Aggregate Gains from AI and Their Distribution: Global Evidence from Usage Data
IMF Working Papers, July 10, 2026
Source details
- Canonical URL
- Aggregate Gains from AI and Their Distribution: Global Evidence from Usage Data
Other formats
Bibliographic details
- Authors: Rachel Yuting Fan, Ha Minh Nguyen
- Published: July 10, 2026
- Series: IMF Working Papers
- DOI: https://doi.org/10.5089/9798229051279.001
Core findings
- Using five waves of the Anthropic Economic Index (January 2025 to February 2026), the authors construct two novel measures from observed AI usage across countries.
- The AI concentration index (ACI) indicates:
- In developing economies, virtually all AI usage-based value is generated in a small professional enclave (ACI near 1.0).
- High-income economies average 0.4 to 0.5, with concentration declining in many countries.
- The labor cost equivalent (LCE) values the time currently saved by AI at $2.7 trillion annually (3.4% of GDP), presented as an indicative measure of the labor cost of that time.
Measures and data
- Data source: Five waves of the Anthropic Economic Index (January 2025 to February 2026).
- Novel measures constructed:
- AI concentration index (ACI): quantifies how concentrated AI usage-based value is across occupations within countries.
- Labor cost equivalent (LCE): values time saved by AI in dollar terms and as a share of GDP.
Distributional patterns and predictors
- Concentration patterns:
- Developing economies: ACI near 1.0, indicating concentration of AI usage-based value in a small professional enclave.
- High-income economies: ACI averages 0.4 to 0.5, with evidence of declining concentration in many countries.
- Predictors of concentration:
- Income predicts concentration.
- Regulatory readiness predicts concentration.
- Lacking an official English language slows broadening of AI usage, posing a barrier for developing countries.
Key statistics and publication metadata (from source)
- LCE: $2.7 trillion annually.
- LCE as share of GDP: 3.4% of GDP.
- Time span of index waves: January 2025 to February 2026.
- Pages: 36.
- Volume: 2026.
- Issue: 147.
- Series: Working Paper No. 2026/147.
- DOI: https://doi.org/10.5089/9798229051279.001.
- ISBN: 9798229051279.
- ISSN: 1018-5941.
- Publication date: July 10, 2026.
- Authors: Rachel Yuting Fan, Ha Minh Nguyen.
Implications for policy and research
- Broadening AI usage outside a small professional enclave appears limited in developing economies; policy and regulatory readiness matter for diffusion.
- Language barriers (absence of an official English language) are identified as a tangible obstacle to broader AI adoption in developing countries.
- The $2.7 trillion LCE (3.4% of GDP) provides an indicative aggregate scale of current time savings from AI, informing labor cost and national accounts considerations.
IMF Working Papers — "Aggregate Gains from AI and Their Distribution: Global Evidence from Usage Data" (Rachel Yuting Fan, Ha Minh Nguyen), July 10, 2026.
Content in this bundle
- Working Paper