Tuesday, February 22, 2011
My favorite data visualization for this week is a work in progress brought to you by Ashley Ingram, a William and Mary student who is double majoring in Economics and Sociology. Ashley and I had been talking about whether U.S. foreign aid allocations shifted after the Cold War to target poorer countries rather than being driven so strongly by geo-political concerns. Naturally, I started talking about academic papers that showed this, that, or the other using cross national time series analysis. Ashley did something much cooler. She used AidData to map all the foreign aid given by the U.S. government for each year since 1985 into a series of cartograms where the dollar value of aid substitutes for the actual geographic area of land. Then she embedded these maps so you can see change in aid flows over time. If that wasn't cool enough, she also added data on average income by country so that light colored countries are poorest and dark colored countries are relatively rich. If U.S. aid flows shift dramatically to poorer countries after the Cold War, our maps should be getting lighter after 1990. And, they are...sort of...
A larger and higher-resolution version of this file is available for download.
Here are the 8 things that jump out at me when I look at this map. My colleagues who looked at this saw very different things and I'd be really interested to hear what other people think they see, and what it means.
1. The continent of Africa, which contains many of the world's poorest countries, does get larger over time, especially late in the time series.
2. Some of the increased aid to Africa is flowing to the richest countries in Africa (like South Africa right after the end of Apartheid in the early 90s and a second surge of dollars for HIV/AIDS programs from 2006-08).
3. The map does get lighter over time, but there are noticible blips of increasing darkness, like the large infusions of aid to relatively rich countries of Eastern Europe and the former Soviet Union in the 1990s.
4. Some really obvious things are constant and do not change -- Egypt and Israel remain massive splotches on the map every year as a result of the Camp David settlement.
5. Post invasion Iraq and Afghanistan receive massive and sudden increases in aid that literally distort the map. I guess Paul Wolfowitz was wrong about those Iraqi oil revenues paying for reconstruction.
6. Aid to Pakistan is very volatile. Lots of aid until 1990. Very little until 2001. Lots of aid after 2001. This pattern appears to be driven by two things: pursuit of nuclear weapons, which the U.S. deems unacceptable in 1990; and the U.S. invasion of Afghanistan and the war on terror, which apparently makes nuclear weapons in Pakistan no longer unacceptable.
7. U.S. aid to Latin America declines in most countries, especially the middle income countries, but Peru and Colombia defy gravity.
8. Relative to other regions, East Asia seems remarkably stable and gets very little aid over time. The most interesting changes occur when the U.S. attempts to buy North Korean compliance with various nuclear deals (see the polka-dotted country appear suddenly in 1999 and 2002).
Very rarely do such data visualizations answer important questions definitively, and this one doesn't either. But I have found repeatedly that seeing data in a different way does raise good questions and causes people to think creatively. The exact same data in a flat table, a line graph, a regression table, and a cartogram look very different and suggest different things. I did not expect some of the things listed above. Some of them raise interesting questions and/or suggest hypotheses to be explored. Some of them are downright misleading. A few observations to get our conversation started...
A. While Pakistan, Afghanistan, and Peru contribute to the "lightening" of the map, presumably the aid is not being targeted for these countries simply because they are in the bottom two quartiles in terms of poverty. Correlation (even in multi-colored maps) does not equal causation.
B. Poverty is not uniformly distributed within countries, so even if "less poor" countries are receiving a lot of aid, it does not mean the aid is not going to poor people (though it probably is not). In order to test this hypothesis (or visualize it) we would need sub-national data like this.
C. What the heck is that "rich" dark blue blob on the map under Jordan? I know that Yemen gets a lot of foreign aid, but I keep hearing on CNN that Yemen is very poor. It is not Kuwait, Saudi Arabia, or Oman (I checked and the U.S. gives little or no aid to those countries for most years).
Bottom line: The thing that you REALLY need to make useful visualizations is better data. Right now the gee-whiz tech tools are better than the underlying data (at least the data that is publicly available). This is especially true at the sub-national level.
I apologize if I sound like a broken record, but I am really pleased about the adoption of the IATI standard and I am looking forward to the day when donor governments, recipient governments, and aid implementing organizations make more of their data publicly accessible.
This post was originally published on Full Disclosure, which is the DevEx aid transparency blog. For a link to the original post click here.
Monday, February 7, 2011
Ransom and Bain used AidData to identify the universe of agricultural projects allocated to developing countries between 1978 and 2003. Then they independently coded a sample of these projects in order to identify the type of agricultural aid and whether the projects directly or indirectly targeted women. The Figure below shows the percentage of agricultural projects that fell into one of six broad categories in a given year (white bar) and the percentage of projects within that sub-sector that targeted women/gender (shaded bar). Across all six types of agriculture aid the pattern looks similar -- no targeting at the beginning of the time series, a surge in the 1990s, and then drop off in 2002-2003. The figure below shows one of the six categories (Ag Infrastructure).
It turns out that while the discourse about mainstreaming gender and targeting vulnerable female populations has continued to increase over time, dollars for this purpose have not. Ransom and Bain conclude on a pessimistic note: "Our findings suggest that the rhetoric of gender mainstreaming outstrips efforts to develop projects aimed at women and gender inequality and that the concept may be being used to legitimize a decline in focusing explicitly on women." For the whole paper click here.
Their findings are similar to Powers et al who compare actual budgets of World Bank projects to the rhetoric of environmental mainstreaming within the Bank and show that rhetoric often outstrips funding reality in that sector as well.
Tuesday, February 1, 2011
The International Aid Transparency Initiative (IATI), launched in 2008 during the Accra High Level Forum on Aid Effectiveness, is approaching a major milestone. Next Wednesday, its Steering Committee will meet to finalize the second phase of the IATI data standard (the first phase was agreed last summer). Meanwhile, just this week, the UK’s aid agency—the Department for International Development (DFID)—became the first donor to publish its data on the IATI data registry in the agreed format. Many more donors will publish their information in this format in the lead up to the Busan High Level Forum to be held in Nov./Dec. 2011.
IATI has been signed by 18 donors and endorsed by 18 partner countries, including Liberia and Bangladesh most recently. In advance of next week’s meeting, 35 CSOs advocating greater transparency and comparability of aid information have sent letters to IATI signatories encouraging them to make sure IATI delivers. In particular, they emphasize the need to address the priority concerns of countries receiving aid, including the timely provision of data on current and projected aid flows, and the use of common data formats that are compatible with national budgets and systems.
As donors, partner countries, and other development organizations make progress toward speaking the same language—using the same definitions and classifications for different types of aid flows—it makes AidData’s job much easier. Soon, AidData users will be able to download data in IATI format, so stay tuned. Being able to aggregate, disaggregate, and mash up aid information from many sources is key to answering the most basic and the most sophisticated questions about where aid goes and what impact is has.