AI's effect on equity markets is ambiguous: productivity gains lift corporate cash flows, but worker displacement and systemic alignment risk can push required returns higher; in deep markets alignment-related tail risk dominates, while in shallow markets withdrawal of retail investors produces persistent elevation of the equity risk premium.

Main Finding

AI-driven labour displacement affects the equity risk premium (ERP) through three co-equal, additively separable channels—(i) a productivity channel that is equity-bullish, (ii) a participation compression channel that can be equity-bearish by raising the required risk premium via a shrinking investor base, and (iii) an alignment risk channel (ARP) specific to agentic AI that raises risk premia via model-uncertainty/fat-left-tail output risks. The net effect on ERP depends on which channel dominates; deep financial markets are dominated by ARP (Regime D) while shallow markets are dominated by participation compression (Regime E). The model delivers analytical characterisations of nonlinear financial fragility, regime-contingent divergence in risk premia, and an equilibrium “alignment squeeze,” and predicts persistent ERP elevation (participation hysteresis) under certain conditions. India’s SIP flows are proposed as an empirical proxy for the endogenous participation rate.

Key Points

• Three channels (co-equal and additively separable):
  • Productivity channel: AI raises firm cash flows and fundamentals → pushes equity prices up (bullish).
  • Participation compression channel: displacement pushes marginal households below market entry cost κ, shrinking the investor pool; consumption risk concentrates on fewer investors, raising required ERP via a Basak–Cuoco style mechanism (bearish).
  • Alignment risk channel (ARP): agentic AI can produce correlated misalignment shocks with fat left tails and no reliable historical frequency; modelled via Hansen–Sargent multiplier (robustness/ambiguity) preferences → priced factor that raises ERP even if fundamentals improve.
• Proposition 1: equilibrium ERP decomposes additively into three terms corresponding to the channels.
• Two regimes:
  • Regime D (deep markets): ARP dominates, elevated ERP driven by alignment/ambiguity risk.
  • Regime E (shallow markets): participation compression dominates, displacement reduces investor participation and elevates ERP despite productivity gains.
• Analytical results:
  • Non-linear financial fragility: small changes in displacement or market depth can produce large ERP moves.
  • Regime-contingent divergence: identical productivity boosts can produce opposite ERP responses across regimes.
  • General equilibrium alignment squeeze: alignment risk raises the premium independently of the participation wedge, compressing valuations.
• Additional dynamics:
  • AI capital concentration amplifies the regime divide and can lock shallow markets into a permanent displacement attractor.
  • Participation hysteresis: ERP may remain elevated after employment normalises because market re-entry is costly.
• Empirical implication: high-frequency systematic investment plan (SIP) flows (e.g., in India) are a natural observable for the model’s endogenous participation rate and can serve as an empirical laboratory to test the displacement–participation mechanism.

Data & Methods

• Model architecture:
  • Heterogeneous-agent general equilibrium framework with a fixed equity market entry cost κ that determines investor participation.
  • Firms experience AI-driven productivity changes affecting cash flows.
  • Participation compression implemented by moving marginal households below κ, concentrating consumption risk among active investors (Basak–Cuoco mechanism).
  • Alignment risk modelled using Hansen–Sargent multiplier preferences to capture aversion to model misspecification / ambiguity and produce fat-left-tail aggregate output shocks.
• Analytical approach:
  • Closed-form decomposition of equilibrium ERP into three additively separable components (Proposition 1).
  • Characterisation of regimes via market depth and the relative strength of ARP vs participation wedge.
  • Comparative statics and phase-regime analysis showing non-linearities, attractors, and hysteresis.
• Empirical strategy suggestion:
  • Use country- or market-level high-frequency flows into retail equity vehicles (e.g., India’s SIP flows) as an endogenous, observable proxy for investor participation.
  • Testable predictions include relationships between displacement measures, participation flows, and contemporaneous ERP movements across markets of varying depth.

Implications for AI Economics

• Asset pricing and valuation:
  • AI can simultaneously improve fundamentals and raise ERP; equity valuations may fall or rise depending on market depth and alignment risk—so standard fundamental-only valuation is incomplete.
  • Fat-tail alignment risks imply higher discount rates even when cash flows rise; valuations must incorporate ambiguity/robustness premia.
• Financial stability and distributional concerns:
  • Participation compression concentrates financial risks, raising systemic fragility and amplifying inequality between equity holders and non-holders.
  • Shallow markets are at risk of permanent displacement equilibria and prolonged high ERPs; policy should consider market-depth resilience.
• Policy and regulation:
  • Monitoring and mitigating alignment risk (safety, verification, coordination) is macro-financially important because ARP can dominate in deep markets.
  • Policies to lower equity market entry costs or broaden participation (e.g., easier retail access, pooled instruments, subsidised entry) can reduce participation-compression-driven ERP spikes.
  • Antitrust and capital-concentration policies matter: concentrated AI capital intensifies regime divergence and potential for persistent fragility.
• Empirical research agenda:
  • Use SIP and other retail-flow measures to test the displacement–participation channel; cross-country comparisons (market depth) can test regime predictions.
  • Estimate the magnitude of ARP using models of ambiguity aversion and tail-risk pricing conditional on observed AI incidents or near-miss events.
• Investment and risk management:
  • Investors and portfolio managers should incorporate regime-dependence: risk premia may rise despite better fundamentals when participation or alignment channels dominate.
  • Stress-testing and scenario analysis should include correlated alignment shocks with fat tails rather than relying solely on historical loss distributions.