## Instilling Trust through Statistics

## Source details

**Canonical URL:** [Instilling Trust through Statistics](https://www.imf.org/-/media/files/publications/fandd/article/2025/12/kroese.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/fandd/article/2025/12/kroese.pdf.md)
- [Structured JSON version](/-/media/files/publications/fandd/article/2025/12/kroese.pdf.json)

---

### Background and illustrative cases
- Argentina: In the mid-2000s, official inflation statistics began to diverge from independent estimates; by 2007, private analysts pegged inflation as much as triple the officially reported rate. The credibility of Argentina’s national statistics office collapsed, leading investors to lose confidence and pull out funds.
- Nigeria: After waiting 20 years to rebase its national accounts, in 2010 Nigeria announced its economy was almost 60 percent bigger than previously estimated.
- These episodes illustrate how weakened statistical institutions—due to chronic underfunding and political interference—can compromise the integrity of economic data and the decisions built upon them.

### Major challenges facing NSOs
- Data environment complexity:
  - Data generation at unprecedented scale from social media, smart devices, and private sector sources.
  - Enabled by surging computing power and artificial intelligence, but tools can also produce misleading or fabricated results.
  - Large language models still hallucinate; leading models consistently fail to produce accurate figures when prompted to produce a table of economic growth rates using the latest IMF World Economic Outlook—even when given the source. Most of the numbers are close but incorrect.
- Institutional pressures:
  - Chronic underfunding and political interference undermine independence and quality.
  - Declining response rates to traditional surveys: the response rate to the UK’s Labour Force Survey dropped below 15 percent in 2023, leading to a temporary suspension of the official releases that underpin employment estimates.
- Measurement challenges:
  - Need to rebased GDP and update series regularly to remain useful.
  - Evolving economy with new sectors like the gig economy and digital services requiring fresh measurement approaches.
- Resource constraints:
  - By one estimate, the budget of the US Bureau of Labor Statistics has declined by an inflation-adjusted 22 percent since 2010.
- Data access and sharing limitations:
  - NSOs require access to administrative and private sector data; legal frameworks and coordination are often insufficient.

### Fundamental principles for strong NSOs
- Independence:
  - Legal frameworks must protect NSOs from external interference, allowing them to choose methodologies and publish findings based on professional judgment.
  - Leaders should be empowered to make decisions grounded in statistical expertise; staff should adhere to the highest ethical standards, including safeguarding confidential data and using it solely for statistical purposes.
- Adequate funding:
  - NSOs need resources to compete for talent, invest in technology, and conduct research into emerging data challenges (response rates, new survey techniques, innovative data sources), while maintaining day-to-day production of core statistics.
- Access to data:
  - Governments should facilitate sharing of administrative data with NSOs.
  - Legal frameworks should enable secure and confidential data exchange between NSOs and other official statistics producers, such as central banks.
- Standards and comparability:
  - Integrated and well-defined data, aligned with global standards, ensure statistics are accurate and comparable across countries.

### Capabilities and modernization priorities
- Innovation and methods:
  - Integrate alternative data sources (administrative records, satellite imagery, private sector data).
  - Use big data and AI techniques while ensuring methodological rigor and validation.
- Technology and discoverability:
  - Be AI-ready, with well-structured metadata and access via application programming interfaces so information can be easily found and used by modern tools and platforms.
  - Cooperation with AI developers to ensure official statistical data are more discoverable.
- Coordination:
  - Establish national coordination committees to oversee statistical governance and promote collaboration among official statistics producers.
- Partnerships:
  - Collaborate with other official statistics producers, academic institutions, international organizations (such as the IMF), and technology companies as data providers and dissemination channels.
- Communication and public engagement:
  - Invest in data visualization, interactive dashboards, and plain-language summaries.
  - Embrace open data initiatives and modern communication channels—such as social media and data portals—to engage the public and counter misinformation.
  - Improving visibility and transparency can build public trust and improve survey response rates.

### Policy recommendations
- Enshrine and protect NSO independence through legal frameworks that prevent political interference.
- Ensure sustainable and adequate funding to allow NSOs to:
  - Compete for talent.
  - Invest in technology and research.
  - Maintain core statistical production.
- Facilitate secure data sharing:
  - Governments should enable administrative and private data access under confidentiality safeguards.
  - Promote legal frameworks for secure exchange among official statistics producers.
- Modernize statistical infrastructure:
  - Make data AI-ready with structured metadata and APIs.
  - Adopt big data and AI methods with rigorous validation to guard against misleading outputs.
- Strengthen governance and coordination:
  - Create national coordination committees for statistical governance.
  - Promote international cooperation to align methodologies and build capacity.
- Improve communication and transparency:
  - Use visualization, interactive tools, and plain-language outputs to reach diverse audiences and counter misinformation.

### Key statistics and specific data points preserved from the source
- By 2007, private analysts pegged inflation in Argentina as much as triple the officially reported rate.
- Nigeria waited 20 years to rebase its national accounts and then, in 2010, announced its economy was almost 60 percent bigger than previously estimated.
- The budget of the US Bureau of Labor Statistics has declined by an inflation-adjusted 22 percent since 2010.
- The response rate to the UK’s Labour Force Survey dropped below 15 percent in 2023.

*Bert Kroese, IMF chief statistician and director of the Statistics Department, F&D, December 2025.*

---


_Source: https://www.imf.org/-/media/files/publications/fandd/article/2025/12/kroese.pdf_