China's major economic data continue to improve

Problems with China’s economic data are well known,[1] but what is often overlooked is that China’s National Bureau of Statistics (NBS) continues to improve the quality of some of the most important macroeconomic data, in most cases by adopting international standards. This process began as long ago as 1985, when China began to measure output according to the System of National Accounts (SNA), the international standard for measuring economic output, investment, and other economic variables, and abandoned using net material product, a measurement concept it had borrowed from the Soviet Union in the 1950s. This was a major step forward since net material product overlooked the contribution of large parts of the service sector. More recently the NBS has revised its data on fixed asset investment, international trade in goods, household income and expenditure, and value-added in financial services. In each case the revised data are an improvement, frequently by bringing China’s data more closely into alignment with widely used international standards.

Fixed Asset Investment

China’s NBS adopted the Soviet measure of fixed asset investment (FAI) in the 1950s and after a break during the Cultural Revolution resumed publishing FAI data in 1981. Since 1998, the Bureau has compiled and released this data on a monthly basis. Even after China began measuring investment according to the metric of gross fixed capital formation, a key component in the SNA, analysts continued to use FAI as a proxy for gross fixed capital formation because the latter is published only annually and with a lag of several months. However, the methodology for compiling FAI was badly flawed because of double counting, falsification, and other well-known problems, some of which were openly acknowledged by China’s statistics officials (Holz 2020).

The growth of FAI began to outstrip the growth of gross fixed capital formation in 2007, and by 2015–16 reported FAI was about 90 percent higher than gross fixed capital formation (see figure 1). The NBS began to eliminate double counting and other flaws on a trial basis in four cities in 2013. By 2016 some provinces began reporting annual FAI data that were lower than the data they had reported for the previous year. As more provinces issued corrected data in subsequent years, the diagram shows the value of FAI converged toward that of gross fixed capital formation. The convergence will never be complete since there are some differences in coverage in the two data series, [2] but the higher frequency FAI data are now a much better proxy for gross fixed capital formation.

Goods Trade

Two Chinese agencies publish data on exports and imports. The State Administration of Foreign Exchange (SAFE) compiles China’s balance of payments data, which include goods trade statistics, based on the methodology and standards set forth in the Balance of Payments Manual (BPM) of the International Monetary Fund (IMF). The General Administration of Customs of China (GACC) compiles trade statistics following the methodology set forth in the United Nations International Merchandise Trade Statistics (IMTS). Goods trade data in SAFE’s balance of payments data starting in 2022 reflect trade between residents and nonresidents based on the principle of ownership transfer of goods. The GACC compiles trade data based on the cross-border movement of goods without considering whether ownership is transferred between residents and nonresidents.

SAFE first released China’s balance of payments, including data on imports and exports, in 1985 when it provided figures for the years 1982–84 based on the methodology set forth in the IMF’s BPM 4th edition published in 1977. In 2015 SAFE adopted the methodology of the BPM 6th edition first published in 2009, which provided more data disaggregation in some of the components. GACC independently has published data on imports and exports since 1980.[3] The differences in the trade data between the two series were typically quite small until 2022, when the data on the goods trade surplus released by SAFE increasingly fell short of the GACC numbers, creating what The Economist characterized as a “yawning gap”.[4]

The discrepancy between the two series was not, however, an attempt by the authorities to understate the size of China’s trade surplus. Rather, it reflected China’s adoption of a 2021 IMF guideline to use in accounting for trade in balance of payments statistics that involve what it calls “factory-less manufacturing” (IMF Committee on Balance of Payments Statistics 2021). The Fund’s goal was to update the measurement of trade to reflect modern global production arrangements.

In three recent semiannual balance of payments reports, SAFE has provided substantial explanations of its adoption of the new IMF methodology (SAFE 2022; 2023; 2024).

A typical example of the kind of modern production the new methodology is designed to measure more accurately is when a multinational company in country A supplies to a contractor in country B the parts and components as well as the intellectual property (say in the form of blueprints) needed to produce a final good that is then sold by the multinational (typically its sales agent in country B) to country C. If the multinational retains ownership of the parts and components, the intellectual property, as well as the final good, then the sale of the final good to country C should be treated as an export from country A, where the multinational is headquartered, and an import for country C. The services provided by the contractor should be recorded as an import of country A and an export of country B. In contrast to this new methodology, because the goods would cross the border between country B and country C, customs would record the sale of the good to country C as an export from country B.

In a world where global production arrangements are ever more complex, the customs methodology substantially overstates the contribution of the transaction just described to growth in country B, since the real income accruing to the contractor in country B is typically a very small share of the value of the export. Most of the value of the export accrues to the multinational headquartered in country A, particularly when the value of the intellectual property owned by the multinational contributes significantly to the value of the good exported to country C. Given that this type of factory-less manufacturing is widespread in China, the value of China’s exports reported by SAFE beginning in 2022 is far less than reflected in the data from the GACC. The revised methodology, however, provides a much more accurate measure of the contribution of trade to China’s GDP growth, which is why China’s statistical authority uses SAFE’s data on goods trade, not GACC’s data, in calculating its expenditure GDP (SAFE 2023, 25).

Household Income and Expenditure

Beginning in 2020 (with 2018 data), to come into compliance with SNA methodology, the NBS began to include social transfers in kind (STIK) as part of household disposable income and expenditure. These transfers from the government to households are mostly for free or subsidized health care and education when they are directly attributable to specific households.[5] The data were first reflected in the 2018 flow of funds and in revised accounts of household and government consumption expenditure going back to 1978.[6] As shown in figure 2, in the early years of reform these social transfers were relatively small but doubled to account for 6 to 7 percent of GDP by 2022, the most recent available data. STIK accounted for an average of 12 percent of GDP among the 30 member countries of the Organization for Economic Cooperation and Development (OECD) in 2022.[7] Average per capita GDP of OECD members in 2022 was about twice that of China, so it is not surprising that social transfers in China as a share of GDP were half that of the OECD average. A comparison with Mexico, the least developed OECD member with a per capita GDP slightly larger than China’s, may be more relevant: Mexico’s social transfers in kind in 2022 were less than 6 percent of GDP.

The Chinese data on STIK call for reevaluating the conventional wisdom that household consumption in China is weak. Taking STIK into account, as required by SNA, boosts household consumption in 2022 from the previously reported 37 percent to 43 percent of GDP. This is still low compared to high income countries in the west but close to the 45.5 percent average for upper middle-income countries, the World Bank category that includes China (World Bank 2023).[8]

Financial Services Value Added

Value added by the financial services industry (commercial and investment banks, securities firms, insurance companies, and so forth) is one of 13 components of services specified in the SNA. Until 2024 China’s statistical authorities estimated the value added of these services indirectly, based on the size of bank deposits and loans, and other gross indicators of financial services activities. Beginning in 2024 the NSB began to measure the value added of these services based on the incomes of service sector firms minus their costs.[9] This net measure is a much better reflection of value added in China’s service sector and is consistent with SNA methodology.

Conclusion

The paradox of China’s economic data thus lies in its dual character: The disappearance of certain data points, sometimes those related to economic weakness, raise concerns about the authority’s selective transparency. Yet at the same time, Beijing has better aligned its data with international standards and improved their quality. While its push for quality improvements shows its commitment to more reliable statistics, its selective withholding of information inevitably undermines confidence in the overall picture. If Beijing really wants to build trust in its economic data, it needs to ensure greater transparency, even though some statistics may expose vulnerabilities in its economy.

References

Gatley, Thomas, and Ernan Cui. 2019. The Case of the Mysterious Vanishing Statistics. Gavekal Dragonomics, October 17.

Holz, Carsten A. 2020. Understanding PRC Investment Statistics. Center for Economic Studies and ifo Institute (CSFifo) Working Paper No. 8110.

IMF Committee on Balance of Payments Statistics. 2021. “C.4 Merchanting and Factory-less Producers; Clarifying Negative Exports in Merchanting; and Merchanting of Services.”

SAFE. 2022. 2022 1H Balance of Payments Report. September 29.

SAFE. 2023. 2022 Balance of Payments Report. March 31.

SAFE. 2024. 2024 1H Balance of Payments Report. September 30.

World Bank. 2023. “Which Way Forward? Navigating China’s Post-Pandemic Growth Path.” December.

Xu, Xianchun. 2023. A Probe into China’s Official Statistics (in Chinese). Beijing: Social Sciences Academic Press (China).

Data Disclosure

The data underlying this analysis are available here (xlsx).

Notes

1. Such as data falsifications at local levels, an issue openly acknowledged by China’s top leaders and legislators. Concerns over data reliability prompted the Chinese legislature to amend the statistics law last year, aiming to combat data fraud and strengthen enforcement. Another major concern is the decline in transparency in recent years. Some data series have been quietly discontinued by the authorities, including bank lending to firms by ownership and industrial employment (Gatley and Cui 2019).

2. For example, FAI includes the value of sales of land and other assets, which do not increase the productive capacity of an economy and thus are not included in gross fixed capital formation.

3. Prior to 1980 China’s trade data was released by the Ministry of Foreign Trade.

4. “The Chinese authorities are concealing the state of the economy,” The Economist, September 5, 2024.

5. In kind transfers from employers to workers are accounted for in the flow of funds as part of compensation from employment (Xu 2023, 50).

6. This change in accounting did not affect China’s GDP since total consumption expenditure was unchanged; the increase in household consumption was exactly offset by a reduction in government consumption.

7. See here for social transfers in kind to households as shares of GDP in OECD countries.

8. It is unclear whether the World Bank’s figure for average household consumption as a share of GDP in upper middle-income countries includes STIK.

9. Wang Shiyu, “The optimization of the value-added accounting method in financial services may continue to ‘squeeze out water’ in monetary and credit data,” Caixin (in Chinese), May 14, 2024.