and try to find out how similar projects turned out in neighbors’ or friends’ homes. This is a logical strategy, because we want to protect our property, and the project will involve a considerable investment of resources. As citizens, we should expect a similar approach from our—local or national—governments when undertaking a major infrastructure project.
But measuring the impact of infrastructure projects can be difficult, as their characteristics and scale depend on particular features of each locale, which makes it difficult to establish a counterfactual to compare and isolate the effect of such projects. Hence the importance of proposing and replicating creative options to overcome the difficulties associated with impact assessment of an infrastructure project.
For example, as it is difficult to establish a potential counterfactual, especially when we have a single treated unit with very few untreated observations (recurring scenario in infrastructure interventions), a “synthetic control” can be established, which is statistically similar to the treated unit— provided there is retrospective information available on relevant variables for a sufficient period before the intervention. This requires taking the weighted average of pre-intervention results from a set of untreated units to construct a hypothetical counterfactual that simulates how the variables of interest of the intervened unit would have evolved if it had not been intervened. This method is being used for the impact assessment of the revamping of the Belgrano Sur Train, a project co-funded by CAF to improve mobility and connectivity of transport services in the Buenos Aires Metropolitan Region towards the central area of the city. To establish a synthetic control, six branches of the Roca, Urquiza, Mitre and Sarmiento railways were identified, which will not be intervened in the coming years, leveraging also existing sources of information. Note that synthetic control will be used not only to assess the impact of completed projects, but also to determine the number of households close to improved railway services and number of users, who will be surveyed.
Another difficulty is the cost of surveying information to measure the impact of infrastructure projects. In this regard, artificial intelligence or machine learning solutions are very useful. Georeferenced data (GPS) could be used, for example, to monitor travel times and route compliance by certain transport units. We can even use satellite photos to build proxy measures. An assessment by Mitnik, Sanchez and Yañez-Pagans (2018) used the brightness of satellite night light as a proxy measure of the impact of transport infrastructure investments on economic activity in Haiti. This allowed surveyors to estimate brightness increases that were deemed to reflect GDP growth, with impacts focused on 2-km buffers around intervened roads.
Finally, the study emphasizes that sometimes difficulties can be turned into opportunities. For example, in 2006 the authorities of the metropolitan area of Acayucan (Veracruz, Mexico) decided to advance a program to surface residential streets for the first time. Since budgetary and time constraints would only allow to work on 28 of the 56 identified streets, the city council approved the impact assessment of the project and the randomization of the 28 streets to be intervened. The assessment by González-Navarro and Quintana-Domeque (2016) was based on primary information survey, and concluded that first-time paving of residential streets in low-income neighborhoods of developing countries adds value to the properties around the interventions, which directly benefits their owners. In addition, in just two years, this rise in real estate wealth translated into a significant increase in household appliance purchases, home improvements and vehicle ownership.
Latin America faces significant infrastructure challenges, especially considering that of the 19 countries in the region among the 140 countries in the 2018 Global Competitiveness Report, Chile is best ranked at 41st place, followed by Mexico (49th) and Ecuador (59th). Of the other countries, 11 are ranked between 66th-98th, and five below 100. Haiti stands out as the nation with the worst infrastructure in the entire ranking. This suggests that large investments are needed to improve competitiveness of regional infrastructure, which should ideally involve assessments to determine whether projects have the expected impact and are cost-effective.
In the end, Latin America is our home and we must ensure that the works financed for our progress generate yield the best possible results.
