Mapping For Results

The following is a post by one of our research assistants at the College of William and Mary, Alena Stern ('12)

In about five minutes, any person armed with a computer and an internet connection can determine the exact amount of development assistance directed to the education sector in Uganda over the past 8 years. A quick query of the AidData website reveals that, since 2002, $643,103,401 (Constant 2000 USD) has been committed to the education sector in Uganda. Having this information at one’s fingertips is a huge step forward in aid transparency, but many people in the development field view this as merely a first step and see a need for more specific and accessible information on where the funding actually goes. Although AidData makes accessing a list of projects in the Ugandan education sector much easier, if I wanted to determine where within Uganda these projects are located, I would face a much more difficult task.

First, I might find myself looking for information that doesn’t exist. Many donors refuse to make their project documents public, making it nearly impossible to determine just where a project is implemented.

However, the problem is not always a dearth of information. With certain donors, the sheer volume of information actually makes accessibility a challenge. Let’s take the World Bank for example. Thanks to the Open Data Initiative announced in April 2010, all World Bank documentation is now publicly available. However, to find the sub-national location of a project, I would have to sort through hundreds of pages of PADs (Project Appraisal Documents), EAs (Environmental Assessments), ISDSs (Integrated Safeguard Data Sheets), PIDs (Project Information Documents), PPs (Project Papers) and other official documents. And then, in many cases, I still might be frustrated to discover that there is no concrete geographical information. I know this challenge well, as I spent six weeks of my summer working as a researcher on the Mapping for Results Initiative.

This new geo-coding initiative represents a partnership between AidData and the World Bank Institute. Over 6 weeks, our team of 13 interns geo-referenced all 1,216 active World Bank projects across 42 countries in Sub-Saharan Africa, 27 countries in Latin America and the Caribbean, Indonesia, and the Philippines along with a selection of African Development Bank projects. In total, we coded about 12,000 specific geographic locations, but hopefully this is just the beginning.

We geo-reference by recording each individual location targeted by a development aid project as mentioned in the documents referenced above, down to the most precise geo-graphic level possible--sometimes even down to the neighborhood level. In June of 2010 there was no requirement within the World Bank to systematically report this information in a standardized way. In response to the difficulties this presented to our research team, the World Bank is now experimenting with a pilot program to introduce standardized location reporting into future project documentation. After finding all the location names, we then “reference” each location by finding its latitude and longitude, so it can be universally referenced no matter how many times administrative divisions may change (which can be many, many times). For a longer and more entertaining discussion of geo-coding and this project, see this video, which has lots of cool maps, created by Aileen Boniface (Virginia Tech), Patricia Austria (College of William and Mary), and Kelsey Ranta (Georgetown University).

Why is geo-referencing important? First, geo-referencing allows for better donor coordination. Simply knowing what projects are underway in a certain country is not enough to avoid project duplication within a given region. If only national-level information is known, it is possible that the bulk of donor activity will be clustered in one region of the country while other regions are neglected. Empirically, this does occur, as this map of World Bank projects in Kenya overlaid on a map of poverty levels by district illustrates:

Second, geo-referencing is important because nations are not homogenous. As the variation in poverty levels within Kenya illustrates, levels of need are not constant across a country. This certainly holds true with sector-specific indicators of need, such as infant mortality in the health sector or primary school enrollment in the education sector. We could have a much richer picture of whether aid flows where it is most needed by looking beyond whether aid is flowing to the neediest countries to whether hospitals are being built in the districts with the worst health indicators, power plants are being constructed in the districts with the lowest levels of electrification, and wells are being built in the districts with least access to clean water. If aid were consistently targeted to the areas where it is most needed, aid dollars could more effectively deliver development results.

Finally, geo-referencing is critical in the effort to improve accountability and dialogue with recipients. Reinikka and Svensson (2005) illustrate the power of localized information in increasing the percentage of government expenditures on education that actually reach intended beneficiaries. After a public expenditure tracking survey (PETS) stated that in the mid 1990s, the average Ugandan school only received 20% of central government spending intended for the school, the government instituted a newspaper campaign to inform Ugandan citizens of what their schools were entitled to in central government expenditure. A second PETS in 2002 stated that in 2001 the average school received 80% of central government expenditure. Though the causality between the newspaper campaign and the reduction of local capture may not be confirmed, this example certainly illustrates the power of localized information to improve accountability and create results. Without the reduction in local capture, nearly $106 million of the $177 million of aid to education since 2002 would have been lost.

Geo-referencing creates the localized information that empowers recipients to hold their governments accountable and makes it possible to easily visualize aid information so it is accessible and understandable. Greater transparency doesn’t help recipients if the information they need is tied up in hundreds of pages of text. Information accessibility is even more important than availability. As illustrated by the map above, geo-referencing can produce visualizations that make the inequities of aid distribution within a recipient country immediately apparent.

Through this initiative, the World Bank has been a leader in making its data not only available, but also accessible. We hope that the Mapping for Results Initiative will encourage other donors to undertake the project of geo-referencing their data.
Tags: behind the curtaindata visualizationssouth-south cooperationgeo-referencingaid coordination


Thanks for the video and the static image of Kenya. Sorry if I've missed an obvious link to this, but where and when are you going to make publicly available this data (eg in a google earth or maps mashup?)

Hi, this is Mike Findley and I work on the AidData project. Thanks for your comment on the availability of the georeferenced data. <br /><br />The data should be publicly available in late September or early October. All parties are working vigorously to get the data and maps ready. Some interactive maps will be available through the World Bank at that time and the underlying data should be accessible for further integration as users see fit. <br /><br />AidData will also post some maps (at that will include additional data that we have coded, beyond what was accomplished with the Mapping for Results Initiative (an additional 50,000 projects were coded from AidData). <br /><br />Mike Findley

Thanks - Mike for answering and mister z for asking.<br /><br />A follow-up question:<br /><br />Will the data be available under some standard Open license (preferrably CC-by-sa or less restrictive - but at least CC license or other standard Open license) or will it be licensed under the same non-standard license that licenses its data? <br /><br />I think that this is of major importance because if it's the latter then the data can't be used in all of the hundreds or rather thousands of projects & services that use standard Open content licenses such as Wikipedia or OpenStreetMap.<br /><br />Publicly available (even with snappy + free -- as in no cost -- APIs) and Open are two very different things and I hope that people at AidData know this difference & understand it well -- even though your site (including the blog) seems to be licensed under "traditional" All Rights Reserved copyright license.<br /><br />Now, I know that publishing _existing_ data, especially a huge amount of it, under an Open content license is not something that can be done easily. <br /><br />So, I'm not starting a discussion about the World Bank's Open Data Initiative currently lingering on the border of public data initiate rather than Open (and free as in free speech) data; let's hope that the initiative has set a direction and that will sooner than later become a place for Open data (and not just free and publicly+easily available, which in itself is great). <br /><br />Rather, I'm just saying that these "historic restrictions" shouldn't come in the way of licensing new data with an Open license (in projects that truly want to embrace openness and freedom such as AidData AFAIK) as new data doesn't have to carry those (institutional, existing/historic agreements, etc) restrictions.<br /><br />And so, I'm interested to hear about the license of the the data that will be published. Just as interesting are the (possible) discussions that have taken place around the topic.<br /><br />These license choices have significance also because there aren't _that_ many initiatives yet that embrace Open paradigms as in open source approaches (and not "just" transparency, accountability, good governance, etc). Or at least most of the bulk of things are still done with pretty much completely "closed" paradigms with just enough transparency to keep the troublemakers quiet.<br /><br />So, when AidData or any other similar venture that seems to want to embrace the Open are actually doing something neat (as you are) then this sets a precedent.

OK, after that comment -- and after watching the video clip -- it's clearly appropriate to comment on the big picture:<br /><br />This is a fantastic pilot! I dearly hope that it will make a permanent mark to the way development projects are prepared, executed and documented across the whole international/global/etc development field.<br /><br />I think it's especially great -- and not so surprising -- that the team who pulled together something like this was a young enthusiastic group of people and not "seasoned professionals".<br /><br />In a way it shows what we've seen from so many other break-through projects that it's not (traditional) expertise, experience nor money -- not even technology in itself that changes things, but rather, _how_ things are done.<br /><br />Just as Clay Shirky has well said: "Revolution doesn't happen when society adapts new tools. It happens when society adapts new behaviors." ( <br /><br />There's nothing (cutting-edge) new about the tools of geo-coding. But it's new behavior to map all projects' geo-locations and make them readily available in one place, for free; hopefully also Openly and Freely. <br /><br />And _that_, I think, would be the biggest change in behavior. <br /><br />Making things truly Open and Free, enabling Open participation for anyone and everyone (with the least possible barriers -- and in this it's rather people with some appropriate wiki/OpenGeo project who should be consulted and not the "seasoned professionals", no hard feelings) and doing so ensuring maximum access to information, including the voice of those who are being helped, empowering people to the fullest, maximizing transparency and through that also accountability and so on (@5:58, 3:00, 5:43, and 3:15 in the video, respectively ;).<br /><br />Truly Open and Free parts are crucial in ensuring the highest probability for all of the goals of the project mentioned in the video to become reality. Projects such as Ushahidi, Sahana, OpenStreetMap and related projects + other similar projects have shown very successfully how incredibly powerful Open solutions can be when _used_ in an Open manner.<br /><br />So, I hope the pilot helps get the idea "mainstreamed" to the Bank's activities. And hopefully you find a way to Open & Free the data + make the platform in which the data is presented Openly collaborative and participatory.