Fiscal Policy Can Help Broaden the Gains of AI to Humanity

IMF Blog, June 17, 2024

Source details

Canonical URL: Fiscal Policy Can Help Broaden the Gains of AI to Humanity

Other formats

Bibliographic details

Authors: Era Dabla-Norris, Ruud de Mooij
Published: June 17, 2024

Overview

Publication date: June 17, 2024
Authors: Era Dabla‑Norris, Ruud de Mooij
Core argument: Fiscal policy can play a major role in supporting a more equal distribution of gains and opportunities from generative-AI, but doing so requires significant upgrades to social-protection and tax systems worldwide.
Context: New generative-AI technologies offer large productivity and public-service gains, but rapid, large-scale transformation raises risks of job losses and greater inequality. Governments should adopt an agile approach to prepare for highly disruptive scenarios.

Social protection and labor-market responses

Risks and transitions:
- AI could eventually boost overall employment and wages, but could also put large swaths of the labor force out of work for extended periods, creating painful transitions.
Unemployment insurance and wage support:
- More generous unemployment insurance could cushion negative impacts, allowing displaced workers to find better skill matches.
- Most countries have considerable scope to broaden coverage and generosity of unemployment insurance, improve portability of entitlements, and consider forms of wage insurance.
Training, reskilling, and targeted assistance:
- Sector-based training, apprenticeships, upskilling and reskilling programs should play greater roles in preparing workers for AI-era jobs.
- Comprehensive social-assistance programs will be needed for workers facing long-term unemployment or reduced local labor demand due to automation or industry closures.
Emerging-market and developing economy considerations:
- Workers in such countries are less exposed to AI, but are also less protected by formal social-protection programs because of larger informal sectors.
- Innovative approaches leveraging digital technologies can facilitate expanded coverage of social-assistance programs in these countries.

Tax policy: whether and how to tax AI

On taxing AI directly:
- A specific tax on AI (for example, a "robot tax") is not advisable because only people can pay taxes; such a tax could reduce the speed of investment and innovation, be hard to implement, and, if ill-targeted, do more harm than good.
Rebalancing tax incentives and corporate taxation:
- In recent decades, some advanced countries scaled up corporate tax breaks on software and computer hardware to drive innovation; these incentives also tend to encourage firms to replace workers through automation.
- Corporate tax systems that inefficiently favor the rapid displacement of human jobs should be reconsidered given the risk of magnified dislocations from AI.
- Many emerging market and developing countries have corporate tax systems that discourage automation, which can be distortive by preventing investments that would enable catching up in the global AI economy.
Redistributive taxation to offset rising inequality:
- Generative-AI can lead to higher income inequality and concentration of wealth; taxes on capital income should be strengthened to protect the tax base against a further decline in labor’s share of income and to offset rising wealth inequality.
- Strengthening corporate income taxes could help reverse the long-run trend since the 1980s of a declining tax burden on capital income in advanced economies and a rising burden on labor income.
- Policy instruments cited as options:
- The global minimum tax agreed by over 140 countries, which establishes a minimum 15-percent effective tax rate on multinational companies, is a step in the right direction.
- Other possible measures include a supplemental tax on excess profits, stronger taxes on capital gains, and improved enforcement.

Policy recommendations and priorities

Upgrade social-protection systems:
- Broaden coverage and generosity of unemployment insurance.
- Improve portability of entitlements.
- Consider wage insurance and expand comprehensive social-assistance for long-term unemployed or locally displaced workers.
Invest in human capital:
- Scale sector-based training, apprenticeships, upskilling and reskilling programs to prepare workers for AI-era jobs.
- Increase public investment in education and social spending to broaden gains from AI (financing requires more public revenue).
Reform tax systems to support workers and redistribution:
- Reconsider tax incentives that unintentionally encourage automation and worker displacement.
- Strengthen taxes on capital income and corporate taxes where appropriate.
- Implement and enforce measures such as the 15-percent global minimum tax, supplemental taxes on excess profits, and stronger capital gains taxation.
Tailor approaches for emerging-market and developing countries:
- Use digital technologies to expand social-assistance coverage in economies with large informal sectors.
- Balance corporate tax settings to avoid deterring productive automation while enabling catch-up investments.

Broader implications and international cooperation

Recognition of public funding role:
- Recent AI breakthroughs build on years of investment in fundamental research, including through publicly funded programs.
Long-term shaping of AI:
- Decisions made now by policymakers will shape the evolution of AI for decades; priority should be ensuring applications broadly benefit society, leveraging AI to improve education, health, and government services.
Need for global collaboration:
- Given AI’s global reach, greater cooperation among countries will be increasingly important.

Based on a staff discussion note contributed to by Fernanda Brollo, Daniel Garcia‑Macia, Tibor Hanappi, Li Liu, and Anh Dinh Minh Nguyen; blog authors: Era Dabla‑Norris and Ruud de Mooij; June 17, 2024.