Fed risks rising structural joblessness amid rapid AI adoption

Regional Fed President Raphael Bostic told Reuters that the labour market may be entering a phase where firms need fewer employees because of automation and generative AI, a shift that can push the economy’s stable unemployment benchmark higher. He argues monetary policy has limited knobs to reverse changes that stem from permanent shifts in labour demand, so interest-rate decisions should reflect that new reality rather than temporarily masking it.

Bostic contrasted the short-run cyclical job picture with longer-term structural forces, urging policymakers to avoid using rate cuts to paper over a reallocation problem that properly belongs to fiscal policy and labour-market programs. He pointed to the Fed’s median long-run unemployment estimate of 4.2% and the headline jobless rate near 4.3% as the current calibration points that would shift if AI-driven productivity is durable and concentrated.

The Reuters interview arrives against a backdrop of divergent public commentary from other senior officials and nominees. Governor Michael Barr has laid out three scenarios — gradual diffusion, an investment-driven unwind, and an aggressive automation shock — to show how outcomes can range from manageable reallocation to persistent labour displacement. By contrast, supporters of nominee Kevin Warsh argue that sustained productivity gains could be disinflationary and, if verified across sectors, create room for sizable policy-rate cuts.

Beyond debate over theory, empirical signals are already mixed. Aggregate payroll additions were modest in 2025 (roughly 181,000 net for the year) while roughly 1.8 million Americans remained jobless for six months or longer. Corporate trackers recorded over 1.2 million job‑cut notifications in 2025 and planned layoff announcements in January topped 108,435, even as planned hires in that month were unusually low (around 5,306).

Technology-capital flows amplify the distributional pinch: market estimates put global AI infrastructure spending near $1.5 trillion in 2025, concentrated among a small set of hyperscalers. That concentration can limit broad diffusion of productivity gains and shift income toward capital owners rather than wage earners, a point emphasised by private-sector voices such as Anthropic’s Dario Amodei, who warns of rapid, acute shocks that compress adaptation windows.

For the Fed, the policy implications are twofold. If AI permanently raises structural unemployment or the neutral real rate (r*), the effective floor for policy rates will be higher and the scope for easing narrower — a dynamic that would lift long-term yields, term premia and borrowing costs. Alternatively, if productivity gains diffuse widely and pass through to lower prices and wages, they could be disinflationary and justify a calmer path to easing. The empirical difference hinges on persistence, breadth, and the public- and private-sector choices that shape diffusion.

Bostic framed the question as one of institutional boundaries and political economy: monetary policy can’t substitute for large-scale retraining, expanded benefits, immigration adjustments, or competition and industrial policy to broaden AI’s gains. He warned that absent such coordination, regional and demographic pockets — particularly early-career workers and those in labour‑intensive sectors — will experience earlier and more persistent pain.

Markets are already pricing this uncertainty into futures and fixed-income curves: higher expected r* or structural unemployment would push out the timing and magnitude of cuts, while a confirmed productivity dividend would narrow financial conditions and could reignite debate on easing. Operational frictions at the Fed — reserve management, Treasury issuance schedules and money-market plumbing — add further sequencing constraints to any shift in stance.

Bostic’s comments also intersect with internal Fed governance and external politics: as leadership transitions and confirmation fights play out, new decision-makers will need to build empirical consensus across the 12‑member FOMC to avoid fragmented communication. He recommended sustaining descriptive staff research and careful governance to maintain credibility when structural shocks change the policy calculus.

In short, Bostic reframes AI-driven productivity gains as a coordination and distribution challenge as much as a macroeconomic one: markets and policymakers should watch hiring composition, long‑tenured unemployment, diffusion metrics and capital-concentration indicators for early signs that the Fed’s natural-unemployment benchmark needs updating.