1. [para. 1][para. 2] China's rapid economic transformation, driven by digital technologies and new business models, is creating growing challenges for the statistical system that measures it. The country's GDP grew 4.7% in the first half of the year, with new growth drivers including high-end manufacturing, the digital economy, and modern services contributing more than 40% to growth. However, some economists believe aspects of the digital and service economies are not fully reflected in GDP figures.
2. [para. 3][para. 4] This challenge is not unique to China. Most countries still use the 2008 System of National Accounts (SNA), developed before the expansion of the internet, blockchain, cloud computing, and AI, leaving statistical systems struggling to adapt to a changing economy. The UN Statistical Commission formally adopted the 2025 SNA last year, with revisions covering globalization, digitalization, well-being, and sustainability. A key change treats data used in production for more than a year as a fixed asset, allowing related expenditures to be recognized as fixed capital formation and reflected in GDP.
3. [para. 6] China has established a research group to revise its national accounts system and is developing domestic standards based on the latest SNA and China's national conditions. Implementation around 2030, as recommended by the UN, will be difficult, partly due to challenges in valuing data and the relatively small number of Chinese companies currently including data in their financial statements.
4. [para. 7][para. 8][para. 9] More broadly, digitalization is exposing gaps in China's statistical system. Employment data illustrate this problem: China's surveyed urban unemployment rate, first officially published in 2018, replaced the previously criticized registered rate, but a gap has emerged between the surveyed rate and market perceptions of unemployment, leading to reduced attention to the indicator.
5. [para. 10][para. 11] The surveyed unemployment rate follows International Labour Organization standards, defining as employed anyone aged 16 or above who works at least one hour during the reference week. This definition doesn't capture employment quality—workers displaced from traditional industries may temporarily turn to flexible employment like ride-hailing or food delivery, without affecting the surveyed rate even as income, stability, and job quality decline.
6. [para. 12][para. 13] A senior researcher suggests supplementing existing indicators with the labor force participation rate and the distribution of weekly working hours to provide a more complete picture. Income statistics face a similar problem, relying heavily on averages—when differences emerge, the average's representativeness declines significantly, explaining why data differs from subjective feelings.
7. [para. 14][para. 15] Service consumption accounted for 46.1% of per-capita consumption expenditure last year, but service consumption statistics remain limited. Existing methods struggle to capture a service economy dominated by small enterprises and individual businesses, while rapid merchant entry and exit make annual statistical sampling less responsive.
8. [para. 17][para. 18] China has begun addressing the gap: since August 2023, the NBS has published year-on-year growth in service retail sales monthly, and since May 2026, has published the same type of indicator for total retail sales of consumer goods and services. However, only cumulative growth rates are published, without overall scale, monthly growth rates, or breakdowns by category.
9. [para. 19][para. 20][para. 21] Under the current system, data is not treated as a fixed asset, so related corporate expenditures count as intermediate consumption, not fully reflecting data's contribution to gross fixed capital formation. Digitalization can also reduce measured consumption, as consumers now access services like navigation or ticket-booking digitally for free. The 2025 SNA would treat data used in production for more than a year as a fixed asset under intellectual property products, while shorter-lived data remains intermediate consumption.
10. [para. 22][para. 23] Data is difficult to value because it can be reproduced and sold to multiple users at different prices. A senior statistics scholar noted that formulating domestic standards based on international ones is relatively easy, but truly implementing them is quite difficult.
AI generated, for reference only
Sign in with Google
Sign in with Facebook
Sign in with 财新