The indicator calculates the share of China's outward foreign direct investment stock (OFDIs) on a country's total stock of inward FDI (Total FDIs).
Chinese FDI statistics present the following known problems:
a) Data recording method: China gathers data through local commerce bureaus where companies register their foreign investments instead of mandatory enterprise surveys. This can result in substantial underreporting by firms that wish to sidestep approval procedures.
b) Investment amount: stringent capital controls have favoured overstating or understating of investment amount to disguise hot money flows.
c) Round-tripping: sending capitals to Hong Kong or other tax havens only to repatriate them to enjoy preferential tax treatments. This is, however, not an issue in our case because tax havens are not the subject of our study.
d) Destination country: Chinese data report the first and not the last destination country of investment flows. This is the least problematic issue in our opinion because most of the hot money is likely going towards developed countries after leaving Hong Kong or other tax havens, and to a lesser extent towards the Middle East and North Africa.
The indicator reports the value of overseas projects contracted to Chinese companies. Data are aggregate of two time series 'Engineering construction projects' and 'Labor service cooperation'.
China reports data for overseas projects in three time series: 1) engineering construction projects, 2) labor service cooperation, and 3) design consulting. China classifies these activities as ‘Foreign Economic Cooperation’ activities, which usually include civil engineering constructions financed by foreign investors, projects financed by the Chinese governments through its foreign aid programs and construction projects of Chinese diplomatic missions, trade offices and other institutions abroad (Brautigam 2011, pp. 216-217). Design-consulting data are not accounted for because of their discontinuity.
China Trade And External Economic Statistical Yearbook (2016 and previous years).
Whenever the data was not available it has been considered as 0.