Global Trade Linkages and the Cross-Country Distribution of the Gains From AI

Why It Matters

The analysis shows that AI’s macroeconomic impact is uneven and heavily mediated by trade links, highlighting the need for policies that boost domestic AI capacity while maintaining openness. Countries that fail to adopt AI risk losing market share despite lower import prices.

Key Takeaways

• AI adoption averages 4% in OECD, ranging 0.2% Mexico to 7% Ireland
• Projected per‑capita income gains vary 0.1‑0.95% annually (medium pace)
• Foreign AI adoption cuts import prices, boosting welfare for open economies
• Without domestic AI, competitiveness loss offsets most foreign benefits
• Trade‑related knowledge spillovers modestly raise gains but don’t close gaps

Pulse Analysis

Artificial intelligence is poised to become the next engine of productivity, but its effects will not be uniform across the globe. By integrating micro‑level task productivity gains with a multi‑country, multi‑sector general‑equilibrium trade model, the authors reveal that AI adoption rates differ dramatically—about 4 % on average in OECD nations, yet as low as 0.2 % in Mexico and as high as 7 % in Ireland. These divergent trajectories translate into per‑capita real‑income growth ranging from a modest 0.1 percentage point per year to nearly 1 percentage point in the most advanced economies, comparable to the ICT boom of the 1990s. The model also isolates the trade channel: foreign AI adoption drives down the price of imported intermediates and final goods, delivering a sizable welfare boost to highly open economies that rely on AI‑intensive services.

The study underscores that foreign spillovers alone cannot compensate for a lack of domestic AI deployment. Experiments where a single country abstains from AI adoption show a sharp decline in competitiveness, eroding most of the import‑price benefit. Even with knowledge spillovers—where firms learn AI practices from trade partners—the uplift is modest, suggesting that learning effects, while present, will not close the productivity gap between early and late adopters. The findings imply that the distribution of AI gains will be shaped as much by national innovation ecosystems as by global trade patterns.

Policymakers therefore face a dual mandate: accelerate domestic AI uptake through investments in digital infrastructure, data ecosystems, skill development, and reliable energy, while preserving low barriers to digital‑services trade. Such a strategy ensures firms can access cutting‑edge AI models abroad and translate them into domestic productivity gains. In a world where AI can add up to a full percentage point to annual income growth, the balance between openness and home‑grown capability will determine which economies capture the lion’s share of the AI dividend.