We should start building fiscal insurance for the AI era
The Hill · C · trust 42/100
Comments: by Douglas Elmendorf and Louise Sheiner, opinion contributors - 08/16/26 12:00 PM ET Comments: Link copied by Douglas Elmendorf and Louise Sheiner, opinion contributors - 08/16/26 12:00 PM ET Comments: Link copied FILE – The OpenAI logo is displayed on a cellphone with an image on a computer monitor generated by ChatGPT’s Dall-E text-to-image model, Dec. 8, 2023, in Boston. (AP Photo/Michael Dwyer, File) Predictions about the economic effects of artificial intelligence vary enormously. Some analysts foresee rapid productivity growth and broadly shared prosperity. Others anticipate widespread job displacement, sharply greater inequality, or a large shift of income from workers to owners of capital.
No one knows which of these futures will materialize. But uncertainty should not obscure the magnitude of the potential disruptions. AI could transform the labor market and the distribution of income on a scale that would generate pressure for major changes in fiscal policy.
Yet, fiscal institutions cannot be designed overnight. Collecting and interpreting evidence takes time, while developing, legislating, and implementing new policies takes longer. Following the “China shock,” much of the policy response came only after geographically concentrated economic and social damage was deeply entrenched. And if the gains from AI accrue primarily to a small number of capital owners at the top of the income distribution, the resulting concentration of economic and political power could make some responses far more difficult later.
Policymakers should therefore begin developing forms of “fiscal insurance” against the most consequential risks — not because we already know which policies will be needed, but because credible, well-crafted options should be devised and available for broad use before the need for them becomes urgent.
The first risk is large-scale worker displacement. AI may create new jobs as well as eliminate existing ones. But workers who lose jobs may struggle to find new positions that match their skills or location. Some could face prolonged unemployment, large and persistent earnings losses, or withdrawal from the labor force. Research on displaced workers shows that the damage can extend well beyond lost wages, affecting workers’ health, families and communities.
The U.S. already has the beginnings of a policy response in Trade Adjustment Assistance , which has provided training, employment services, and income support to some workers displaced by international trade. But the program has been narrow and administratively burdensome, requiring workers to establish that their job losses resulted from a particular cause.
That model makes even less sense in an economy being reshaped by AI. It may be nearly impossible to determine whether a particular layoff resulted from AI, some other technological change, international trade, weak demand, corporate restructuring, or a combination of forces.
Policymakers should therefore begin developing a modernized system of adjustment assistance that is available to most or all workers experiencing significant displacement, without requiring them to prove a specific cause of their job loss. Such a system might include temporary income support, occupational and soft-skills training, links to nearby employers or broader job-search assistance, and wage insurance, which temporarily offsets part of the decrease in earnings when a worker takes a lower-paying new job.
Important questions remain about eligibility, generosity, duration, and interactions with unemployment insurance. In addition, evidence on training programs is mixed, so further experimentation to learn which approaches are most helpful to displaced workers would be valuable. A modernized system is not ready for enactment at national scale. But if AI produces major job losses, policymakers should not have to begin designing a response from scratch.
A second risk is that AI sharply increases the share of national income going to capital rather than labor. Because ownership of capital is highly concentrated, that outcome could further widen income and wealth disparities and increase the political influence of those who already own substantial assets.
One familiar policy response would be to raise taxes on capital income, wealth, inheritances, or consumption. These tools are fairly well understood, and the revenue that could be raised, especially if AI generates large increases in national income, could help finance policies that cushion worker displacement or expand prosperity.
A less familiar policy response would be to broaden ownership of financial assets. This approach could prove more socially appealing and politically durable than tax increases: people might consider a system of broader ownership to be appropriate, and an established public claim on capital returns is harder to reverse than a tax rate, because surrendering an ownership stake faces steeper hurdles.
One possibility of this sort would be a sovereign wealth fund that would acquire assets on behalf of the public. But such a fund would raise difficult questions about how it would be financed, who would control its investments, how political interference would be prevented, and whether its administrators would acquire excessive economic power.
Another possibility would be to place equity stakes in individual accounts. That approach could give people a direct and visible claim on capital income, but its effects would depend critically on the design. Loose restrictions on managing the accounts would give individuals more control but could erode the goal of durable, broadly held ownership; tight restrictions would preserve that goal but be more paternalistic and harder to administer.
These policy options for responding to potential worker displacement and shift in income to capital require more development and testing before they could be deployed confidently on a national…
Read the original at The Hill →
Open in TruthVane →