Interesting Fed of New York paper on AI impact on implementing an effective monetary policy.
Gist: AI doesn't change what central banks are trying to achieve, but it makes inflation, growth, and financial risks harder to interpret, increasing the likelihood of policy mistakes and requiring more flexible, data-driven policymaking.
Five key takeaways:
1. Inflation becomes harder to read: AI can change how firms produce goods and services, altering the relationship between economic activity and inflation. As a result, traditional indicators such as unemployment, labor shortages, or output gaps may become less reliable guides to future price pressures. Central banks may need to rely more on direct measures of costs, pricing behavior, and supply conditions.
2. Policy benchmarks become more uncertain: Rapid AI adoption makes it harder to estimate key variables such as potential output and the natural rate of interest. Productivity gains may take years to materialize because of implementation costs, organizational restructuring, and temporary inefficiencies. This increases the risk that policymakers mistake structural changes for cyclical fluctuations and set policy too tight or too loose.
3. Monetary policy transmission changes: AI can accelerate some transmission channels by speeding up information processing, expectation formation, and price adjustments. At the same time, technology adoption and business reorganization can delay the realization of productivity gains. The result is a more uneven and less predictable policy transmission mechanism than historical experience would suggest.
4. Financial risks may increase: AI can amplify financial vulnerabilities by encouraging reliance on similar models, faster information diffusion, and expectations-driven investment behavior. This may contribute to asset price booms that are disconnected from realized economic performance. Sudden corrections could tighten financial conditions and disrupt credit markets even when inflation remains under control.
5. The biggest risk is an AI-driven stagflation scenario: During the transition to AI, firms may face higher costs from investment, reorganization, and temporary productivity shortfalls, creating inflationary pressure. At the same time, financial markets may price in large future productivity gains, leading to elevated valuations and leverage. If those gains are delayed or disappoint, central banks could face both inflation and financial instability simultaneously, with interest rates alone insufficient to address both problems.
Core message: AI may raise productivity and expand economic capacity over the long run, but it also complicates economic management by making key policy signals noisier and less predictable. Moreover, while AI could exert disinflationary pressure over time through efficiency gains, the transition itself may be inflationary due to adoption costs, reorganization frictions, and large upfront investments.
------------------------------
Carlos Salas
Portfolio Manager & Freelance Investment Research Consultant
------------------------------