It’s Not Just Big Data: Data Granularity and Aid Targeting

Reposted from Duck of Minerva.

Prevalence of HIV across East Africa with purple hot spots of high prevalence

Prevalence of HIV across East Africa with purple hot spots of high prevalence

Earlier this spring, I had a chance to talk to Mark Dybul, the head of the Global Fund to Fight AIDS, TB, and Malaria and former administrator of PEPFAR, the U.S. bilateral AIDS program. At the time, he expressed optimism about using geo-referenced data on HIV/AIDS prevalence to better to target AIDS foreign assistance. In advance of the recent AIDS conference in Australia, researchers (which include Dybul) released a new study in The Lancet ($) that modeled that potential in Kenya by focusing on the hot spots of high HIV/AIDS prevalence (see above East Africa map, purple represent high prevalence levels). Dybul’s comments were music to my ears. For the past year, I’ve been part of the AidData Research Consortium’s project (ARC) to develop sub-national foreign assistance data. Already that project has worked to help geo-reference World Bank, African Development Bank and Asian Development Bank projects as well as foreign assistance from all donors in a number of countries. As many of you know, I’ve been part of climate vulnerability mapping for the better part of five years through my work on Africa through the Minerva Initiative and the CCAPS program at the Strauss Center. This fall we will embark on a new Minerva project to look at disaster vulnerability and complex emergencies in South and Southeast Asia. In this post, let me say a few more words on the importance of data granularity and aid targeting. The logic behind data granularity is fairly simple. If you know where the problems are concentrated geographically, you can target aid accordingly, which may be particularly important in an era of tight aid budgets. The Economist ran a story on the importance of data granularity and these findings:

One watchword here is granularity. The world’s AIDS maps, which once recorded rates only on a country-by-country basis, now do so region by region. This means effort can be focused on the worst-affected places within a country, not just on the worst-affected countries. Deborah Birx, America’s global AIDS co-ordinator, thinks such focus is essential. It will, she believes, keep downward pressure on the infection rate without too much extra expense. A study published in the Lancet on July 19th, to coincide with the conference, supports her in this. Sarah-Jane Anderson of Imperial College London and her colleagues have crunched the numbers for Kenya and concluded that focusing on the worst-affected parts of that country could, over 15 years, reduce the number of new infections by 100,000 at no extra cost.

You see HIV/AIDS is not distributed evenly over the country. The Lancet author conclude that future infections are likely to come from a handful of areas:

Across the 47 counties in Kenya, estimated HIV prevalence varies substantially, from less than 1% to 22% in the year 2013. In the model, the estimated number of new infections is concentrated in five counties (Migori, Homa Bay, Kisii, Nairobi, and Kericho) that account for almost 40% of all new HIV infections in the country.

While political criteria may ultimately prove more important in the allocation of aid and spending both within and between countries, granular georeferenced aid on baseline problems like HIV and where aid money actually gets spent at least provide some intellectual ballast as to where resources ought to be directed. This is a little easier if aid allocation is based on a single indicator such as HIV/AIDS prevalence. As we have discovered in our work on Africa, this is harder if you are trying to develop a multi-dimensional index of underlying need such as vulnerability, where modeling choices can greatly influence outcomes and different modeling groups may generate very different portraits of underlying vulnerability. In the HIV/AIDS context, the Global Fund is championing this approach and, as the main multilateral donor, it can largely authoritatively decide which maps of need are appropriate. Sarah-Jane Anderson and her co-authors in The Lancet piece write of the approach and significance:

A uniform strategy, which does not use available intelligence on the epidemic, will fail to be as effective as a strategy that does. By use of a public health approach that focuses resources based on an epidemiological understanding of subnational geographical areas and key affected populations, and selects the package of interventions most likely to have an effect according to the drivers of each HIV stronghold, the efficiency and effectiveness of programming could be greatly increased.
Right now, it’s easier to assess where the need is than it is to assess where the money is going. The map above was developed using data on HIV prevalence levels from the USAID-funded Demographic and Health Surveys (DHS) which are fabulous and rich source of sub-national data for a number of developing countries. While donors like the Global Fund may geo-code projects internally, it’s been tough sledding for those of us interested in foreign assistance to acquire this data and to push for more data transparency in this space. Trends are moving in that direction so perhaps in a couple of years, we’ll be able to judge if the new Global Fund strategy of targeting aid is working or not.

Josh Busby is an Associate Professor at the LBJ School of Public Affairs and a member of the AidData Research Consortium (ARC).

Tags: dataaidGlobal FundHIVAIDSARChealthgeospatial