If you needed reliable and comprehensive data on development finance flowing into your country, where would you go? We often assume that the most detailed and accurate records are maintained by the organizations directly involved. The OECD-DAC donors regularly report their data through the Creditor Reporting System (CRS), and recipient governments maintain their own databases of external financing, some of which are open to the public. So you would go to one of those official sources, right?
A new effort at AidData to track development finance from China to Africa with media-based data collection methods has challenged this basic assumption. Much to our surprise, we found during the course of our data collection work that media-based methods often reveal major weaknesses in official reporting systems. In Malawi, for instance, the Ministry of Finance maintains a database of all incoming aid flows. The publicly released dataset—covering 80% of Malawi’s total known aid receipts from 2000-2011—lists 538 commitments (totaling $5.3 billion) from 30 donor agencies, including two Chinese-funded projects worth $133.78 million. AidData’s media-based data collection methods captured these two projects, but also uncovered an additional 14 Chinese projects, not captured in the Ministry of Finance's database, worth an additional $164.8 million over the same time period. Of these 14 projects, AidData was able to uncover actual project values for 9 of them. These projects are provided in the table below. We came to a very similar conclusion last year while piloting our media-based methodology on Saudi Arabia to capture its official development finance to Yemen.
Going forward, we will integrate into the database additional projects not found in the media reports, and vet and refine the data through crowdsourcing and expert feedback. We also hope to apply this methodology to a DAC donor (that releases official records) to detect possible biases in media-based data collection methods. After all, you can't reduce or account for biases until those biases are known. We thank Bill Savedoff of CGD for making this point at the recent launch of our methodology, database, working paper, and online platform.
We encourage others to also explore the project-level data at china.aiddata.org and 'kick the tires' of our methodology (version 1.0). If you think you have uncovered a weakness or a systematic bias in the methodology, send us a note at firstname.lastname@example.org. We also welcome guest bloggers who have conducted analysis with the data and would like to share their work.