A spatial perspective on World Bank health projects in India
While development indicators improved in some regions over time, some regions were continuously lagging behind.
As part of a team of summer interns at Development Gateway, I geo-mapped multilateral development projects to help assess aid effectiveness and donor coordination. A similar effort by a team of interns last year led to the successful launch of the World Bank Mapping for Results (MfR) platform, making the World Bank sub-national project information accessible to the public at large. Since then, the World Bank’s Open Data Initiative has launched a new portal for financial information and supported the Kenya Open Data Initiative for sub-national recipient country budget data. In a short time, mapping of development projects has emerged as a new way to increase transparency and accountability in the international development world, and has been included as a part of the IATI (International Aid Transparency Initiative) standard. Several development organizations have followed suit and agreed to make geographic project information public. I decided to use my internship opportunity to analyze the World Bank’s health and social service projects in India to see if I could discern any patterns. The results were interesting.
India is a significant recipient of IDA (International Development Association) loans and the data for active project locations were readily available at the World Bank MfR website. I obtained sub-national development indicators from Indian district level health surveys (DLHS), and using a basic geo-mapping program (www.geocommons.com), available for free online, overlaid them on the World Bank active health project locations. The percentage of fully immunized children at the district level was the first development indicator I used. The dots represent the location of the World Bank projects, with the size of each dot signifying the amount of aid. For Figure 1, the lighter regions on the map, belonging to lower quartiles, indicate low rates of child immunization.
Figure 1: Percentage of Fully Immunized Children (District Level)
As soon as I mapped the data a clear pattern emerged. The districts in the northern regions were lagging behind in terms of immunization. When I mapped the next indicator, percentage of district level population living at a low standard of living, a similar pattern was visible: many regions with low immunization rates had a high prevalence of populations with a low standard of living (Figure 2). (I reviewed the same indicators from a similar survey conducted 5 years prior. Though development indicators had improved in some regions over time, I could clearly identify regions that were continuously lagging behind.)
The next step for this analysis was to overlay the locations of World Bank project activities on these indicator layers. For Figure 2, darker regions or regions in higher quartiles, indicate a high percentage of the district population qualified as having a low standard of living.
Figure 2: Percentage of Population with a Low Standard of Living
In the map of district immunization rates (Figure 1), the active World Bank projects cover the Northern districts in the second lowest quartile, but districts in the bottom quartile (lowest immunization rate and lightest shade) in adjoining regions have no active World Bank projects. On the other hand, there are projects in the Southern states, like Tamil Nadu, which fall in the third and fourth quartile (high rates of immunization). The map with the standard of living indicator (Figure 2) presents a slightly different picture. The World Bank active projects cover the majority of Northern districts in the quartile with the highest percentage of the population belonging to the low standard of living category (the darkest region).
This initial analysis therefore suggests several interesting questions for researchers How effective is the targeting of these projects? If the projects were mapped alongside a different set of indicators, rather than these two, what sort of picture would emerge? What if we looked at the trends in key indicators over time, relative to the start/end date of these projects, rather than a simple snapshot? Geo-enabling aid activity information makes this kind of spatial analysis much more feasible.