On September 7th The Economist published an article, "China's Relationship with Africa is Growing Murkier," with a graph on "murky" Chinese lending into Africa that really muddied the water (Figure A).
 |
| Figure A |
The Economist implied that Boston University and the World Bank were underestimating China's sovereign loan commitments in Africa because they were not shining their light in all the dark corners: specifically, they were not including all Chinese creditors and all kinds of loans. But instead, it appears to be The Economist that needed a better flashlight.
As Chinese-loan-data-geeks, my Cornell University colleague Yufan Huang and I are very familiar with the World Bank's IDS, AidData, and Boston University. We spent a good part of Monday trying to replicate the numbers in The Economist's chart.
Here's what we found.
- We guessed from the shape of the World Bank data curve that The Economist was only selecting "official" Chinese creditors, and not "private". Yet many African governments (like Angola's) report China Development Bank loans to the World Bank as "private". [That data is in the IDS, but analysts need to select "official" and "private".]
- Working on that hunch, we were able to replicate the curve of the World Bank's data on sovereign loan commitments using "official", but not the amounts (Figure B).
 |
| Figure B (CARI chart) |
- What The Economists' graph showed as $20 bn, should actually be labeled $10 bn. The data labels on the right axis were off by 100% for the World Bank data but, strangely, this was not the case for the other databases.
- When we added "private" Chinese loan commitments to the World Bank data (Figure C), we were able to bring the curve much closer to the Boston University curve. [Although the article implied that Boston University only tracked Chinese policy banks, their Chinese Loans to Africa database includes all Chinese creditors, official and "private", that make loan commitments to African governments and their state-owned companies.]
 |
| Figure C (CARI chart) |
- BU, World Bank, and AidData have a few methodological differences, discussed below.* Most importantly, AidData includes Chinese loans to private Chinese corporations investing in Africa. Interesting, but not the same as sovereign, i.e. public and publicly guaranteed debt. This adds some $26 billion to AidData's figures that do not show up for the World Bank or BU. They also include $4 bn in central bank swap lines.
 |
| Figure D (The Economist) |
With our concerns in hand, The Economist agreed to revise their chart (Figure D). The new chart corrects the AidData numbers (note the color of the lines are changed), although it still overcounts currency swaps tracked by AidData by including all the rollovers as new credits, and they still include Chinese loans to Chinese corporations. However, they do not correct the World Bank data, and thus imply, wrongly, that the World Bank is significantly undercounting Chinese sovereign loan commitments in Africa.
Here's the sovereign lending chart they should have included, in our view (Figure E). This removes private borrowers and swaps from AidData, and Chinese loans to regional banks in Africa from AidData and BU.
Figure E suggests that Chinese loans to African sovereigns are not so murky -- the World Bank**, Boston University, and AidData all seem to agree on this.
 |
| Figure E (CARI) |
This may seem a data geek tempest in a teapot: the equivalent of quarreling about how many angels can dance on the head of a pin. But what worries me most is that the skilled folks at The Economist didn't wonder why their data was so off the charts.
It suggests that huge overestimations of what China is doing in Africa -- and perhaps elsewhere -- are now seen as plausible.
This feeds a larger, fear-based narrative about Chinese lending overseas. And that complicates the goal we share, of trying to understand the real story of how China engages in Africa.
A. Syndicated Loans. Boston University follows the World Bank's International Debt Statistics reporting protocol. The World Bank asks countries to report syndicated loans, record the lead bank, and report "whether the syndicate comprises institutions of only one or of several countries." If Chinese banks are part of an international syndicate with other multinational banks, this does not show up as a Chinese loan in either source. AidData does include these loans. Their methodology "assumes that each bank provided equal contributions to the syndicated loan" and they include the Chinese portion in their data.
*Methodology Differences
B. Swap Lines. Boston University does not includes People's Bank of China central bank swap lines in their development finance data, although they have a separate database on swaps. The World Bank only began requiring countries to report swap lines of greater than one year duration in 2020, so their data wouldn't include most of these. AidData includes swaps, but in their most recent database, they record only about $4 bn in PBOC swap lines in Africa, covering only Nigeria and Egypt (in a separate swap database they also records PBOC swaps in Morocco and South Africa).
C. Lines of Credit. The World Bank records Chinese lines of credit as a commitment in the year it happens. Boston University and AidData record instead the individual projects under those lines of credit. This explains why in Figures B & C, the World Bank data for 2011 spikes up and falls back, compared with BU and AidData.
D. Loans to Private Borrowers. Neither the World Bank nor Boston University report data on Chinese loans to private Chinese or other private sector borrowers in Africa, unless that loan has a public guarantee. AidData does include these private sector loans, coding them as either "potential public sector liability" (total of $6.8 bn) or purely private (total of $26 bn). This explains the difference between AidData and Boston University/World Bank in our Figures B & C.
**Sam Pa, China International Fund, and Angola
We wondered why the World Bank had a big bump of close to $20 bn ($10 bn once the data label is corrected) in Chinese loan commitments for 2005, when neither CARI/Boston University nor AidData had this figure. Digging into the World Bank data revealed that they had included $9.8 billion in "Chinese official loan commitments" for Angola -- the exact amount that was allegedly pledged by the notorious Hong Kong company China International Fund (CIF). This private company, which later moved its headquarters to Singapore, failed in its boast that it would arrange that financing. (Their leader, Sam Pa, was arrested in 2015 under Xi Jinping's anti-corruption campaign and hasn't been heard from since). Despite its deliberately confusing name, CIF was a private Hong Kong company, not a Chinese official creditor, so it is included in error.
