The Open Data Challenge is an extraordinary opportunity for graduate students to submit papers that use open data collected for the MCC's independent evaluations. As the winner of the first Open Data Challenge, I would like to share my experience on how having access to MCC's open data has allowed me to improve the quality of my research.
On February 2, 2016, the Millennium Challenge Corporation hosted an event named Open Data: Big Impact, to which I was invited to present the paper I submitted to the Open Data Challenge. This opportunity to visit the MCC and learn about its projects has allowed me to go inside the work of a foreign aid agency and witness a little bit of how it runs its operations. It has also allowed me to discuss my research results with the people who care about them the most: the funders. The comments that I received during this event and the ability to access additional data have allowed me to extend the depth of my analysis in a way that would not have been possible without the MCC's commitment to making their data publicly available.
Rural Roads and Improved Access to Markets
In most developing countries the condition of the rural road network is substantially lower than that of the main roads. Due to the poor condition of rural roads, rural households face increased travel times and transportation costs and are isolated in terms of access to social and economic facilities. For that reason poverty levels are likely to be significantly higher among those living in remote or isolated locations. Most national household survey data show a significant regional dimension to the incidence of poverty, with greater proportions of poor households in remote, less favored, or weakly integrated areas.
The improvement of rural roads is broadly recognized as a fundamental precondition for the development of rural areas. By providing mobility, transport enables rural populations to access markets, and to benefit from social and educational interactions, health care and technology transfer. Improving access to markets encourages rural farmers to modernize with fertilizers, mechanized equipment, and new seed varieties, which, in turn, raises yields, lowers unit costs, and increases demand for inputs. Rural transport also provides access to labor markets and thus the opportunity to earn non-farm income.
Roads play a central role in rural development, but despite large amounts spent on them, little is known about the size and distribution of their benefits. That is why I was determined to make this the topic of my dissertation. Before finding out about the MCC's compact with the Government of Nicaragua, I considered working with data from a few other similar projects. One in particular, from Vietnam, provided a lot of data related to a rural road rehabilitation funded by the World Bank, but it provided very little background information about the project. In contrast, the Open Data Challenge did not just provide me open access to survey data from Nicaragua, but a lot of background information too, in particular, an independent impact evaluation conducted by Dr. Jonathan Alevy, and the possibility of having someone to contact in case I had any specific questions.
The improvement of rural roads is broadly recognized as a fundamental precondition for the development of rural areas. By providing mobility, transport enables rural populations to access markets, and to benefit from social and educational interactions, health care and technology transfer. Improving access to markets encourages rural farmers to modernize with fertilizers, mechanized equipment, and new seed varieties, which, in turn, raises yields, lowers unit costs, and increases demand for inputs. Rural transport also provides access to labor markets and thus the opportunity to earn non-farm income.
Roads play a central role in rural development, but despite large amounts spent on them, little is known about the size and distribution of their benefits. That is why I was determined to make this the topic of my dissertation. Before finding out about the MCC's compact with the Government of Nicaragua, I considered working with data from a few other similar projects. One in particular, from Vietnam, provided a lot of data related to a rural road rehabilitation funded by the World Bank, but it provided very little background information about the project. In contrast, the Open Data Challenge did not just provide me open access to survey data from Nicaragua, but a lot of background information too, in particular, an independent impact evaluation conducted by Dr. Jonathan Alevy, and the possibility of having someone to contact in case I had any specific questions.
The Millennium Challenge Corporation's Compact with Nicaragua
In 2005, the MCC signed a five-year compact (MCA-N) with the Government of Nicaragua to invest in the western part of the country in the departments of Chinandega and León. The MCA-N consisted of three projects:
Before a country can become eligible for an MCC assistance program, the MCC’s Board examines its commitment to just and democratic governance, investments in its people and economic freedom as measured by different policy indicators. After the municipal elections of November, 2008, in response to a pattern of actions by the Government of Nicaragua inconsistent with the MCC's eligibility criteria, the MCC’s Board terminated a portion of the MCA-N, reducing compact funding from $175 million to $113.5 million. Funding was terminated for all activities in the Property Regularization Project and for activities in the Transportation Project that were not already under contract. The Rural Business Development Project was not affected.
The MCA-N Transportation Project ended up substantially rehabilitating 42 miles of roads in 2008 and 2009: two secondary roads, Somotillo-Cinco Pinos in the north and León-Poneloya-Las Peñitas linking the urban center of León to oceanfront communities, and a secondary trunk road, Villanueva-Guasaule connecting the northern city of Villanueva to the Honduran border at El Guasaule. These road upgrades were expected to decrease vehicle operating costs and travel time. The rehabilitation included resurfacing and extensive drainage and grading improvements. The communities within the zone of influence of the road upgrades were expected to benefit in the form of lower prices and increased availability of consumer goods.
- A Transportation Project that allocated $57.9 million to the rehabilitation of 42 miles of rural roads.
- A Rural Business Development Project that helped farmers develop and implement a business plan built around a high-potential activity like growing beans, sesame, cassava, or vegetables, and producing milk.
- A Property Regularization Project that would eliminate the institutional and regulatory barriers preventing productive investment in property in León.
Before a country can become eligible for an MCC assistance program, the MCC’s Board examines its commitment to just and democratic governance, investments in its people and economic freedom as measured by different policy indicators. After the municipal elections of November, 2008, in response to a pattern of actions by the Government of Nicaragua inconsistent with the MCC's eligibility criteria, the MCC’s Board terminated a portion of the MCA-N, reducing compact funding from $175 million to $113.5 million. Funding was terminated for all activities in the Property Regularization Project and for activities in the Transportation Project that were not already under contract. The Rural Business Development Project was not affected.
The MCA-N Transportation Project ended up substantially rehabilitating 42 miles of roads in 2008 and 2009: two secondary roads, Somotillo-Cinco Pinos in the north and León-Poneloya-Las Peñitas linking the urban center of León to oceanfront communities, and a secondary trunk road, Villanueva-Guasaule connecting the northern city of Villanueva to the Honduran border at El Guasaule. These road upgrades were expected to decrease vehicle operating costs and travel time. The rehabilitation included resurfacing and extensive drainage and grading improvements. The communities within the zone of influence of the road upgrades were expected to benefit in the form of lower prices and increased availability of consumer goods.
Submitting a paper to the Open Data Challenge
The paper I submitted to the Open Data Challenge is titled: Access to modern markets and the impacts of rural road rehabilitation: Evidence from Nicaragua. The main goal of this paper is to analyze changes in the prices of consumer goods collected from establishments (stores) located along the rehabilitated roads. One of the most useful elements of this intervention is that although all but three of the road rehabilitation projects were cancelled, the collection of surveys along these cancelled projects was not cancelled and the data collected in these non-treated areas is also available. The independent impact evaluation conducted by Dr. Jonathan Alevy obtains the impact of the road rehabilitation by comparing store prices from areas where a rehabilitation took place to store prices from areas where it did not. His analysis provided evidence that the distribution of some perishable and fragile food items had improved as a result of the transportation project, but that the overall effects of the project had been modest. In particular, it was found that prices for cheese and eggs declined about 20% in the treated areas.
My paper follows a different strategy by providing a model of spatial arbitrage between the big city of León and the rural area around the rehabilitated road León-Poneloya-Las Peñitas in Western Nicaragua, one of the roads that were rehabilitated as part of the MCA-N Transportation Project. This particular road connects the city of León, which is the second largest in Nicaragua, with the fishing port of Poneloya and the coastal village of Las Peñitas and it is the only direct access to these coastal communities. Before the project its 12.2-mile surface was in very poor condition. Its rehabilitation included the improvement of pavement structures, minor and major drainage structures, sidewalks, shoulders, signage, and buses bays.
Alevy’s impact evaluation supports the idea that the price of perishable and fragile items in the area of influence of the rehabilitated roads had decreased as a result of the intervention. However, my paper has not found strong evidence that this has been the case in the area of influence of the road León-Poneloya-Las Peñitas. In fact, the results in my paper seem to support the opposite idea. Storable consumer goods like cooking oil, toilet paper, matches, and toothpaste seem to be the ones that were affected the most by the intervention.
My paper follows a different strategy by providing a model of spatial arbitrage between the big city of León and the rural area around the rehabilitated road León-Poneloya-Las Peñitas in Western Nicaragua, one of the roads that were rehabilitated as part of the MCA-N Transportation Project. This particular road connects the city of León, which is the second largest in Nicaragua, with the fishing port of Poneloya and the coastal village of Las Peñitas and it is the only direct access to these coastal communities. Before the project its 12.2-mile surface was in very poor condition. Its rehabilitation included the improvement of pavement structures, minor and major drainage structures, sidewalks, shoulders, signage, and buses bays.
Alevy’s impact evaluation supports the idea that the price of perishable and fragile items in the area of influence of the rehabilitated roads had decreased as a result of the intervention. However, my paper has not found strong evidence that this has been the case in the area of influence of the road León-Poneloya-Las Peñitas. In fact, the results in my paper seem to support the opposite idea. Storable consumer goods like cooking oil, toilet paper, matches, and toothpaste seem to be the ones that were affected the most by the intervention.
Extending my analysis with Open Data
The paper I submitted to the Open Data Challenge analyzed prices from stores located along the rehabilitated road León-Poneloya-Las Peñitas and compared them to prices in León. Originally I did not know the precise location of these stores. Now that I have access to the stores' GPS coordinates I am able to locate establishments more precisely along the road. The following map displays the location of these stores:
This map displays the location of the surveyed establishments along the rehabilitated rural road from León to Poneloya and Las Peñitas in the Pacific coast. It also displays the location of surveyed establishments along the rural road from León to Loma Pelada where the road rehabilitation was cancelled.
Thanks to comments received during my visit to the MCC, I have now introduced stores located along the road Santa Teresa-Las Brisas, which was NOT rehabilitated, as a relevant comparison group because:
In other words, changes in prices at the stores located along the rehabilitated road León-Poneloya-Las Peñitas can be attributed more certainly to the road improvement if this same effect cannot be found along the non-rehabilitated road Santa Teresa-Las Brisas.
One of the main concerns when picking a comparison group is whether treated and untreated areas shared similar characteristics before the intervention. In this case, a simple method to verify it is to test whether average prices for consumer goods were similar before the road rehabilitation work began. I decided to compare average prices for consumer goods in Carlos Canales (treated) to those in Chacaraseca and Loma Pelada (control) because these towns are located at a similar distance away from León. The average distance from stores in Carlos Canales to León is 5.85 kilometers and the average distance from stores in Chacaraseca and Loma Pelada to León is 5.10 kilometers (this difference in average distance to León is not statistically-different from zero). Only three out of twenty-two prices for consumer goods had a statistically-significant difference in prices between these these two areas: eggs, detergent, and bathroom soap (eggs and detergent are more expensive in Carlos Canales, and bathroom soap is cheaper in Carlos Canales). The difference in average prices between these two areas was not statistically-different from zero for the rest of the consumer goods, which suggests most prices were similar before the road rehabilitation project began.
Household surveys collected for the evaluation of the Rural Business Development Project are also available to evaluate whether Carlos Canales and Chacaraseca and Loma Pelada share similar characteristics. The following map displays the location of surveyed households located around the study area. There were several households surveyed in León and close to Carlos Canales, Chacaraseca, and Loma Pelada. Unfortunately, none of the surveyed households are located close to Poneloya and Las Peñitas.
- They are also connected to León by a single low-quality road.
- They are also isolated from other big cities, which suggests most of their trade will be directed to and from León.
In other words, changes in prices at the stores located along the rehabilitated road León-Poneloya-Las Peñitas can be attributed more certainly to the road improvement if this same effect cannot be found along the non-rehabilitated road Santa Teresa-Las Brisas.
One of the main concerns when picking a comparison group is whether treated and untreated areas shared similar characteristics before the intervention. In this case, a simple method to verify it is to test whether average prices for consumer goods were similar before the road rehabilitation work began. I decided to compare average prices for consumer goods in Carlos Canales (treated) to those in Chacaraseca and Loma Pelada (control) because these towns are located at a similar distance away from León. The average distance from stores in Carlos Canales to León is 5.85 kilometers and the average distance from stores in Chacaraseca and Loma Pelada to León is 5.10 kilometers (this difference in average distance to León is not statistically-different from zero). Only three out of twenty-two prices for consumer goods had a statistically-significant difference in prices between these these two areas: eggs, detergent, and bathroom soap (eggs and detergent are more expensive in Carlos Canales, and bathroom soap is cheaper in Carlos Canales). The difference in average prices between these two areas was not statistically-different from zero for the rest of the consumer goods, which suggests most prices were similar before the road rehabilitation project began.
Household surveys collected for the evaluation of the Rural Business Development Project are also available to evaluate whether Carlos Canales and Chacaraseca and Loma Pelada share similar characteristics. The following map displays the location of surveyed households located around the study area. There were several households surveyed in León and close to Carlos Canales, Chacaraseca, and Loma Pelada. Unfortunately, none of the surveyed households are located close to Poneloya and Las Peñitas.
The MCA-N Rural Business Development Project collected households surveys among farmers who participated in their training. Some of these households are located close to the rural roads that were part of the MCA-N Transportation Project and allow to learn a lot more about households in the area, even if these surveys were collected for a different purpose.
A significant difference in characteristics can be found between households in León and outside of León. Surveyed households outside of León are more likely to produce and consume their own produce. The comparison between rural areas is less clear. There is a very limited number of households located precisely at the villages along the roads. The one surveyed household located in Carlos Canales, for example, received training on how to raise livestock as part of the Rural Business Development Project and in the survey reported producing and consuming its own milk, cheese, and eggs. On the other hand, three surveyed households in Chacaraseca received training on how to grow sesame and also report producing and consuming their own milk, cheese, and eggs. No significant difference between these two rural areas can be identified by looking at these few observations. As we would expect, households located in rural areas outside of León are significantly more likely to produce and consume their own vegetables, grains, meat, dairy and fruits.