2024-04-29-国际清算银行-人工智能对产出和通货膨胀的影响（英）_42页_1mb

报告摘要

BIS Working Paper: Impact of Artificial Intelligence on Output and Inflation

Key Findings

AI Adoption Impact:
- AI significantly boosts aggregate output, consumption, and investment in both short and long run.
- The increase in GDP stabilizes around 35% above baseline in the long run, driven by capital stock expansion and productivity gains.
Inflation Response:
- Unanticipated AI: Initially disinflationary due to supply expansion, but inflation increases ~0.75pp after 4 years due to demand-driven effects.
- Anticipated AI: Immediate inflation rise as households accelerate consumption, delaying investment and amplifying inflationary pressures.
Sectoral Effects:
- Direct Exposure ≠ Long-Term Impact: Industry AI usage initially correlates weakly with future output growth, as spillovers and linkages dominate.
- Consumption vs. Investment Goods: Output growth is twice as large when AI affects consumption-good sectors (e.g., ~50% vs. ~25% in manufacturing).
- Labour-Intensity: More labour-intensive industries record smaller value-added increases due to wage pressures.
Policy Implications:
- Foster AI adoption in consumption-good sectors to maximize output gains.
- AI could create a Goldilocks scenario for monetary policy: short-term disinflation followed by manageable inflation control via tightening.
- Public policy should prioritize sector-specific adoption to offset secular challenges (e.g., aging, supply chain disruptions).
Uncertainty Notes:
- AI effects differ under alternative assumptions (e.g., labour/capital-augmenting), but long-run outcomes remain positive across models.
- Cross-country variability in sectoral impacts highlights the need for targeted policy interventions.