Many of the common perceptions about Chinese investment in Africa are not well supported by evidence (Brautigam, 2009). For this reason, we devote this section to describing our data sources, acknowledging weaknesses where they exist and explaining why different data sources may lead to somewhat different conclusions. Where appropriate, we draw on the work of others to support our arguments.
Getting a handle on the extent and type of Chinese investment in Africa is challenging. One reason is that planned investment projects – the data captured for example in MOFCOM OFDI project - typically differ significantly from realized investment projects – the data captured by MOFCOM OFDI flow. This problem is not unique to Chinese investment. However, using data on planned and approved projects can wildly overstate actual investment. For example, Shen (2015) compares data on the number of Chinese FDI projects from six African countries to the corresponding data from MOFCOM and finds that the number of projects according to host country data was at least 2.5 times greater than the amount recorded in the MOFCOM OFDI project data. Statistics produced by the Ethiopian Investment Commission (2016) reveal an even greater disparity between planned projects and projects in the operational phase. For example, in 2013 only 5% of all projects had moved from the planned stage to the implementation or operational stage. More recent fieldwork by Brautigam et al (2017) reveals a very low correspondence between project approvals recorded in the MOFCOM OFDI project data and actual investments on the ground.
Because the Chinese government requires all state-owned and private firms investing abroad to submit detailed information on a regular basis, official Chinese data (MOFCOM OFDI flow) tend to be relatively reliable compared to other sources when it comes to tracking actual investments by Chinese companies in Africa. However, there are some important caveats. First, the MOFCOM OFDI flow data is likely to miss smaller projects since in the past it has only tracked projects valued above $10 million. The aggregate value of these ‘missing’ projects is almost impossible to estimate without digging in country by country. While projects under $10 million are likely to account for a small share of total investment, the size of unrealized projects is unknown since the MOFCOM OFDI project data does not record project values (Brautigam et al 2017). In addition, the MOFCOM OFDI flow data misses investments that go to Africa through offshore tax havens.
An alternative source of information on Chinese investment in Africa is project level data also collected by MOFCOM[1]. The project level data is a database of all registered Chinese firms with investments in Africa. This database includes projects above $100 million that have been approved at the central level and projects between $10 and $100 million that have been approved through provincial offices. The database includes the name of the parent company and its African subsidiary, the scope of its business, and the date its application was approved by MOFCOM. This level of detail makes it possible – at least in theory – to code investment by ISIC sector. However, most projects span several sectors, making coding difficult. An additional drawback of these data is that the dollar amounts of the projects are not revealed. Finally, like the data on planned investments collected by African governments, approved investments often do not come to fruition, meaning that one must be careful about drawing conclusions about Chinese investment in Africa based on these data alone (Brautigam et al 2017).
Thus, to describe broad trends in Chinese investment in Africa we draw primarily on the MOFCOM OFDI flow data. These same data are also deemed the most reliable source of Chinese OFDI by UNCTAD and are used in UNCTAD’s World Investment Report. The official source on overseas investment and cooperation in China is the China Commerce Yearbook, which is published by the Ministry of Commerce. This ministry is charged with collecting and reporting all such data and is our primary data source. In addition, the China Statistical Yearbook includes investment data consistent with the China Commerce Yearbook. All of these data are publicly available online and we provide links to the websites in our references. For the most recent data (2015), which has not been published in the yearbooks, the official sources are the China Annual Bulletin of Statistics of Contracted Projects and Labor Cooperation with Foreign Countries, which can also be accessed online on the website of the National Bureau of Statistics of China.
We use UNCTAD data on bilateral foreign direct investment to compile investment into Africa from all other countries. As noted, the UNCTAD data for China comes from official sources and matches exactly the Chinese government data on outflows to Africa. It is therefore a useful source for comparing inflows from China into Africa to inflows from other countries into Africa.
Of course, an alternative approach would be to collect data on FDI inflows from African countries. This is done by the IMF through the Coordinated Direct Investment Survey (CDIS), in which individual countries are asked to submit data on their inward and outward direct investment stocks. However, there are large disparities between the CDIS and the MOFCOM OFDI flow/UNCTAD data, with the former indicating much higher levels of Chinese FDI as compared to the latter. Based on work by Shen (2015) and Brautigam et al (2017), our view is that a large part of the discrepancy is likely to be the difference between planned and actual investments. This is because most of the African data comes from understaffed investment agencies which are often not able to keep up with what is happening on the ground.
By comparison, international trade data is a lot easier to find and tends to be more reliable. Part of the reason for this is that there is no such thing as ‘planned trade’. A second reason has to do with the fact that exports and imports mostly pass through official customs agencies in at least two countries making it possible to verify transactions without extensive fieldwork. There are of course issues with smuggling and invoicing but having two sets of accounts can alleviate some of these issues. Our source for all of the trade data we use in this paper is UN COMTRADE.
[1] These data were available from NBS until 2016 but they have since been removed and so do not appear to be publicly available.
