Outlet assessment
World Bank (research and policy publications)
Reliability
- Assessment confidence
- medium
- Check
- full assessment · provisional
- Assessed
- 2026-11-14
- Reevaluation
- due after 13 May 2027 (180 days; only if the outlet is still in use. How)
- Kind
- ngo, INT
- Leaning
- market-oriented; cooperative with Beijing
How scores are set: outlet assessment.
Kind: ngo. It is an international financial institution (IBRD and IDA, founded 1944, Washington DC, 189 member countries). This profile covers its research, policy and data publications on documents.worldbank.org: Policy Research Working Papers, China Economic Updates, East Asia and Pacific Economic Updates, Poverty and Equity Briefs, country reports and Prosperity Data360 snapshots. It does not cover project or procurement documents. Deep research uses these publications for economic statistics and estimates about China.
Summary#
- Reliability overall is 84 (70-89 band). In the sample, 18 of 18 checkable first statements were accurate or mostly accurate. The rate is 100%, with a Wilson 95% interval of 82.4-100 (n = 18 checkable of 24 drawn). Of those 18, 10 were accurate and 8 mostly accurate.
- No statement was false or misleading. Minor errors were common, though, and usually came from rounding or from citing an old data vintage. Examples:
- PGM reserves given as “over 96 percent” African; USGS 2025 gives about 79%.
- National development financial institutions said to provide “the majority” of public climate finance; the CPI table gives 43%.
- China’s 1992 tariff given as “over 40 percent”; WDI gives 39.7%.
- The outlet’s statements about China’s own macro figures matched official data: GDP 2025 at 5.0%, Q4 2025 at 4.5%, Q1 2026 at about 5.0%, and the China Gini of 36.1.
- Not graded here: forecasts. Their dates have not passed.
- The outside record holds a documented case of manipulated China data. Leadership pressure raised China’s Doing Business 2018 ranking, and the report series was discontinued in 2021.
- Assessment confidence is medium (provisional). It rests on what was checked: 18 checkable items, all accurate or mostly accurate (95% interval 82.4-100), and 3 wording comparisons. The real limits are the sample size (under 24 checkable items) and thin perception (6 posts). The 6 items whose references could not be read, the 5 missing wording comparisons and the blocked platforms are not counted; they are queued for retry (see Pending checks).
Ownership#
- Chain: documents.worldbank.org publications → World Bank (IBRD/IDA) → 189 member governments.
- Control sits with the Board of Governors and the Board of Executive Directors.
- IBRD voting power as of 2026-06-30: United States 15.98%, Japan 6.88%, China 5.75%, UK 3.75% (IBRD voting table).
- “As changes to the bank’s Charter require an 85% supermajority, the U.S. can block any major change” (Wikipedia).
- No single controlling owner, so no group profile was staged.
- China is the third-largest shareholder, a borrower (project documents dominate the China document list, e.g. Hunan methane-reducing paddy rice PforR) and a co-author of flagship work (World Bank and DRC 2022, cited in the sampled middle-class paper).
- Policy Research Working Papers carry authors’ names and a disclaimer: “The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank … or the governments they represent.”
- Flagship reports state “The World Bank does not guarantee the accuracy … of the data.”
- Editorial independence is unknown. The WilmerHale investigation shows shareholder pressure reaching data in one product, Doing Business 2018. No evidence was found either way for the current China research products.
Accuracy#
- Statistics: 13 checkable, 13 accurate or mostly accurate. Minor errors were rounding or old-vintage figures:
- EMDE share of world GDP is 45% in the outlet against 41% in the IMF WEO, a grouping difference.
- East Asia and Pacific poverty in 1990 is 987.1M in the outlet against 1,031.5M in current PIP data for all of EAS.
- Fact-reports: 3 of 3. The one mostly-accurate item was the Yingcheng compressed-air storage plant: it was the world’s largest when the note was published, but a 60 MW non-fuel plant (Jiangsu, 2022) came before it.
- Research findings: 2 of 2. The Deng et al. 2025 figures were reproduced exactly, but the reference was missing from the bibliography. One abstract said “consistent positive” where its own table has one negative terrain class.
- Corrections: no corrections were seen in the sample. The Documents & Reports system shows no correction notices for these items.
- Not checkable (6): Thailand March CPI, GMD country count, the processing-trade duty-free fact, the IEA demand-response share, China Ecosystem Assessment survey counts, Macao WTTC/arrivals. In each case the reference could not be reached.
Outside record#
- Doing Business data irregularities (2017-2021). WilmerHale, report to the Board, 2021-09-15:
- “high-ranking Chinese government officials repeatedly expressed their concerns to President Kim … that the country’s ranking … 78 in Doing Business 2017 … did not accurately reflect its economic reforms.”
- The data changes “boosted China’s score by nearly a point and increased its ranking by seven places to 78.”
- The Bank “discontinued [Doing Business] on September 14, 2021 following an audit documenting how bank leadership pressured experts to manipulate the results of the 2018 and 2020 reports” (Wikipedia).
- This is the main documented case of China-related data being shaped by shareholder relations.
- Media-rating groups and fact-checkers could not be read (search down). Not done.
Orientation#
- Economic policy (moderate, medium confidence). The outlet frames China around market-oriented rebalancing:
- “domestic demand remains subdued, the difficulties in the property sector persist” (EAP Update Apr 2026).
- “在现有政策框架下,经济向消费驱动型增长模式转型的进程缓慢推进” (China Economic Update Jul 2026).
- Toward the Chinese government (moderate, medium confidence). The outlet is cooperative:
- It repeats official achievements: “China alone lifted 800 million people out of extreme poverty”, attributed to WB and DRC 2022.
- Its Chinese-language reports adopt official policy vocabulary, for example the 生态保护红线 (ecological protection red line) framework.
- The Hong Kong justice snapshot reports “Human Rights and Rule of Law Data is not available”.
- No political or human-rights critique appeared in the sample.
- It is critical in economic terms: it warns about the property sector and weak consumption.
- Institutional interest (mild). Its own survey reports high trust in the World Bank Group among China stakeholders. The respondent list came from “the WBG country team”.
Wording#
- Event: China’s 2025 GDP. World Bank: “growth is projected to decline from 5.0 percent in 2025”, framed as deceleration. Other outlets framed it as target met:
- Yicai Global: “Meeting Annual Target”.
- Bastille Post: “hitting annual target”.
- CPEC News: “meeting the annual growth target”.
- Event: Thailand’s April 2025 CPI. World Bank: “Inflation turned negative for the first time in over a year, prompting the Bank of Thailand to lower its policy rate”. The Nation: “raising deflation concerns”.
- Event: Yingcheng compressed-air storage plant. World Bank: “a breakthrough for long-duration storage technology”. ESS News, more neutrally: “previously held the title”.
- Pattern: in macro reporting the outlet uses forecast and deceleration language. Its project and knowledge notes use promotional wording. Only 3 of the 8 required events could be compared.
Perception#
- Hacker News (5 posts):
- 4 use World Bank data as an authority for China figures, for example “Here is a more reliable source from worldbank”.
- 1 cites the Doing Business scandal: “Management tried to change the model to raise the rankings of China and Saudi Arabia”.
- The World Bank’s own FY25 China Country Opinion Survey (n=163): “Trust the WBG to do what is right for China” 8.5/10.
- Not readable: X, Reddit, YouTube, review sites and forums, media-rating groups.
By audience (platform profiles, 2026-11-15): center left audiences, 5 posts (Hacker News 5): 3 trusts, 1 mostly distrusts, 1 mixed; mean trust 75.0, gap -9.0 points against measured reliability 84. All posts come from left-of-center audiences (one audience side), so perception is capped at medium and no cross-audience reading is possible yet; the next top-up samples center and right-leaning venues (Free Republic, GETTR, patriots.win). Platform pages: Hacker News.
How it is weighed#
- Trust more: reproductions of official or harmonized data (PIP, WDI) and summaries of NBS figures. These matched the sources.
- Trust less: exact secondary percentages in thematic reports (reserves shares, climate-finance shares). Recheck them against the cited primary source; minor errors were common.
- Likely bias:
- Rankings and indices where China, as a shareholder, has lobbied (Doing Business precedent).
- Self-assessments of World Bank programs and engagement.
- Silence on political or human-rights topics, which is selection rather than inaccuracy.
- Counting: Policy Research Working Papers are author views, not institutional positions. Do not count a PRWP and a World Bank flagship that reuse the same data as independent confirmations.
Used in posts#
- Layout fixture: a post tied to the China project: 3 reference(s) · filter references
- PDF World Bank Document (World Bank, 2019-08)
- PDF World Bank Document (World Bank, 2020)
- World Bank Poverty and Inequality Platform API: China, all years (World Bank, 2026)