---
id: "lib-2026-03-00-ai-meets-trade"
type: "library"
title: "AI meets trade"
lang: "EN"
org: "OECD"
source_type: "Report"
date: "2026-03-01"
week: "2026-W09"
url: "https://nosov.ai/library/lib-2026-03-00-ai-meets-trade/"
original_url: "https://www.oecd.org/content/dam/oecd/en/publications/reports/2026/03/ai-meets-trade_6001acf4/13081644-en.pdf"
tier: "A"
tier_name: "Strong evidence"
tier_reason: "OECD Economics Department working paper with a Harvard Growth Lab co-author that runs a multi-country general equilibrium model with explicit scenarios; solid modelling, but a working paper rather than a flagship OECD report."
summary_basis: "full-text"
original_minutes: "120"
tags: ["ai", "trade", "policy", "economics"]
---

# AI meets trade

Tier A (Strong evidence): OECD Economics Department working paper with a Harvard Growth Lab co-author that runs a multi-country general equilibrium model with explicit scenarios; solid modelling, but a working paper rather than a flagship OECD report.

Summary based on the full text.

## In short

OECD Artificial Intelligence Papers No. 57 (March 2026) by Francesco Filippucci, Peter Gal, Katharina Laengle and Matthias Schief of the OECD Economics Department and Muhammed A. Yildirim of the Harvard Growth Lab. It estimates how much AI could raise per capita real income in OECD and G20 economies over the next decade, and how trade spreads those gains across borders. Gains vary widely by country. Lagging adopters get some benefit from cheaper AI-intensive imports, but staying competitive, especially in AI-exposed sectors, requires domestic adoption.

## Key numbers

- 0.1-0.95 pp: AI contribution to annual per capita real income growth across OECD countries over 10 years, central scenario
- ~0.03 pp: annual gain for the median OECD country from AI adoption abroad only, with no domestic adoption
- ~0.2 pp: competitiveness loss for countries like the UK and Israel if they do not adopt, about a quarter of central-scenario gains
- up to half: share of total AI welfare gains coming from abroad in Latin American and Central Eastern European OECD economies

## What it reports

- Central (medium adoption, expanded capabilities) scenario: AI adds 0.1-0.95 percentage points a year to per capita real income growth across OECD countries over 10 years. Slow adoption: 0.03-0.4 pp; rapid adoption: 0.2-1.2 pp.
- A mobile-phone-like diffusion pace gives about 50% larger gains on average; an electricity-like pace about 40% lower.
- The high end includes the US, knowledge-intensive economies (Switzerland, UK, Netherlands, Luxembourg) and strong adopters (Israel, Korea, Ireland, Sweden).
- Micro evidence used as input: performance gains from 14% in customer service to 56% in coding, around 40-45% in professional writing and consulting.
- Projected AI adoption in 2034 ranges from about 30% to 60% of businesses in leading adopters, and 2% to 10% in Mexico and Colombia.
- If a country does not adopt while others do, the median OECD country gains only about 0.03 pp a year. For the UK (finance) and Israel (ICT services), non-adoption costs about 0.2 pp, roughly a quarter of central-scenario gains.
- Strong knowledge spillovers through trade could triple gains for some Latin American OECD members and add 30-40% for several Central Eastern European members.

## How they got it

Sector-level productivity gains are built bottom-up from task-level performance studies, sector AI exposure and estimated current and future firm adoption following S-shaped paths of earlier general purpose technologies. These sector TFP gains feed a multi-country, multi-sector general equilibrium trade model. Results are decomposed into domestic and foreign sources, with separate simulations for non-adoption and knowledge spillovers.

## Limits

- Results depend on assumed adoption paths and capability scenarios; the authors say they are not forecasts and should not be added to existing growth projections.
- Micro-level gain estimates mostly measure time saved on tasks, which may overstate firm-level productivity.
- Adoption data for China and India rely on ad hoc surveys; the paper assumes adoption 50% above US levels in its high scenario.

## Why it matters for companies

For firms in knowledge-intensive sectors such as finance, professional services or ICT, the model shows that rivals in high-adoption countries will compete on price. Buying cheaper AI-enabled services from abroad does not offset losing market share at home or in exports. European firms in particular face a repeat of the 1995-2005 ICT gap, when the US gained 1-1.5 pp of labour productivity a year and Europe much less. Adoption speed is the part that companies and governments can influence.

Original: https://www.oecd.org/content/dam/oecd/en/publications/reports/2026/03/ai-meets-trade_6001acf4/13081644-en.pdf